الرئيسية [World Scientific Handbook in Financial Economics Series] The World Scientific Handbook of Futures...
أبلغ عن مشكلةThis book has a different problem? Report it to us
إختار نعم إذا كان إختار نعم إذا كان إختار نعم إذا كان إختار نعم إذا كان
نجحت في فتح الملف
يحتوي الملف على كتاب (يُسمح أيضًا بالرسوم الهزلية)
محتوى الكتاب مقبول
يتوافق الإسم و المؤلف و لغة الملف مع وصف الكتاب. تجاهل الخانات الأخرى لأنها ثانوية!
إختار لا إذا كان إختار لا إذا كان إختار لا إذا كان إختار لا إذا كان
- الملف تالف
- الملف محمي بواسطة الوسائل التقنية لحماية حق المؤلف DRM.
- الملف ليس كتاب (علي سبيل المثال، xls, html, xml)
- الملف عبارة عن مقال
- الملف مقتطف من كتاب
- الملف عبارة عن مجلة
- الملف نموذج إختبار
- الملف سبام
هل تعتقد أن محتوى الكتاب غير مناسب ويجب حظره
لا يتطابق الإسم أو المؤلف أو لغة الملف مع وصف الكتاب. تجاهل الخانات الأخرى.
Change your answer
July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 CHAPTER 22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. HOW TO LOSE MONEY IN DERIVATIVES: EXAMPLES FROM HEDGE FUNDS AND BANK TRADING DEPARTMENTS∗ Sebastien Lleo and William T. Ziemba Abstract: What makes futures hedge funds fail? The common ingredient is over betting and not being diversiﬁed in some bad scenarios that can lead to disaster. Once troubles arise, it is diﬃcult to take the necessary actions that eliminate the problem. Moreover, many hedge fund operators tend not to make decisions to minimize losses but rather tend to bet more doubling up hoping to exit the problem with a proﬁt. Incentives, including large fees on gains and minimal penalties for losses, push managers into such risky and reckless behavior. We discuss some speciﬁc ways losses occur. To illustrate, we discuss the speciﬁc cases of Long Term Capital Management, Niederhoﬀer’s hedge fund, Amaranth and Société Générale. In some cases, the failures lead to contagion in other hedge funds and ﬁnancial institutions. We also list other hedge fund and bank trading failures with brief comments on them. Keywords: Hedge fund trading disasters, over-betting, Long Term Capital Management, Amarath and Société Générale. Understanding How to Lose Helps One Avoid Losses! We begin by discussing how to lose money in derivatives which leads to our discussion of hedge fund disasters and how to prevent them. The derivative futures industry deals with products in which one party gains what the other party loses. These are zero sum games situations. Hence there will be large winners and large losers. The size of the gains and losses are magniﬁed by the leverage and overbetting, leading invariably to large losses when a bad scenario occurs. This industry now totals over $700 trillion of which the majority is in interest and bond derivatives with a s; maller, but ∗ Some of the material in this chapter is adapted from chapters in Ziemba and Ziemba (2013) which were modiﬁed updates of columns originally published in Wilmott. 689 page 689 July 3, 2015 8:28 690 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. substantial, amount in equity derivatives. Figlewski (1994) attempted to categorize derivative disasters and this chapter discusses and expands on that: 1. Hedge In an ordinary hedge, one loses money on one side of the transaction in an eﬀort to reduce risk. To evaluate the performance of a hedge one must consider all aspects of the transaction. In hedges where one delta hedges but is a net seller of options, there is volatility (gamma) risk which could lead to losses if there is a large price move up or down and the volatility rises. Also accounting problems can lead to losses if gains and losses on both sides of a derivatives hedge are recorded in the ﬁrm’s ﬁnancial statements at the same time. 2. Counterparty default Credit risk is the fastest growing area of derivatives and a common hedge fund strategy is to be short overpriced credit default derivatives. There are many ways to lose money on these shorts if they are not hedged correctly, even if they have a mathematical advantage. In addition, one may lose more if the counterpart defaults because of fraud or following the theft of funds, as was the case with MF Global. 3. Speculation Derivatives have many purposes including transferring risk from those who do not wish it (hedgers) to those who do (speculators). Speculators who take naked unhedged positions take the purest bet and win or lose monies related to the size of the move of the underlying security. Bets on currencies, interest rates, bonds, and stock market index moves are common futures and futures options trades. Human agency problems frequently lead to larger losses for traders who are holding losing positions that if cashed out would lead to lost jobs or bonus. Some traders increase exposure exactly when they should reduce it in the hopes that a market turnaround will allow them to cash out with a small gain before their superiors ﬁnd out about the true situation and force them to liquidate. Since the job or bonus may have already been lost, the trader’s interests are in conﬂict with objectives of the ﬁrm and huge losses may occur. Writing options, and more generally selling volatility or insurance, which typically gain small proﬁts most of the time but can lead to large losses, is a common vehicle for this problem because the size of the position accelerates quickly when the underlying security moves in the wrong direction as in the Niederhoﬀer example below. Since trades between large institutions frequently are not page 690 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 691 collateralized mark-to-market, large paper losses can accumulate without visible signs such as a margin call. Nick Leeson’s loss betting on short puts and calls on the Nikkei is one of many such examples. The Kobe earthquake was the bad scenario that bankrupted Barings. A proper accounting of trading success evaluates all gains and losses so that the extent of some current loss is weighed against previous gains. Derivative losses should also be compared to losses on underlying securities. For example, from January 3 to June 30, 1994, the 30-year T-bonds fell 13.6%. Hence holders of bonds lost considerable sums as well since interest rates rose quickly and signiﬁcantly. 4. Forced liquidation at unfavorable prices Gap moves through stops are one example of forced liquidation. Portfolio insurance strategies based on selling futures during the October 18, 1987 stock market crash were unable to keep up with the rapidly declining market. The futures fell 29% that day compared to −22% for the S&P500 cash market. Forced liquidation due to margin problems is made more diﬃcult when others have similar positions and predicaments and this leads to contagion. The August 1998 problems of Long Term Capital Management (LTCM) in bond and other markets were more diﬃcult because others had followed their lead with similar positions. When trouble arose, buyers were scarce and sellers were everywhere. Another example is Metallgellschaft’s crude oil futures hedging losses of over $1.3 billion, which is discussed below. They had long-term contracts to supply oil at ﬁxed prices for several years. These commitments were hedged with long oil futures. But when spot oil prices fell rapidly, the contracts to sell oil at high prices rose in value but did not provide current cash to cover the mark-to-the-market futures losses. A management error led to the unwinding of the hedge near the bottom of the oil market and the disaster. Potential problems are greater in illiquid markets. Such positions are typically long-term and liquidation must be done matching sales with available buyers. Hence, forced liquidation can lead to large bidask spreads. Askin Capital’s failure in the bond market in 1994 was exacerbated because they held very sophisticated securities which were only traded by very few counterparties, so contagion occurred. Once they learned of Askin’s liquidity problems and weak bargaining position, they lowered their bids even more and were then able to gain large liquidity premiums. 5. Misunderstanding the risk exposure As derivative securities have become more complex, so has their full understanding. The Shaw et al. (1995) Nikkei put warrant trade page 691 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 692 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba (discussed in Ziemba and Ziemba (2013), Chapter 12) was successful because we did a careful analysis to fairly price the securities. In many cases, losses are the result of trading in high-risk ﬁnancial instruments by unsophisticated investors. Lawsuits have arisen by such investors attempting to recover some of their losses with claims that they were misled or not properly briefed on the risks of the positions taken. Since the general public and thus judges and juries ﬁnd derivatives confusing and risky, even when they are used to reduce risk, such cases or their threat may be successful. A great risk exposure is the extreme scenario which often investors assume has zero probability when in fact they have low but positive probability. Investors are frequently unprepared for interest rate, currency or stock price changes so large and so fast that they are considered to be impossible to occur. The move of some bond interest rate spreads from 3% a year earlier to 17% in August/September 1998 led even savvy investors and very sophisticated Long Term Capital Management researchers and traders down this road. They had done extensive stress testing with a VaR risk model which failed as the extreme events such as the August 1998 Russian default had both the extreme low probability event plus changing correlations. Several scenario-dependent correlation matrices rather then simulations around the past correlations from one correlation matrix is suggested. This is implemented, for example, in the Innovest pension plan model which does not involve levered derivative positions (see Ziemba and Ziemba, 2013, Chapter 14). The key for staying out of trouble especially with highly levered positions is to fully consider the possible futures and have enough capital or access to capital to weather bad scenario storms so that any required liquidation can be done orderly. Figlewski (1994) observes that the risk in mortgage-backed securities is especially diﬃcult to understand. Interest only (IO) securities, which provide only the interest part of the underlying mortgage pool’s payment stream, are a good example. When interest rates rise, IOs rise since payments are reduced and the stream of interest payments is larger. But when rates rise sharply, the IO falls in value like other ﬁxedincome instruments because the future interest payments are more heavily discounted. This signal of changing interest rate exposure was one of the diﬃculties in Askin’s losses in 1994. Similarly the sign change between stocks and bonds during stock market crashes as in 2000 to 2003 has caused other similar losses. Scenario-dependent matrices are especially useful and needed in such situations. page 692 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 693 6. Forgetting that high returns involve high risk If investors seek high returns, they will usually have some large losses. The Kelly criterion strategy and its variants provide a theory to achieve very high long-term returns, but large losses will also occur. These losses are magniﬁed with derivative securities and especially with large derivative positions relative to the investor’s available capital. 7. How over-betting occurs Figure 22.1 shows how the typical over-bet situation occurs assuming a Kelly strategy is being used. The top of the growth rate curve is at the full Kelly bet level that is the asset allocation maximizing the expected value of the log of the ﬁnal wealth subject to the constraints of the model. To the left of this point are the fractional Kelly strategies which under a lognormal asset distribution assumption use a negative power utility function rather than log. So αwα for α < 0 gives the fractional Kelly 1 1 . So u(w) = −1 weight f = 1−α w corresponds to 2 Kelly with α = −1. Overbetting is to the right of the full Kelly strategy and it is clear that betting more than full Kelly gives more risk measured by the probability of reaching a high goal before a lower level curve on the ﬁgure. It is this area way to the right where over-betting occurs. And virtually all of the disasters occur because of the over-betting. Figure 22.1. Relative growth and probabilities of doubling, tripling, and quadrupling initial wealth for various fractions of wealth bet for the gamble win $2 with probability 0.4 and lose $1 with probability 0.6. page 693 July 3, 2015 8:28 694 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. It is easy to over-bet with derivative positions as the size depends on the volatility and other parameters and is always changing. So a position safe one day can become very risky very fast. A full treatment of the pros and cons of Kelly betting is in Ziemba (2014). Stochastic programming models provide a good way to try to avoid problems 1–6 by carefully modeling the situation at hand and considering the possible economic futures in an organized way. Hedge fund and bank trading disasters usually occur because traders over-bet, the portfolio is not truly diversiﬁed and then trouble arises when a bad scenario occurs. We now discuss a number of sensational failures including Metalgesllshart (1993), LTCM (1998), Niederhoﬀer (1997), Amaranth Advisors (2006), Merrill Lynch (2007), Société Générale (2008), Lehman (2008), AIG (2008), Citigroup (2008), MF Globl (2012), and Monti dei Paschi (2013). Stochastic programming models provide a way to deal with the risk control of such portfolios using an overall approach to position size, taking into account various possible scenarios that may be beyond the range of previous historical data. Since correlations are scenario-dependent, this approach is useful to model the overall position size. The model will not allow the hedge fund to maintain positions so large and so under-diversiﬁed that a major disaster can occur. Also the model will force consideration of how the fund will attempt to deal with the bad scenario because once there is a derivative disaster, it is very diﬃcult to resolve the problem. More cash is immediately needed and there are liquidity and other considerations. Ziemba and Ziemba (2013, Chapter 14) explores more deeply such models in the context of pension fund as well as hedge fund management. Litzenberger and Modest (2009), who were on the ﬁring line for the LTCM failure, propose a modiﬁcation of standard ﬁnance CAPM type theory modiﬁed for fat tails and CVaR or expected tail losses for the losses. Ziemba (2003, 2007, 2013) presents his approach using convex risk measures and three scenario-dependent correlation matrices depending upon volatility using stochastic programming scenario optimization. Both of these approaches would mitigate such losses. The key is not to over-bet and have access to capital once a crisis occurs and to plan in advance for such events. The Failure of Long Term Capital Management (1998) There have been many hedge fund failures but LTCM stands out as a particularly public one. The ﬁrm started with the talents of the core bond page 694 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 695 traders from John Merriwether’s group at the Salomon Brothers who were very successful for a number of years. When Warren Buﬀett came on board at Salomon, the culture of this group clashed with Buﬀett’s apparently more conservative style. In truth, Buﬀett’s record is Kelly like (see Ziemba, 2014) and not all that diﬀerent from Merriwether’s group in terms of position size but Buﬀett’s risk control is superior. He always has lots of cash to bail out any troubling trades. A new group was formed with an all star cast of top academics including two future Nobel Laureates and many top professors and students, many linked to MIT. In addition, top government oﬃcials were involved. The team was dubbed too smart to lose and several billion was raised even though there was no real track record, fees were very high (25% of proﬁts plus a 3% management expense fee), and entry investment was $100 million minimum. The idea, according to Myron Scholes, was to be a big vacuum cleaner sucking up nickels all over the world as the cartoon suggests. There were many trades, but the essence of the bond risk arbitrage was to buy underpriced bonds in various locales, sell overpriced bonds in other locales, wait for the prices to revert to their theoretical eﬃcient market prices, and then unwind the position. These trades are similar to the Nikkei put warrant risk arbitrage Thorp and Ziemba did except that the leverage they used was much much greater (Shaw et al., 1995). We can call these bond trades buy Italy and sell Florence. As shown in the graph, the interest rate implied by the bond prices is higher in Italy than in Florence. But the theory is that Florence, a smaller place, would have more risk. Hence, the trade should have an advantage and be unwound when the prices reverted to their true risk priced values. LTCM analysts made many such trades, most much more complex than this, all across the world. They also had many other complex and innovative trades. Their belief was that markets were eﬃcient and, when temporarily out of whack, would snap back quickly. The continuous lognormal assumptions of option pricing hedging led them to take very large positions which according to their theory were close to riskless. page 695 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 696 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba The plan worked and net returns for the part of the year 1994 that the fund operated were 19.9% net. The years 1995 and 1996 had similar high returns of 42.8% and 40.8% net, respectively. Indeed for the principals whose money grew fee-less, the net returns were 63% and 57%, respectively, with taxes deferred. There was so much demand for investment in the fund, which in 1997 was eﬀectively closed to new investors, that a grey market arose with a 10% premium. By 1997, it became harder to ﬁnd proﬁtable trades and the gains fell to 17.1%. This was a good record for most but not satisfactory to LTCM’s principals; among other things the S&P500 returned 31% excluding dividends. Their action was to return $2.7 billion of the $6.7 billion to the investors, a huge mistake! The principals then put in an additional $100 million raised by personal bank loans, another mistake. The banks were happy to lend this money basically unsecured. Banks and others were quite keen to loan to or invest with this group and the investors were not happy to be forced out of the fund. Still, at the start, $1 on February 24, 1994, was $2.40 net at the end of 1997. The year 1998 was diﬃcult for the fund and then turned into a disaster following the August 17 Russian ruble devaluation and sovereign bond default. Bonds denominated in rubles trading for say 60 fell rapidly to 3 whereas Russian bonds denominated in marks or dollars only fell a few percent as they were not subject to the eﬀects of the ruble devaluation. So long 60 short 95 say became long 3 short 92 say. Also there were defaults in currency hedging contracts which added to the losses because that hedge failed. Such losses occur from time to time in various markets, and hedge funds which over-bet can be particularly vulnerable to it. The problem for LTCM was that they had $1.25 trillion of positions in notional value (that was over 2% of the world’s derivatives in 1998) with a market value of $129 billion ﬁnanced by $125 billion of borrowed money. They had $4 billion in equity with a leverage ratio of 32. Although the trades were all over the world and hence it seemed they were diversiﬁed, they in fact were not. What happened was a scenario-dependent correlation situation like that modeled in the Innovest pension fund application described in Ziemba and Ziemba (2013), Chapter 14. There was an underlying variable that frequently lurks its ugly head in disasters that being investor conﬁdence. The graph on the side illustrates the problem: all the bond rates increased for non-high quality debt. For example, emerging market debt was trading for 3.3% above U.S. T-bonds in October 1997, then 6% in July 1998 and then an astounding 17% in September 1998. page 696 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 How to Lose Money in Derivatives 697 8 6 4 2 Source: Salomon Smith Bamey The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. J F M A M J J A S O 0 The spread between emerging market debt and U.S. T-bonds, 1998 LTCM was unable to weather the storm of this enormous crisis of conﬁdence and lost about 95% of their capital, some $4.6 billion including most of the principals’ and employees’ considerable accumulated fees. The $100 million loan actually put some of them into bankruptcy, although others came out better ﬁnancially. It did not help that they unwound liquid positions ﬁrst rather than across all liquidity levels as the two Nobel Prize winners Robert Merton and Myron Scholes recommended, nor that many other copy-cat ﬁrms had similar positions, nor that LTCM had created enemies by being so skilled and so brash, nor that the lack of monitoring of margin by brokers eager for their business allowed the positions to grow to over-bet levels. A pivotal decision was returning $2.7 billion to investors. They could have kept the funds in liquid low-risk assets to buﬀer their mounting losses. However had they kept the funds they might have made even more risky plays. Returning this money reﬂected their greediness. They simply wanted to make a higher rate of return with similar positions on a smaller capital base.1 Smart people bounce back and possibly learn from their mistakes. Various ex-LTCM members have joined new hedge funds and other ventures. The lessons are: • Do not over-bet, it is too dangerous. • VaR type systems are inadequate to measure true risk but see Jorion’s (2007) excellent book on VaR and Dunbar’s (2000) discussion of the VaR calculations used by LTCM. LTCM analysts did a very careful analysis but the problem was that the risk control method of VaR which is used in 1 Using the Kelly criterion, you should never bet more than the log optimal amount and betting more (as LTCM did) is dominated as it has lower growth rates and higher risk. This point is not understood by even the top academic ﬁnancial economists who insist on using positive power as well as negative power and log utility functions. The positive power ones are dominated and reﬂect over-betting. page 697 July 3, 2015 8:28 698 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. • • • • • World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba regulations does not really protect hedge funds that are so highly levered because you are not penalized enough for large losses. Indeed if you lose $10 million, it is penalized the same as losing $100 million if the VaR number is $9 million of losses. CVaR partially addresses this limitation but what you really need are convex penalties so that penalties are more than proportional to losses. You really do need to use scenario-dependent correlation matrices and consider extreme scenarios. LTCM was not subject to VaR regulation but still used it. Be aware of and consider extreme scenarios. Allow for extra illiquidity and contract defaults. LTCM also suﬀered because of the copycat ﬁrms which put on similar positions and unwound them at the same time in August/September 1998. Really diversify (to quote Soros from the Quantum Funds, “we risked 10% of our funds in Russia and lost it, $2 billion, but we are still up 21% in 1998”). Historical correlations work when you do not need them and fail when you need them in a crisis when they approach one. Real correlations are scenario-dependent. Sorry to be repetitive, but this is crucial. Good information on the demise of LTCM and the subsequent $3.5 billion bailout by major brokerage ﬁrms organized by the FED are in a Harvard Business School case by André Perold (1998), and articles by Philippe Jorion (2000) and Franklin Edwards (1999). Eventually the positions converged and the bailout team was able to emerge with a proﬁt on their investment. The currency devaluation of some two-thirds was no surprise to WTZ. In 1992, we were the guests in St. Petersburg of Professor Zari Rachev, an expert in stable and heavy-tail distributions and editor of the ﬁrst handbook in North Holland’s Series on Finance (Rachev, 2003) of which WTZ was the series editor. On arrival, I gave him a $100 bill and he gave me four inches of 25 Ruble notes. Our dinner out cost two inches for the four of us; and drinks were extra in hard currency. So we are in the Soros camp; make bets in Russia (or similar risky markets) if you have an edge without risking too much of your wealth. Where was the money lost? The score card according to Dunbar (2000) was a loss of $4.6 billion. Emerging market trades such as those similar to the buy Italy, sell Florence lost $430 million. Directional, macro trades lost $371 million. Equity pairs trading lost 306 million. Short long-term equity options, long short-term equity lost $1.314 billion. Fixed income arbitrage lost $1.628 billion. page 698 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 699 The bad scenario of investor conﬁdence that led to much higher interest rates for lower quality debt and much higher implied equity volatility had a serious eﬀect on all the trades. The long-short equity options trades, largely in the CAC40 and Dax equity indices, were based on a historical volatility of about 15% versus implieds of about 22%. Unfortunately, in the bad scenario, the implieds reached 30% and then 40%. With smaller positions, the fund could have waited it out but with such huge levered positions, it could not. Equity implieds can reach 70% or higher as Japan’s Nikkei did in 1990/1991 and stay there for many months. Niederhoﬀer’s Hedge Fund Disaster and the Imported Crash of October 27 and 28, 1997 The Asian Financial crises was a series of banking and currency crises that developed in various Asian countries beginning in mid-1997. Many East and Southeast Asian countries had currency pegs to the U.S. dollar which made it easy for them to attract ﬁnancing but lacked adequate foreign reserves to cover the outstanding debt. Their pegs to the U.S. dollar and low interest rates encouraged mismatches in currency (debts were in U.S. dollars, loans in local currency) and maturities. Spending and expectations that led to borrowing were too high and Japan, the main driver of these economies, was facing a consumer slowdown so its imports dropped, so that eﬀectively these countries were long yen and short dollars. A large increase in the U.S. currency in yen terms exacerbated the crisis, which began after speculators challenged the Thai Baht and spread through the region. The countries had to devalue their currencies, interest rates rose, and stock prices fell. Also, several hedge funds took signiﬁcant losses. Most notably, Victor Niederhoﬀer’s fund, which had an excellent previous record with only modest drawdowns, but his large long bet on cheap Thai stocks that became cheaper and cheaper quickly turned $120 million into $70 million. Further buying on dips added to losses. Finally the fund created a large short position in outof-the-money S&P futures index puts including the November 830’s trading for about $4–6 at various times around August–September 1997. The crisis devastated the economies of Malaysia, Singapore, Indonesia, etc. Finally it spread to Hong Kong, where the currency was pegged to the U.S. dollar at 7.8. The peg supported Hong Kong’s trade and investment hub and was to be defended at all costs. In this case, the weapon used was higher interest rates which almost always lead to a stock market crash after a lag; see Lleo and Ziemba (2012). The U.S. S&P500 was not in the danger page 699 July 3, 2015 8:28 700 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba zone in October 1997 by WTZ’s models nor, we presume, by those of others. Also, trade with Hong Kong and Asia, though substantial, was only a small part of the U.S. trade. Many U.S. investors thought that this Asian currency crisis was a small problem because it did not aﬀect Japan very much. In fact, Japan caused a lot of it. The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. A wild week, October 20−25, 1997 The week of October 20–25, 1997 was diﬃcult for equity markets with the Hang Seng dropping sharply. The S&P was also shaky. The November 830 puts were 60 cents on Monday, Tuesday, and Wednesday but rose to 1.20 on Thursday and 2.40 on Friday. The Hang Seng dropped over 20% in a short period including a 10% drop on Friday, October 25. The S&P500 was at 976 substantially above 830 as of Friday’s close. A further 5% drop in Hong Kong on Monday, October 27 led to a panic in the S&P500 futures later on Monday in the United States. They fell 7% from 976 to 906 which was still considerably above 830. On Tuesday morning, there was a further fall of 3% to 876 still keeping the 830 puts out of the money. The full fall in the S&P500 was then 10%. But the volatility Exploded and the 830’s climbed to the $16 area. Refco called in Niederhoﬀer’s puts mid-morning on Tuesday, resulting in the fund losing about $20 million. So Niederhoﬀer’s $70 million fund was bankrupt and actually in the red as the large position in these puts and other instruments turned the $70 million into minus $20 million. The S&P500 bottomed out around 876, moving violently in a narrow range, then settling. By the end of the week, it returned to the 976 area. So it really was a tempest in a teapot. Investors who were short equity November 830 puts (SPXs) were required to put up so much margin that they had small positions so they weathered the storm. Their $4–$6, while temporarily behind at $16, did eventually go to zero. So did the futures puts, but futures shorters are not required to post as much margin. If they did not have adequate margin because they had too many positions, they could have easily been forced to cover at a large loss. Futures margins, at least for equity index products, do not fully capture the real risk inherent in these positions. We follow closely the academic studies on risk measures and none of the papers we know address this issue properly. When in doubt, always bet less. Niederhoﬀer is back in business having proﬁted by this experience. (Whoops — maybe not, see the postscript!) One of Ziemba’s Vancouver neighbors lost $16 million in one account and $4 million in another account. The diﬀerence being the time given to cash page 700 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 701 out and cover the short puts. Ziemba was in this market also and won in the equity market and lost in futures. I did learn how much margin you actually need in futures which now Ziemba uses in such trading which has been very proﬁtable with a few proprietary wrinkles to protect oneself that he needs to keep conﬁdential. A hedged strategy had a 45% geometric mean with 72 of 77 winners with six quarters ruled too risky by an option price market sentiment danger control measure out of the 83 possible plays in those 22 years and a seven symmetric downside Sharpe ratio. Ruling out the six risky quarters, one of the naked strategies won 76 out of 77 times from 1985 to 2006. In those six quarters, the S&P500 actually fell in four. The cumulative S&P500 loss in the six quarters was −41.7%. The lessons for hedge funds are much as with LTCM. Do not over-bet, do diversify, watch out for extreme scenarios. Even the measure to keep one out of potentially large falls mentioned above did not work in October 1997. That was an imported fear-induced stock market crash which was not really based on the U.S. economy or investor sentiment. Most crashes occur when interest rates relative to price earnings ratios are too high. Almost always when that happens, there is a crash (a 10% plus fall in equity prices from the current price level within one year); see Ziemba (2003) and Lleo and Ziemba (2012) for the 1987 U.S., the 1990 Japan, the U.S. in 2000, the U.S. in 2001, which predicted the 22% fall in the S&P500 in 2002 and China, Iceland and the U.S. in 2006–2009 are leading examples. Interestingly the measure moved out of the danger zone following the 2000 crash. Then, in mid-2001, it was even more in the danger zone than in 1999 because stock prices fell but earnings fell more. In 2003, the measure then moved into the buy zone and predicted the rise in the S&P500 in 2003. No measure is perfect but this measure adds value and tends to keep you out of extreme trouble. When long bond interest rates get too high relative to stock returns as measured by the earnings over price yield method, there almost always is a crash. Ziemba–Schwartz (1991) used a diﬀerence method and the results of that are in Ziemba (2003). Ziemba started using these measures in 1988 in my study group at Yamaichi Research, Japan. The study predicted the 1987 crash. It also predicted the 1990 Japan crash. Ziemba told Yamaichi executives about this in 1989, but they would not listen. Yamaichi went bankrupt in 1995; they would have survived if they had listened.2 From 2 They could have paid WTZ a million dollars for an hour’s consulting and still made more than 1000 times proﬁt from the advice. It was more important for them to be nice to his family and him as they were than to listen to the results of a gaijin professor. How could page 701 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 702 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba 1948 to 1988, every time that the measure was in the danger zone there was a fall of 10% or more with no misses. This was 12 of 12 with 8 other 10%+ crashes occurring for other reasons than high interest rates relative to earnings. In late 1989, the model had the highest reading ever in the danger zone and predicted the January 1990 start of the Japanese stock market crash. A mini-crash caused by some extraneous event can occur any time. So to protect oneself, positions must never be too large. Koliman (1998) and Crouhy, Galai, and Mark in Gibson’s (2000) book on model risk discuss this. Their analysis suggests it is a violation of lognormality which I agree it was. Those who had too many positions had to cash out and suﬀer large losses because they had to satisfy the increased margin required due to the drop in price and the increase in implied volatility. Some good references on hedge fund performance, risk and incentives follow for further reading. Kouwenberg and Ziemba (2007) using a continuous time model with a prospect theory S-shaped objective, where losses are more damaging than gains are good, study the eﬀect of incentives on hedge fund manager behavior. The incentive fee encourages managers to take excessive risk but that risk is much less if the fund manager has a substantial amount of their own money in the fund (at least 30%). This suggests that investors should look for funds where the managers eat their own cooking.3 Their empirical results indicate that hedge funds with incentive fees have higher downside risk than funds without such a compensation contract. Average net returns, both absolute and risk-adjusted, are signiﬁcantly lower in the presence of incentive fees. So pick your managers well. An incentive fee is tantamount to a call option on the value of the investor’s assets. Goetzmann, Ingersoll and Ross (2003) and Kouwenberg and Ziemba (2007) show how to calculate the value of that option. The value depends directly on the manager’s optimal investment style with values ranging from 0 (with no investment) to 17% (with 30%+ share) of the investor’s capital. he possibly understand the Japanese stock market? In fact all the economics ideas were there; see Ziemba and Schwartz (1991). WTZ did enjoy these lectures, dinners, and golf but being listened to dominates. 3 But the manager’s personal share of the fund may decline in percentage term as the fund grows! page 702 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 How to Lose Money in Derivatives 703 Overbetting Yields Frequent Trading Disasters “The best way to achieve victory is to master all the rules for disaster, and then concentrate on avoiding them. In America, people get a second chance . . . they don’t get a third.” The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. Victor Niederhoﬀer After Niederhoﬀer’s failure in 1997, his fund was closed and he lost much of his personal fortune, reputation, and happiness. He had failed in 1997 because he greatly over-bet, did not diversify, and a bad scenario wiped him out. Was this a one-time occurrence from which he learned or is it just one of a sequence of similar outcomes? Niederhoﬀer is a multi-talented individual graduating with a PhD in 1969 from the Graduate School of Business, at the University of Chicago where Professor Gene Fama, Merton Miller, and other great ﬁnance theorists and practitioners are on the faculty. Since his work was against the prevailing eﬃcient markets theory and was highly data-dependent, he was more comfortable with the statisticians and was supervised by perhaps the world’s top Bayesian statistician, Arnold Zellner. Earlier at Harvard, his senior thesis “Non-randomness in stock prices: A new model of price movements” challenged random walk theory. He argued that stocks followed patterns such as Monday falls if Friday fell. In 1967, with his PhD thesis unﬁnished and the title “U.S. top squash player,” he headed to the ﬁnance department of the University of California, Berkeley Business School. WTZ was there then as well but never met Niederhoﬀer, being a graduate student. Victor was also a whiz at chess and tennis, dating back to his Harvard undergraduate days. WTZ was friendly with one ﬁnance legend Professor Barr Rosenberg who went on to greatness in a number of investment areas such as founding the Berkeley Program in Finance, the ﬁrm BARRA and later Rosenberg Investments. Both Barr and Victor, like WTZ, were looking for anomalies to beat markets. In 1967, Barr discovered that small caps and low price to book stocks outperformed the broad market. This observation forms the basis of the famous Fama-French (1992) factors 25 years later; see Rosenberg, Reid and Lanstein (1985). While Barr stuck to institutional investing with low or no leverage, Victor was a high-stakes futures trader using lots of leverage. Hence, if he was right, then the gains were very high but if he was wrong and his risk control was faulty, then there could be substantial losses. page 703 July 3, 2015 APR 280.26 6.46% 304.72 7.58% 346 4.70% 1.87% 123.72 3.76% 125.72 122.02 132.43 139.59 149.88 151.81 164.34 194.05 219.11 236.75 236.75 1.62% −12.70% 20.66% 5.40% 7.72% 1.29% 8.25% −5.42% 10.05% 6.59% 56.28% 2004 56.37 1.22% 54.83 8.35% 56.85 0.20% 60.21 5.17% 61.96 4.04% 63.3 2.15% 65.28 3.22% 67.49 3.38% 75.02 2.98% 77.95 105.28 110.17 110.17 3.90% 3.26% 3.67% 50.13% 2003 21.15 2.47% 31.13 7.70% 35.17 1.65% 40.46 1.76% 41.9 3.54% 42.94 2.48% 45.26 4.75% 46.11 1.32% 46.44 0.52% 49.18 5.91% 50.08 1.21% 55.69 1.48% 55.69 40.55% 2.03 1.71% 2.92 2.71% 3.64 −0.65% 5.5 11.81 8.65 8.12 10.01% −1.16% −30.22% −6.81% 8.88 9.29% 10.22 15.20% 10.66 3.55% 12.54 7.86% 12.54 3.12% 1.94 −3.10% 2.72 11.68% 6.6 7.82 9.69% −4.36% 8.99 14.98% 11.07 5.53% 1.07 53.23% Matador Fund 2006 261.24 9.59% 2005 112.24 2002 MAY JUNE JUL AUG SEP OCT NOV DEC YRTTL Manchester Fund 2006 13.9 10.61% 2005 n/a n/a 15.90 7.44% n/a n/a 17.32 8.24% 2 n/a Source: Manchester Trading, LLC (2006). 4.4 4.09% 4.73 5.86 7.50% −0.59% 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. MAR World Scientific Handbook of Futures Markets. . . FEB S. Lleo and W. T. Ziemba JAN 8:28 704 Table 22.1. Performance of the Matador Fund, February 2002–April 2006 and Manchester Fund from March 2005–April 2006. First line is Assets, MM, second line is Monthly Return. page 704 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 705 While teaching at Berkeley, Victor co-founded a small investment bank, Niederhoﬀer, Cross and Zeckhauser (NCZ). Frank Cross was a former Merrill Lynch executive and Richard Zeckhauser, a friend from his Harvard days. Zeckhauser went on to become a well known economist at the Kennedy School of Government and an avid bridge player. NCZ started with just $400, did mail-order mergers, and sold small private companies to buyers. In 1979, Niederhoﬀer went into commodities and had great success, averaging 35% net for 15 years through the mid-1990s. George Soros gave him a private $100 million account in 1981 and Niederhoﬀer traded that until 1993. This account was shut down because, as Soros said, “he temporarily lost his edge . . . he made money while the markets were sloshing along aimlessly. Then he started losing money and had the integrity to close out the account. We came out ahead.” Earlier in 1983, Zeckhauser had quit NCZ to return to full-time teaching and research partially because of Niederhoﬀer’s high level of risk taking, saying that “no matter what your edge, you can lose everything. You hope and believe he will learn his lesson.” Cross died and NCZ became Niederhoﬀer Henkel and was then run by Lee Henkel, the former general council for the IRS. After the 1997 blowout, it was hard for Niederhoﬀer to start again as there was fear of another large drawdown despite his long superior track record. So he began trading for his own account after mortgaging his house. In 2000 he started writing investment columns on websites with Laurel Kenner and in 2001 it paid oﬀ. Mustafa Zaida, a Middle East investor, set up the oﬀshore hedge fund Matador with $2 million and recruited Niederhoﬀer as the trading advisor. To reign in Niederhoﬀer’s exuberance for risk, the fund would invest only in U.S.-based S&P500 futures and options. The claim was that Niederhoﬀer had learned his lesson not to invest in markets he did not understand like Thailand which got him on the road to destruction in 1997. A management fee of 2.5% + 22% of the net new proﬁts was substantial. Yet with good performance, Matador grew to $350 million from non-U.S. investors. Zaida said that “He’s deﬁnitely learned his lesson.” Recall that it was the S&P500 November largely 830 puts that turned $70 million into −$20 million in 1997 after $50 million was lost in Thai equities. Niederhoﬀer always thinks big and bold, so Matador was not enough. In April 2005, Niederhoﬀer started Manchester Partners, LLC for U.S. investors, named for the Silver Cup given to the winner of the Manchester Cup Steeplechase in 1904. This trophy was one of the many art objects Niederhoﬀer had collected over the years and hung onto. Manchester’s fees were 1%+20%, and could trade other than the S&P500 market such as ﬁxed income and page 705 July 3, 2015 8:28 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba currencies. Steve “Mr Wiz” Wisdom was Niederhoﬀer’s risk control aide, hoping to have consistent 25%+ returns with maximum losses of 15–20% in one month. The bond-stock crash measure (Ziemba, 2003; Lleo and Ziemba, 2012), ﬂagged a red signal at the end of 2001 because earnings dropped more than stock prices. Ziemba’s conﬁdential investor sentiment model based on relative put/call option prices ﬂashed red in Q4 of 2002. And indeed there was a substantial fall in the S&P500 in July 2002; Matador lost 30.22% in that month. Still the February 2002 to April 2006 Matador record was a +338% gain, 41% net annualized, $350 million in assets and only 5 losses in 51 months, with a 2.81 Sharpe ratio (see Figure 22.2). This record earned Matador the #1 ranking in 2004, 2005, and 2006 for funds managing $50+ million (see Table 22.2). Manchester had only three monthly losses in the 13 months from its start in April 2005 to April 2006, a cumulative gain of 89.9%. The approach had the following elements (from Manchester Trading, 2006): Scientiﬁc Rigorous statistical methodologies form the foundation of our proprietary pattern recognition process. 400.00 350.00 300.00 250.00 200.00 150.00 100.00 Manchester 50.00 b Ap -02 r Ju -02 n Au -02 g O -02 ct D -02 ec Fe -02 b Ap -03 r Ju -03 nAu 03 g O -03 ct D -03 ec Fe -03 b Ap -04 r Ju -04 n Au -04 g O -04 ct D -04 ec Fe -04 b Ap -05 r Ju -05 n Au -05 g O -05 ct D -05 ec Fe -05 b Ap -06 r06 - Fe The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 706 World Scientific Handbook of Futures Markets. . . Figure 22.2. Performance of the Matador fund, February 2002–April 2006 and Manchester fund from March 2005–April 2006. Source: Manchester Trading, LLC (2006). page 706 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 How to Lose Money in Derivatives 707 30.00% Manchester 20.00% 10.00% -10.00% -20.00% b Ap -02 r Ju -02 nAu 02 g O -02 ct D -02 ec Fe -02 b Ap -03 r Ju -03 n Au -03 g O -03 ct D -03 ec Fe -03 b Ap -04 r Ju -04 n Au -04 g O -04 ct D -04 ec Fe -04 b Ap -05 r Ju -05 n Au -05 g O -05 ct D -05 ec Fe -05 b Ap -06 r06 -30.00% Fe The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 0.00% Figure 22.2b. Table 22.2. (Continued). Ranking of Manchester trading. 2006 #1 performing CTA MarHedge MAPA. 2005 #1 oﬀshore managed futures fund (Tass/Lipper) for funds managing more than $50 million. 2004 #1 oﬀshore managed futures (Tass/Lipper) for funds managing more than $50 million. Cumulative +338% since inception in February 2002: Assets under management $350+ million. Source: Manchester Trading, LLC (2006). Empirical “What can be tested, must be tested.” Validation through testing is the basis for all trade recommendations, impact planning, and margin assessment. Innovative Multidisciplinary inquiry draws from such diverse ﬁelds as speech processing, information theory, and data compression to provide insight and inspiration. Contrarian Crowd behavior tends to create proﬁtable opportunities. We are more often than not counter-trend traders. Focused Undiluted application of our edge leaves the critical diversiﬁcation decision in the hands of our investors. page 707 July 3, 2015 8:28 708 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba For short-term discretionary day trading: • • • • Systematic identiﬁcation of high-probability trades Analysis across multiple markets and multiple time frames Flexible analytical methodology sensitive to changing cycles Tactical execution reducing friction and slippage The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. And for the option trading: • Empirical option pricing vs. implied volatility method • Strategic/opportunistic seller of expensive premium • Forecasting techniques applied to margin pathways enhancing risk modeling • Flexible position across multiple strikes and timeframes • Highly sensitive ongoing measurement of overall liquidity and margin pathway forecasting reﬁnes leverage assessment And what did they learn from the 1997 blowout? • • • • We learned our lesson and got back on our feet fast. We stick with markets and instruments we know. We focus on liquidity. We are alert to the increasing probability of extreme events, measure their potential impact, and prepare for them. • We implement safeguards and continue to reﬁne trading and risk assessment procedures to ensure survival. They say it cannot happen again because • We tailor our risk proﬁle at all times cognizant of the impact and opportunity extreme events can bring about. • We are constantly innovating but remain focused on what works empirically. We don’t stray from our core strategy. • Substantial co-investment by the principals of the ﬁrm is the most powerful statement we can possibly make with regard to our long-term commitment to our partners. Manchester does not like to diversify and their literature says that We choose not to diversify or manage the volatility of our fund to a benchmark or index as we believe our clients and their asset allocation advisors are in a far better position to make accurate and economical diversiﬁcation decisions than we are. (Manchester, 2006) Niederhoﬀer has historically had a long bias in his trades which are frequently unrad with 3–6 times leverage with borrowed money. page 708 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 709 On May 10, 2006, the Russian New Europe (RNE) fund, was trading at a 37% premium to net asset value according to the Barron’s. RNE treated WTZ well over the years with high returns and generous capital gains and dividends. But a 37% premium was extraordinary. The bond-stock model and the short-term investor sentiment option models WTZ uses were both way out of the danger zone and did not predict the subsequent decline. That weekend was a local peak and the S&P500 fell about 7% in the next month with many emerging markets falling 20%+. RNE fell more about 40% to a no-premium level. The twig that got the equity markets going on the downside was the threat of higher Japanese interest rates. This caused some hedge funds with yen carry trades to unwind their positons which meant selling the S&P500 and emerging market equities. It also caused them to look closer at high-yielding emerging market currencies and bonds such as Turkey, South Africa, and Iceland. Although these have high yields, thus making them attractive for carry trades, they also have high current account deﬁcits. Investors feared both higher interest rates and a higher yen in which they had short positions. The Matador fund lost 30.22% in May turning a 2006 gain of 31% to −6% at the end of May. The market was down 3% but Niederhoﬀer was so leveraged that the loss was magniﬁed ten times to some $100 million. This hedge ratio of 10 means that Niederhoﬀer must have been massively long S&P500 futures and/or short S&P500 equity and/or futures puts. This is a huge long position that is not risk-control-safe and subject to large losses with a modest drop in the S&P500. A medium S&P500 drop (see below) would likely have led to losses in the 50% area and a large 10%+ drop to losses of 75%+. Niederhoﬀer said “I had a bad May. I made some mistakes, that’s regrettable . . . but one sparrow does not make a spring; and nor does one bad month.” June 2006 continued badly with the Matador fund down 12% for 2006. When the May to July debacle in the S&P500 ended it was down about 7% but Matador lost 67% and Manchester 45%. Both funds are still trading and the saga continues (see below). WTZ maintains the two rules: do not over-bet and do diversify in all scenarios. One can still make good gains in the S&P500 futures and options and other markets. But somewhat smaller than 30–40% gains are most likely but presumable without blowouts if one has position sizes such that the fund or account will weather a 3–7% decline in 1–4 days or a 10–15% decline over a month. Ziemba’s experience is that proper risk control in the S&P500 market, which is not diversiﬁed, can yield net gains in the 15% to perhaps 25% range but 30–40% seems attainable only with substantial risk that likely will cause page 709 July 3, 2015 8:28 710 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba Table 22.3. Manchester Partners net returns in various time periods versus the S&P500. Recent Returns Latest Month Last 3 Months Last 6 Months Last 12 Months Last 18 Months Inception 5-Apr Partners S&P500 20.44% 1.65% 21.44% 4.38% 80.26% 12.66% 8.40% 12.36% 51.72% 17.86% 83.72% 21.82% The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. Source: Manchester Trading, LLC (2006). a large loss if a bad scenario occurs. Of course, other strategies could yield such higher returns as Blair Hull, Jim Simons, Harry McPike, and others have shown. Niederhoﬀer was given a third chance after all! Table 22.3 shows the Manchester Partners returns to the end of January 2007. The May to July blowout is seen in the 8.4% returns in the last 12 months down from 89.9% as of April 2006. But the fund gained 20.44% in January 2007 and the April 5, 2005 to end January 2007 net returns are back to 83.72%, well above the S&P500. So Niederhoﬀer is back in business once again . . . perhaps till the next time. The Amaranth Advisors Natural Gas Hedge Fund Disaster (2006) On September 19, 2006 the hedge fund Amaranth Advisors of Greenwich, Connecticut, announced that it had lost $6 billion, about two-thirds of the $9.25 billion fund, in less than two weeks, largely because it was overexposed in the natural gas market. Amaranth’s experience shows how a series of trades can undermine the strategy of such a hedge fund and investors’ assets. The Greenwich, Connecticut fund which was founded in 2000, employed hundreds in a large investment space with other oﬃces in Toronto, London, and Singapore. We analyse how Amaranth became so overexposed, whether risk control strategies could have prevented the liquidation and how these trends reﬂect the current state of the ﬁnancial industry. We have argued that the recipe for hedge fund disaster almost always has three parts: 1. A trader over-bets relative to one’s capital and the volatility of the trading instruments used. 2. The trader is not diversiﬁed in all scenarios that could occur. 3. A negative scenario occurs that is plausible ex post and likely ex ante although the negative outcome may have never occurred before in the particular markets the fund is trading. page 710 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 711 One might expect that these two interrelated risk factors (1) and (2) would be part of the risk control assessment of hedge funds. These risks become more pronounced as the total amount of trading grows, especially when trading in billions. But are risks assessed this way? A knowledgeable risk control expert, realizing that the position is not fully diversiﬁed and you need scenario-dependent correlation matrices, would simply tell the traders that they cannot hold positions (1) and (2) since in some scenarios they will have large losses. Eﬃcient market types have a lot to learn about real risk control. Hedges are not essentially risk-free. Even a simple model would say that bets should not be made under conditions (1) and (2) because they are far too dangerous. Medium-sized hedge funds are likely reasonably diversiﬁed. Some type of risk control process is now standard but these systems are mostly based on the industry standard value at risk (VaR) and that is usually not enough protection in (3) as the penalty for large losses is not great enough. On occasion, even at a large fund, a rogue trader will have such a successful trading run that careful risk control is no longer applied. Instead, people focus on the returns generated, the utility function of the trader and that of the partners of the fund, rather than the longer-term utility function of the investors in the fund. Rogue trades — those that violate (1) and (2) — can be made as long as (3) never occurs. In the case of Amaranth’s natural gas bets, their leverage was about 8:1 so $7 was borrowed for every $1 the fund had from its clients. Positions were on exchanges and over the counter and were thus very vulnerable. Those not skilled in risk control can argue that situation (3) which is great enough to wipe them out, simply would not occur because it is far too improbable, that is too far in the tails of the distribution of the underlying asset. They would typically assign zero to the probability of such rare events. Even skilled risk control experts such as Jorion (2006) and Till (2006) refer to LTCM as an 8-sigma event and Amaranth as a 9-sigma event. The problem is that even modiﬁed VaR gives erroneous results and is not safe. Such wipeouts occur with events far more frequent than 8 or 9 sigma: 3-sigma is more like it. Till (2006) argues that daily volatility of Amaranth’s portfolio was 2%, making the September losses 9-sigma, but the possible losses are not stationary. We argue that this analysis is misleading; the 2% is with normal not negative low probability disaster scenarios. Furthermore, diversiﬁcation can easily fail, if, as is typical, it is based on simply averaging the past data rather than with scenario-dependent correlation matrices. It is the diversiﬁcation or lack thereof according to the given scenario that is page 711 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 712 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba (a) Crude oil spot: North Sea Brent; November 1, 2005 to November 28, 2006. (b) NYMEX natural gas futures close, November 1, 2005 to November 22, 2006. Figure 22.3. Energy prices November 2005 to November 2006. crucially important, not the average past correlation across the assets in the portfolio. Figures 22.3(a) and (b) illustrate the nature of the natural gas market. Ziemba and Ziemba (2013, Chapter 32) is a 2012 account of the gas market. Figure 22.3(a) shows crude oil prices from November 1, 2005 to November 28, page 712 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 713 2006. This shows much volatility with prices usually above $60 and at times exceeding the August 30, 2005 post-Katrina high of $70+. The oil prices peaked at $77 in July 2006, then declined to around $60 for much of the fall. This decline coincided with the decline in the price of natural gas in September 2006. At that time, widely watched weather-forecasting centers predicted that the hurricane season would not have major storms and that the winter would be mild. Previously on August 29, September natural gas suddenly rose sharply in the last half hour of trading. Why is not known, but manipulation might have been involved. For Brian Hunter, who was short September and long spring months, both events caused massive losses (see the discussion in section on “The trade and the rogue trader”). Figure 22.4 shows natural gas futures prices in 2006. Starting from over $11/million BTU, the futures prices fell to about $5. The event that triggered the Amaranth crisis was the drop in the price of natural gas from $8 in midJuly to around $5 in September. Since gas prices have climbed to $15 and fallen to $2 in recent years, such a drop is plausible in one’s scenario set and should have been considered. There are fat tails in these markets. There is a large diﬀerence between the daily and long-term moving average price of natural gas, making it a very volatile commodity. Thus such a drop is not a 8–9 sigma event. In the 1990s, natural gas traded for $2–3 per million BTUs. However, by the end of 2000, it reached $10 and then by September 2001 fell back to under $2. Figure 22.3(b) shows the NYMEX natural gas futures prices from November 1, 2005 to November 22, 2006 which like Figure 22.3(a) shows much price volatility. The November 22 price of $7.718 had recovered 50% from the September lows. Figure 22.4. Natural gas futures prices in 2006 to September. Source: Wall Street Journal. page 713 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 714 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba Figure 22.5. Amaranth timeline of a collapse. Source: New York Times, September 23, 2006. Figure 22.6. Daily change in P/L from Amaranth inferred natural gas positions, June 1 to September 15, 2006. Source: Till (2006). Figure 22.5 shows a chronology of the collapse and Figure 22.6 presents a day-by-day recreation of Amaranth’s possible losses including the disastrous last two months and ﬁnal collapse (a loss of $560 million on September 14, 2006) by Till (2006). Davis, Zuckerman and Sender (2007) discuss the bailout saga and some of the winners and losers. They describe how Amaranth scrambled to unload their positions that were losing more and more day by day: Sept 16 Agreed to pay Merrill Lynch approximately $250 million to take over some positions. Sept 17 Agreed to pay Goldman $185 million. page 714 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 How to Lose Money in Derivatives 715 Sept 18 Gave up on Goldman deal when clearing agent J.P. Morgan would not release collateral. Sept 20 Paid J.P. Morgan and Citadel $2.15 billion to take remaining trades after Amaranth absorbed a further $800 million in trading losses. The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. Valuing a Fund Actually the statement that Amaranth had $9.25 billion on September 1 is a bit of a stretch because that was the mark-to-the-market value of their portfolio, the value on which fees were charged. But, in fact, with an estimated 250,000+ natural gas contracts (about 30% of the market), an enormous position built up over the previous two years, the liquidating value of the portfolio was lower even before (3), the crisis. As a comparison, in his heyday in the 1990s, a large position for legendary hedge fund trader George Soros of the Quantum Fund was 5000 contracts. Even with one contract you can lose a lot of money: up to $20,000 in a few days. Indeed much of the previous proﬁts were derived by pushing up of long natural gas prices in an illiquid market. WTZ once had 7% of the ValueLine/S&P500 spread futures market. Even at that level, it is very diﬃcult to get out should the market turn on you. With those January eﬀect trades, one has a fairly well deﬁned exit point and the futures cannot deviate too much from the cash spread but even that level is too high and risky. So the real proﬁts were actually much lower. Those who liquidated Amaranth’s positions bought them at a substantial discount. J.P. Morgan Chase, Amaranth’s natural gas clearing broker made at least $725 million after taking over most of Amaranth’s positions (Davis et al., 2007). Of course, with diﬀerent data forecasts, such discrepancies might still occur occasionally but if they are consistently there, assumptions or risk assessments may be questioned. The trigger for the crisis was a substantial drop in natural prices largely because of high levels of stored gas, coupled with a perceived drop in demand due to changing weather, altering the seasonal pattern of trade. The trading theory was based on the dubious assumption that the natural gas market would underprice winter from summer natural gas prices. Background, adapted from Till (2006) The natural gas market has two main seasons: high demand in winter and generally lower demand in spring and fall. Storage facilitates provide some smoothing of the price. However, in the United States, there is inadequate page 715 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 716 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba storage capacity for the peak winter demand. Therefore, the winter natural gas contracts trade at ever-increasing premiums to summer and fall months to encourage both storage and creation of more production and storage capacity. Basically the market tries to lock in the value of storage by buying summer and fall natural gas and selling winter natural gas forward. The prices of summer and fall futures contracts typically trade at a discount to the winter contracts (contango) thus providing a return for storing natural gas. An owner of a storage facility can buy summer natural gas and simultaneously sell winter natural gas via the futures markets. This diﬀerence is the operator’s return for storage. When the summer futures contract matures, the storage operator can take delivery of the natural gas, and inject it into storage. Later when the winter futures contract matures, the operator can make delivery of the natural gas by drawing it out of storage. Figure 22.7 shows the average build-up of inventories over the year. As long as the operator’s ﬁnancing and physical outlay costs are under the spread locked in through the futures market, this will be proﬁtable. This is a simpliﬁed version of how storage operators can choose to monetize their physical assets. Sophisticated storage operators actually value their storage facilities as an option on calendarspreads. Storage is worth more if the calendar spreads in natural gas are volatile. As a calendar spread trades in steep contango, storage operators can buy the near-month contracts and sell the further-out month contracts, knowing that they can ultimately realize the value of this spread through storage. But a preferable scenario would be for the spread to then tighten, which means that they can trade out of the spread at a proﬁt. Later if the spread trades in wide contango again, they can reinitiate a purchase of the near-month versus far-month natural gas spread. As long as the spread Figure 22.7. Average U.S. natural gas inventories in BCF over the year, 1994–2005. Source: Till (2006). page 716 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 717 is volatile, the operator/trader can continually lock in proﬁts, and if they cannot trade out of the spread at a proﬁt, they can then take physical delivery and realize the value of their storage facility that way. Till (2006) believes that both storage operators and natural gas producers were the ultimate counterparties to Amaranth’s spread trading. In the winter, natural gas demand is inelastic. If cold weather comes early, there is fear that existing storage will not be suﬃcient, so prices are bid up. The fear of inadequate supplies lasts for the entire heating season. Winter 2005 was an example. At the end of the winter, storage could be completely depleted. For example, during February to March 2003, prices had moved up intraday $5.00/MMBtu, but settled only $2.50 higher, which is why Amaranth hoped for a long winter. As a weak hedge they short the summer (April to October). Demand for injection gas is spread throughout the summer and peak usage for electricity demand occurs in July/August. Being more elastic, this part of the curve does not rise as fast as the winter in a upward moving market. This was their hedge. The National Weather Service issued an el niño forecast for the 2006–2007 winter so gas storage was at an all-time record and the spreads were out very wide. This plus the fact that the market basically knew about Amaranth’s positions, led to their downfall, which was a result of their faulty risk control. The trade and the rogue trader Let us take a closer look at the trade that destabilized Amaranth. Brian Hunter, a 32-year-old Canadian from Calgary, had fairly simple trades but of enormous size. He had a series of successful returns. As a youth in Alberta, he could not aﬀord ski tickets but at 24, with training as an instant expert on derivatives from courses at the University of Alberta (including one from a colleague), he headed to a trading career. He was bold and innovative with nerves of steel while holding enormous positions. Typically he was net long with long positions in natural gas in the winter months (November to March) and short positions in the summer months (April to October). Amaranth Advisors was a multi-strategy fund, which is quite fashionable these days since they only have one layer of fees rather than the two layers in a fund of funds. On their website it states: “Amaranth’s investment professionals deploy capital in a broad spectrum of alternative investment and trading strategies in a highly disciplined, risk-controlled manner.” They provide a false sense of security from the assumed diversiﬁcation across strategies. The problem is that diversiﬁcation strategies can be correlated rather than hedged or independent, especially in extreme scenario cases. As a page 717 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 718 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba result, too much can be invested in any one strategy negating diversiﬁcation. In the case of Amaranth, some 58% of assets were tied up in Hunter’s gas trades but risk adjusted, these trades made up 70–90% of Amaranth’s capital allocation. Hunter made huge proﬁts for Amaranth by placing bullish bets on natural gas prices in 2005, the year Hurricane Katrina shocked natural gas reﬁning and production. Hoping to repeat the gains, Amaranth wagered with a 8:1 leverage that the diﬀerence between the March and April futures price of natural gas for 2007 and 2008 would widen. Instead it narrowed. The spread between April and March 2007 contracts went from $2.49 at the end of August 2006 to $0.58 by the end of September 2006. Historically, the spread in future prices for the March and April contracts have not been easily predictable. The spread is dependent on meteorological and political events whose uncertainty makes the placing of such large bets a precarious matter (Wikipedia, 2006). Jack Doueck of Stillwater Capital pointed out that while a good hedge fund investor has to pick good funds to invest in, the key to success in this business, is not to choose the best performing managers, but actually to avoid the frauds and blowups. Frauds can take on various forms including a misappropriation of funds, as in the case of Cambridge, run by John Natale out of Red Bank, NJ, or a mis-reporting of returns as in the case of Lipper, Beacon Hill, or the Manhattan Fund. Blowups usually occur when a single person at the hedge fund has the power to become desperate and bet the ranch with leverage. With both frauds and blowups, contrary to public opinion (and myth), size does not seem to matter: examples include Beacon Hill ($2 billion), Lipper ($5 billion), and Amaranth ($9 billion). Amaranth’s investors will be seeking answers to questions including: to what extent did leverage and concentration play a role in recent out-sized losses? We think the latter; (1) and (2) are the main causes here of the setup before the bad scenario caused the massive losses. Is learning possible? Do traders and researchers really learn from their trading errors? Some do but many do not. Or more precisely, do they care? What lessons are taken from the experience? Hunter previously worked for Deutsche Bank. In December 2003, his natural gas trading group was up $76 million for the year. Then it lost $51.2 million in a single week leading to Hunter’s departure from the Deutsche Bank. Then Hunter blamed “an unprecedented and unforeseeable run-up in gas prices.” At least he thought about extreme page 718 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 719 scenarios. Later in a lawsuit, he argued that while Deutsche Bank had losses, his group did not. Later in July 2006, after having billion dollar swings in his portfolio (January to April +$2B, −$1B in May when prices for autumn delivery fell, +$1B in June), he said that “the cycles that play out in the oil market can take several years, whereas in natural gas, cycles are several months.” The markets are unpredictable but, most successful traders would lower their bets in such markets. Our experience is that when you start losing, you are better oﬀ taking money oﬀ the table, not doubling up in the hope of recouping the losses. It is better to lose some resources and be able to survive than to risk being fully wiped out. However, instead they increased the bets. Amaranth was a favorite of hedge funds of funds, investment pools that buy into various portfolios to try to minimize risk. Funds of funds operated by well known and successful investment ﬁrms Morgan Stanley, Credit Suisse, Bank of New York, Deutsche Bank, and Man Investments all had stakes in Amaranth as of June 30, 2006. From September 2000 to November 30, 2005, the compound annual return to investors, net of all costs was a decent, but not impressive, 14.72%. This is net of their 1.5% management fee and 20% of the net new proﬁts. Amaranth had liquidated a signiﬁcant part of its positions in relatively easy-to-sell securities like convertible bonds, leveraged loans and blank check companies or special-purpose acquisition companies. Liquid investments were sold at a small discount while others, like portfolios of mortgage-backed securities, commanded a steeper discount. As is common among hedge funds, Amaranth severely restricts the ability of investors to cash in their holdings. For example, investors can withdraw money only on the anniversary of their investments and then, only with 90 day’s notice. If they try to withdraw at any point outside that time frame, there is a 2.5% penalty. If investors redeem more than 7.5% of the fund’s assets, Amaranth can refuse further withdrawals. Our experience is that if you lose 50% of a $2 million fund, you will have a hard time relocating to a new fund or raising new money, but if you lose 50% of $2 billion, the job fund prospects are much better. So Hunter moved on to Amaranth whose founder and chief executive, Nick Maounis, said on August 11, 2006, that more than a dozen members of his risk management team served as a check on his star gas trader “what Brian is really, really good at is taking controlled and measured risk.” Nick will forever eat these words. Amaranth said they had careful risk control but they did not really use it. Some 50% of assets in one volatile market is not really very diversiﬁed at any page 719 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 720 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba time and is especially vulnerable in a crash and doubly so if one’s bets make up a large percent of the market. Such a large position is especially dangerous when the other traders in the market know a fund is overextended in this way and many hedge funds such as Citadel and J.P. Morgan were on the other side of the market. Then, when the crisis occurred, spreads widened, adding to the losses. Hunter’s response was to bet more and more (in eﬀect doubling up) until these trades lost so much they had to be liquidated. That is exactly what one should not do based on risk control considerations, but, as discussed below, it makes some sense with traders’ utility functions. Successful traders make a large number of hopefully independent favorable bets which, although they may involve a lot of capital, are not a large percent of the capital nor are they in an illiquid market should one need to liquidate. Warren Buﬀett’s Berkshire Hathaway closed end hedge fund frequently makes $1 billion risky bets but these have a substantial edge (positive expected value) and about 1% or less of Berkshire Hathaway’s more than 140 billion capital. The insurance business brings in a constant ﬂow of billions of dollars in premiums. So Berkshire always has a lot of cash to invest. With Buﬀett keeping billions in cash equivalent reserves for security and good opportunities. A typical Buﬀett trade was a loan of some $945 million to the Williams pipeline company of Oklahoma at some 34% interest in 2002 during the stock market crash, when the oil price was low and the pipeline company was in deep ﬁnancial trouble. Banks refused to bail them out. But Buﬀett knew he had good collateral with the land, pipeline, and buildings. Williams recovered largely due to this investment and better markets and paid oﬀ the loan early. Berkshire Hathaway made a large proﬁt. In the 2007–2009 stock market crisis and decline, Buﬀett made $5–10 billion loans to GE and Goldman Sachs which both were in deep ﬁnancial trouble. In return he got preferred shares paying a 10% dividend plus free warrant options on the stock of GE and GS. Later, when those were cashed in, Berkshire made billions in proﬁts. The problem is that rogue traders are grown in particular organizations and are allowed by the industry. While they are winning, they are called great traders, then they become rogue traders when they blow up their funds. The Hunter case is similar to those of Nick Leeson and Victor Niederhoﬀer but diﬀerent than Long Term Capital Management (LTCM). In the ﬁrst three cases, there was a major emphasis on trade in one basic commodity. The trouble was the risk control, namely our (1) and (2) and combined with the bad scenario (3). As discussed in the next sections the ﬁrm’s and rogue trader’s utility function likely caused this problem by making it optimal page 720 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 How to Lose Money in Derivatives 721 for these utility functions to over-bet. LTCM is much more subtle. The conﬁdence scenario that hit them was the result of faulty risk control based on VaR and historical data. They needed scenario-dependent correlation matrices. The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. Possible Utility Functions of Hedge Fund Traders One way to rank investors is by the symmetric downside Sharpe ratio (DSSR) (see Gergaud and Ziemba, 2012). By that measure, investors with few and small losses and good sized gains have large DSSRs. Berkshire Hathaway has a DSSR of about 0.917 for the period 1985–2000. The DSSR of both the Harvard and Ford Foundations endowments were about 1.0. Thorp’s Princeton Newport’s 1969–1988 DSSR is 13.8. Renaissance Medallion, possibly the world’s most successful hedge fund, had a DSSR of 26.4 during the period January 1993 to April 2005. See also the other funds in the UMASS hedge fund data studied in Gergaud and Ziemba (2012). The results come from the choices made using a utility function. Those who want high DSSRs are investors trying to have smooth and good returns with low volatility and very few monthly losses. Thorp only had three monthly losses in 20 years; the Harvard and Ford endowments and Berkshire Hathaway had 2–3–4 per year. Consider a rogue trader’s utility function.4 The outcome probabilities are: 1. x% of the time the fund blows up and loses 40%+ of its value at some time; the trader is ﬁred and gets another trading job keeping most past bonuses. 2. y% of the time the fund has modest returns of 15% or less; then the trader receives a salary but little or no bonus. 3. z% of the time the fund has large returns of 25% to 100%; then the trader gathers more assets to trade and receives large bonuses. At all times the rogue trader is in (1) and (2), that is, the total positions are over-bet and not diversiﬁed and move markets. There is no plan to exit the strategy since it is assumed that trades can continually be made. Then in a multiperiod or continuous time model, it may well be for the fund managers’ and traders’ speciﬁc utility functions, that it is optimal to take bets that provide enormous gains in some scenarios and huge 4 An academic treatment of a rogue trader is in Lleo and Ziemba (2014). Here we sketch some ideas. page 721 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 722 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba losses in other scenarios. Kouwenberg and Ziemba (2007) show that in a theoretical continuous time model with incentives, risk taking behavior is greatly moderated if the hedge fund manager’s stake in the fund is 30% or more. In the case of Amaranth and similar rogue trading situations, there are additional complications such as the fund manager’s utility function and his wealth stake inside this fund and outside it. Then there is the rogue trader’s utility function and his wealth inside and outside the fund. According to Aumann (2005) in his Nobel lecture: a person’s behavior is rational if it is in his best interests given his information. Aumann further endorses the late Yale Nobel James Tobin’s belief that economics is all about incentives. In the case of Hunter, his share of $1B plus gains (real or booked) was in the $100 million range. What is interesting, and this is similar to LTCM, is that these traders continue and increase bets when so much is already in the bank. Recall in LTCM, that they had a $100 million unsecured loan to invest in their fund. Finally, in such analyses, one must consider the utility functions and constraints of the other investors’ money. In the case of Amaranth, Deutsche Bank who had ﬁrst-hand knowledge of Hunter’s previous trading blowups, was an investor along with other well known ﬁrms. This behaviour is symptomatic of doomed hedge fund managers. Sender and Singer (2003) recount the fall of John Koonmen, who founded Eifuku Master Trust hedge fund in Tokyo in late 2000. The fund reached a peak value of $300 million in 2001 before losing virtually all of the asset under management in just over a week. A high tolerance for risk, if not an outright risk-seeking behaviour, leading to large leverage and overconcentration are to blame for the fund’s demise: as the fund’s capital deeper to $155 millions, Koonmen was still able to invest assets worth $1.4 billion, but decided to allocate them to just a few positions. There were warnings signs, though: Koonmen lost his job at Lehman Brothers after his 1998 trading loss were so large that they had an impact on the bonus of the whole Tokyo equity division. Another sign was the increasing volatility of Koonmen’s portfolio in the Amber Arbitrage Fund, Eifuku’s predecessor. Winners and Losers Who are the winners and losers here? Hunter is a winner and will get relocated soon. He has hundreds of millions, having made about $75 million in 2005 (out of his team’s $1.26 billion proﬁt), and will likely make more later. Of course, his reputation is tarnished but $100+ million in fees over page 722 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 723 the years helps. Like many others, Hunter had to leave 30% of this in the fund so some of the $75 million was lost. There might be some lawsuits but he likely will not be hurt much. At 32, he is set for life ﬁnancially, despite the losses. He is likely to begin again. An executive recruiter has oﬀered to help introduce Hunter to investors. He sees opportunities for Hunter to make a fresh start with high-net-worth investors, possibly in Russia and the Middle East.5 Betting on fallen hedge fund stars is not all that uncommon. John Meriwether, who led Long-Term Capital Management until its 1998 implosion, now runs another hedge fund. Nicholas Maounis, Amaranth’s founder and CEO, was exploring starting a new hedge fund. Instead of being ahead 27% for 2005, his fund had to be liquidated. He lost much of his previous fees by leaving much of it in the fund. Since 2005, there were $70 million in management fees and $200 million in incentive fees, his cut was substantial but like LTCM, he should have diversiﬁed his wealth. Other winners were those on the other side of the trade if they followed proper risk control and could weather the storm created by Amaranth’s plays, and those like Citadel Investment Group, Merrill Lynch and J.P. Morgan Chase, who took over Amaranth’s portfolio and the Fortress Investment group, which helped liquidate assets. J.P. Morgan was named “Energy Derivatives House of the Year, 2006” by Risk magazine. The losers were mainly the investors in Amaranth including various pension funds which sought higher returns to make up for 2000–2003 equity investment mistakes. As of January 30, 2007, they had received about $1.6 billion which is less than 20% of their investment value in August 2006. They will receive a bit more but their losses will exceed 75%. Those who invested in mid-2005 received about 27% of their original investment or about 18% of the peak August value. Other losers are hedge funds like Mother Rock LP which were swept up by the Amaranth debacle including those that lost even though they bet on the right (short) direction because Hunter moved the market long on the way up and those who lost along with Amaranth on the way down. They were long October and short 5 Indeed in late March 2007 it was widely reported that Hunter was soliciting money for a series of commodity funds with the name Solengo Capital. It is believed that cash-rich investors in the Middle East and Europe are likely to invest. To assuage fears of another meltdown, investors will be able to pick speciﬁc managers and commodities. The new fund will impose margin and other restrictions on managers and will eliminate all lock-in restrictions if these controls are violated. The prices of the natural gas contracts Mr Hunter is known to favor had been increasing in anticipation of his return to the market. page 723 July 3, 2015 8:28 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. 724 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba September futures. According to Till (2006), they likely were forced out of their short position August 2, 2006 when the spread rallied brieﬂy but sharply. Another loser was Man Alternative Investments Ltd., a fund of hedge funds listed on the London Stock Exchange in 2001 by the Man Group PLC, which shut down after recent losses tied to Amaranth’s collapse and persistently poor liquidity in the shares. It is a small fund with little active trading interest, a concentrated shareholder base, and positions that were both diﬃcult to build up and unwind. It had about £31.5 million invested in a portfolio selected by Man Group’s Chicago-based Glenwood Capital Investments LLC unit, which is part of Man Group PLC, and has $58 billion in assets under management. The fund lost about one-ﬁfth of its gains during the year from the collapse of Amaranth though it was up 6.5% through October. Archeus Capital, a hedge fund that in October 2005 had assets of $3 billion, announced on October 31, 2006, that it would close, returning $700 million to their investors. The fund, founded and run by two former Salomon Brothers bond traders, Gary K. Kilberg and Peter G. Hirsch, was like Amaranth, a multistrategy fund. However, it had a more conservative approach that focused on exploiting arbitrage opportunities in convertible bonds. Archeus began experiencing redemptions last year after its main investment strategy fell out of favor. The fund’s founders blamed its administrator for failing to maintain accurate records. Their subsequent inability to properly reconcile the fund’s records, led to a series of investor withdrawals from which they were not able to recover. Also, Archeus’s 2006 performance did little to inspire its clients. Through the ﬁrst week of October 2006, Archeus’s main fund was down 1.9% for the year. However, the fund had returned 18.5% since July 2005. Still, during a period when hedge fund returns have come under increased scrutiny and have, on average, lagged the returns of the major stock market indexes, such a return was insuﬃcient to keep investors on board. The $7.7 billion San Diego County Employees’ Retirement Association has retained the class-action ﬁrm Bernstein Litowitz Berger and Grossmann to investigate the Amaranth implosion. Its $175 million investment in Amaranth, which was valued at $234 million in June 2006, is now estimated to be worth only $70 million, thus a $150+ million loss. They should have done better due diligence in advance. Those who bet the ranch on every trade eventually lose it. Investors should have known that was what they were investing in with Amaranth. page 724 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives 725 Following Amaranth’s collapse, investors were seeking someone to blame. Some argued that these bets showed the need for greater or a diﬀerent sort of regulation of hedge funds, or at least of the sort of over-the-counter trades in the natural gas market. Others including Gretchen Morgenson of the New York Times, pointed to the persistence of what many have called the Enron loophole, created in 1993, when the Commodity Futures Trading Commission (CFTC) exempted bilateral energy futures transactions from its regulatory authority. This exemption was extended in 2000 in the commodity futures modernization act to include electronic facilities. Many have argued that Enron used such trades to increase the value of long-term contracts. In the run-up of gas prices in 2005/2006, some analysts and politicians pointed to the role of speculators in changing the demand structure, leading a congressional subcommittee to release a report urging that such trades all be the concern of U.S. regulators. Amaranth’s collapse brings a diﬀerent aspect to this debate, as it shows the limits to such self-regulation by market actors. While it is unclear what policy actions might be taken in this matter, this concern is likely to continue and may change the environment in which such trades are made in the future. However, there are limits to the role that can be played by such regulation. Other small losers are funds of Morgan Stanley and Goldman Sachs who lost 2.5% to 5% from their Amaranth holdings. However as they helped unwind the trades, they may well have recouped their losses when the energy market prices subsequently increased. There is little impact from this on the world economy. The hedge fund industry now has a bit more pressure to regulate position sizes but most regulators steer away from risk control. When you mention risk control, you are usually encouraged to change the subject. What regulators are interested in is operational risk. The exchanges have limits but rogue traders are able to get around these rules. In any event, if VaR were to be used it would most likely not work unless one is blessed with no bad scenarios. As long as risk control is so poorly understood, misapplied, and disregarded, and pension funds and others are desperate for high returns, such disasters will occur from time to time; and this is fully expected. It is simply part of the hedge fund zero sum gain. For every Jim Simons or Blair Hull eaking out steady proﬁts using a lot of careful research, excellent execution, position sizing, and strict risk control, there is a rogue trader trying to make it by over-betting with very little research and a ﬁrm which improperly applies risk control. Improper regulation may well hurt more than help. page 725 July 3, 2015 8:28 726 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba Mettallgeselschaft Reﬁning and Marketing Inc (1993) The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. The story of the Mettallgeselschaft Reﬁning and Marketing (MGRM) disaster is still highly relevant today because it is a complex and passionately debated case. Even 20 years later, several questions remain open: • Was MGRM’s strategy legitimate hedging or speculation? • Could and should the parent company, Mettallgeselschaft AG, have withstood the liquidity pressure? • Was the decision to unwind the strategy in early 1994 the right one? The following discussion expands on description in Lleo (2010). In December 1991, MGRM, the U.S.-based oil marketing subsidiary of German industrial group Mettallgeselschaft AG, sold forward contracts guaranteeing its customers certain prices for 5 or 10 years. By 1993, the total amount of contracts outstanding was equivalent to 150 million barrels of oil-related products. If oil prices increased, this strategy would have left MGRM vulnerable. To hedge this risk, MGRM entered into a series of long positions mostly in the very liquid short-term futures (some for just one month). This practice, known as “stack hedging,” involves periodically rolling over the contracts as they near maturity to maintain the hedge (see Figure 22.8). Stack hedging helps address the maturity gap between the long-term exposure and the short-term hedging instrument. In theory, maintaining the hedged positions through the life of the longterm forward contracts eliminates all risk. But intermediate cash ﬂows may not match because of the daily settlement of futures. This could result in liquidity risk. As long as oil prices kept rising or remained stable, MGRM would be able to roll over its short-term futures without incurring signiﬁcant cash ﬂow problems. In eﬀect, the rollover of the futures would create positive cash ﬂows that MGRM would be able to invest until the maturate of the forward. However, if oil prices declined, MGRM would have to make large cash infusions in its hedging strategy to ﬁnance margin calls and roll over its futures. In reality, oil prices fell through 1993, resulting in a total loss of $1.3 billion on the short-term futures by the end of the year. All of the losses related to the cash inﬂow required by the mechanics of the stack hedging. Mettallgeselschaft AG’s supervisory board took decisive actions: they replaced MGRM’s senior management and unwound the strategy at an page 726 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. How to Lose Money in Derivatives Figure 22.8. 727 Stack hedging. enormous cost. In the end, Mettallgeselschaft AG was only saved by a $1.9 billion rescue package organized early 1994 by 150 German and international banks. Mettallgeselschaft Reﬁning and Marketing (MGRM) is one of the best studied ﬁnancial disasters, mostly because it was the stage of a passionate debate. Mello and Parsons (1995) analysed the disaster shortly after Mettallgeselschaft’s rescue. They generally supported the initial reports in the press that equated the Mettallgeselschaft strategy with speculation and mentioned funding risk as the leading cause of the company’s meltdown. The same year, Culp and Miller (1995a,b) took a very diﬀerent view, asserting that the real culprit in the debacle was not the funding risk inherent in the strategy, but the lack of understanding of Mettallgeselschaft AG’s supervisory board. Culp and Miller further pointed out that the losses incurred were only paper losses that could be compensated for in the long term. By choosing to liquidate the strategy, the supervisory board crystallized the paper losses into actual losses and nearly bankrupted their industrial group. page 727 July 3, 2015 8:28 728 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 S. Lleo and W. T. Ziemba The World Scientific Handbook of Futures Markets Downloaded from www.worldscientific.com by LA TROBE UNIVERSITY on 05/22/16. For personal use only. Edwards and Canter (1995) broadly agreed with Culp and Miller’s analysis: the near collapse of Mettallgeselschaft was the result of disagreement between the supervisory board and MGRM senior management on the soundness and appropriateness of the strategy. The key diﬀerence between Culp and Miller (1995a,b) and Edwards and Canter (1995) is Culp and Miller’s assertion that MGRM’s strategy was self-ﬁnancing, which Edwards and Canter reject. Société Générale (2008) A major event in January 2008 was the rogue trader losses at Société Générale. One thing to observe is that in times of uncertainty, there are more rogue traders. Besides this loss, some $1.4 billion was lost on wheat in two days by a rogue trader at MF Global causing them to lose one fourth of their worth. On January 21 (a U.S. holiday) and 22, 2008 (Monday and Tuesday) nights, the S&P500 futures was some 60 points lower on Globex trading (1265 area) well below previous lows (1406 on August 16, 2007, 1364 on October 17, 2006, and 1273 on March 10, 2008). On both days, the day market recovered, but much damage was done. Jérome Kerviel and SG lost 4.9 billion euro trading index futures in the DAX, FTSE and CAC. By correlation, the S&P500 fell to new lows. Many were hurt. How could a junior trader hold $50 billion euro in positions? The sidebar exhibit is Société Générale’s explanation of the incident. What Is a Subprime Loan and Why Have They Caused so much Trouble in so Many Places? Subprime loans are loans to borrowers who do not qualify for best interest or with terms that make the borrower eventually unqualiﬁed as with zero down payment, zero interest. In general, lending institutions inherently get it wrong. When times are good, they tend to be greedy and try to maximize loan proﬁts but then they are very lax in their evaluation of borrowers’ ability to pay current and future mortgage payments. • Japan in the late 1980s: real estate and stocks, eventually the 10 trillion was lost. page 728 July 3, 2015 8:28 World Scientific Handbook of Futures Markets. . . 9.75in x 6.5in b1892-ch22 729 The World Scien