Tuesday, August 14, 2007

Veryan: Quant hedge fund chaos

Quant fund bloodbath? The steamrollers have been flattening some nickel collectors recently. Damaged collateral is causing lots of collateral damage in the markets. ALL strategies are directional including the "non-directional" ones. Market neutral is not risk neutral especially when there is no mean to revert to. Some "hedges" became "texas hedges" as shorts went up and longs went down. However this overdue shake out will ultimately be a positive for good systematic fund managers.

The best risk management rule I know is "If it can happen, it will happen". The trouble with standard deviation is it just measures STANDARD deviations. The market is never "normal"; it just oscillates from one extreme to the other. Chaos and complexity aren't black swans, they are permanent features of the markets. Volatility is analogous to energy in that it can hide as potential energy but it is always around. Curious how these "once in a 100,000 years" storms occur every few years and blow away anyone who hasn't battened down the hatches. Usually after just a long enough gap in time for neophytes to say volatility is "permanently" contained!

Pundits seem surprised that what started in illiquid credit could affect funds as diverse as liquid equity funds or oil traders. Contagion and hysteria will often affect leveraged strategies. The irrational swamps the rational yet again. Economic expectations and logical assumptions are useless in modeling such an inherently irrational and emotional process. "Rare" events are not very rare, tend to cluster together and lead on to other "rare" events. Fundamentals are irrelevant when fear grips the markets and that will in turn negatively effect those fundamentals.

High frequency trading is actually safer than low frequency trading and it was the "slower" systems that performed worst recently. Some smaller, more agile quant funds using different models and trading in shorter time frames have been arbing the bigger arbs. Just like credit funds, there were losers AND winners in quant recently. Longer term statistical arbitrage is not a black box strategy anymore. More a grey box. It used to be way off the radar screen for most investors and involve relatively small sums of money. But its success has led to significant trade crowding and transparency of what needs to be kept a black box. All arb strategies must eventually get arbed out so you have to keep finding new ones. With every fund AND every strategy there is a point beyond which the dangers of crowding exceed the rewards.

The recent stat arb problems were almost inevitable. 1) Spreads have been getting very tight requiring more leverage to maintain returns but the banks want out necessitating liquidation of mean reversion positions 2) Due to losses in OTHER areas investors are redeeming and quant funds are very liquid 3) Clients now demand more transparency and over time trade secrets have leaked into the marketplace allowing less competent funds to try to emulate the better ones. Loose lips sink ships. 4) Fund employees left to start their own firms but with the SAME strategy DNA eg Goldman Sachs Asset Management to AQR or DE Shaw to Tykhe 5) In some areas of stat arb, capacity was surpassed a while ago but many funds continued to take in assets. It is not just the AUM in a fund, the total AUM in an industry-wide strategy also matters.

"Quant fund" is almost as vague a term as "hedge fund". Some quant funds have done well in this environment. There is not only stat arb within the "quant" space. There is trend following, countertrend trading and volatility arb among others. I am not a quant but I certainly employ plenty of obscure math and proprietary statistical measures to evaluate managers, models and strategies. If you can't quantify your edge you don't have one. If you can't measure your risk you can't manage your risk. You can test and evaluate quantitative investment methods rigorously but discretionary funds rarely have a long enough track record. A bad month or quarter for some prominent, possibly oversized funds does not in any way impact the strong diversification case for quant driven strategies.

The idea that "this has never happened before" is wrong. Volatility is not new. Equity dispersion is not new. Correlation transitioning all over the place is not new. Some newbie quants use just 5 years of historical data so it is interesting that the storm hit exactly as the most volatile month this century dropped off their spreadsheets. DURING July 2002 the Dow fell 18% then rallied 12%. We've seen nothing like that, so far. Stat arbs have shorts and longs so an unwinding means popular shorts go up while popular longs go down. Convergence trades only work if there are reasons they should converge. Historical relationships are just that - HISTORICAL. Beta and correlation just describe the PAST.

If there was one term that needs to be removed from the manager lexicon it would be "market neutral". Marketers and other capital raisers love it. But "market neutral" strategies just transform one set of risks into another set of risks. If you short 250 stocks in the the S&P 500 and buy the other 250, is that market neutral? No. Say you put on a Ford versus General Motors beta neutral pair trade. Is that market neutral? No. All you've done is transform single security directional risk into double security directional spread risk and spreads can be even harder to forecast than the outright. You've introduced correlation issues that are even more difficult to model. You've also got 4 trades to do instead of 2. No fund or strategy is market neutral.

There is skill dispersion in stat arb as in every investment strategy. Mean reversion assumes there actually is a stable mean to revert to. Some models are based on ideas of economic equilibrium and no-arbitrage efficient markets, ie that prices "must" EVENTUALLY come back to someone's guess of fair value. Past market data is widely available and if you throw enough variables and optimizable parameters into a data mining model it is certain you will come up with patterns and factors that predict the past. What separates the good from the bad are those with the skill to find predictors for the future.

Every fund has tough periods and nothing has altered the fact that Renaissance Technologies remains the best quant firm currently operating. It is worth noting that their core fund Medallion is positive for this year and is where the managers keep their own money. Medallion had a rough 1989 and those investors that redeemed would have missed many years of subsequent high capital growth. The fund in the news, the 2 year old Renaissance Institutional Equities Fund, has a longer term trading outlook and reveals more information. More disclosure attracts copycats which can create problems for everyone. Other funds have been trying to reverse engineer Jim Simons and his colleagues' quant strategies since the triple digit gains in the difficult year for many other funds of 1994. RIEF, being newer, larger, less opaque and nimble, was more likely to be impacted by competitors.

Contagion affects all strategies. Bad funds can hurt good funds. AQR and DE Shaw had terrible 1998s mostly due to LTCM blowing up but did fine until last month. EVERYONE loses money sometimes. Say you are a good small cap stock picker. Someone else is not good but also has a position in those stocks. That fund gets into trouble and investors and leverage providers want their money back. The bad fund is forced to dump the stocks you also own. Even though you yourself are good, you lose money. That is exactly what is going on in stat arb now. Fortunately while the short term means problems for everyone in that particular strategy, in the long term the outcome is positive as the worst funds close down while the good funds will thrive.

Interesting how many insist on investing in things they "understand" and then proceed to pile into things they clearly never looked into other than the yield spread and "rating". The market decides what deserves to be considered AAA not some ratings agency. I wish CEOs and CFOs would ask their CIOs and CROs about CDOs and CLOs before denying they have any exposure. But almost by their own admission banks don't know what they don't know so they want there money back NOW from anything liquid. Contrary to popular belief the credit markets have not ground to a halt; cash credit may have temporarily but credit derivatives have been seen massive trading recently. Could the next shoe to drop be in CDSs and CLNs?

The 2 and 20 crowd are NOT to blame for quite rightly adjusting their portfolios. The 2 and 28 crowd, better known as mortgage brokers, were the catalyst. It is noteworthy how regulators won't let hedge funds sell to the mass affluent yet permit sub-prime lenders to aggressively target first-time homebuyers who were clearly not made aware of the risks with option ARMs and low 2 yr/high 28 yr loans and thereby subject them to negative equity, delinquency and foreclosure. As usual it is all embedded options; the option to raise interest rates, the option to not pay those rates, the option to foreclose, the option to try to stay put and call a lawyer. Given the on-selling and repackaging of mortgages and structured investment vehicle conduits nowadays the ultimate owner is not always clear so the sub-prime mess is going to impact the markets for a while.

What started as a small part of the US credit markets has impacted many other areas. Contagion is contagious and bear markets tend to RAISE correlations across risky assets. To anticipate risks you need to be aware of what is coming out of left field. Part of the problem is the silo mentality of a lot of the street; while cross-product expertise has grown the basic stance remains of "I am equity, you are fixed-income, he is currencies and she is commodities" when in fact these days you probably need competence across asset classes. And of course there is the sharp divide between fundamentalists and quants when you really need to know both. There are other strategies and assets not YET impacted by the sub-prime meltdown.

Unlike the experts I haven't a clue if the Dow will be below 10,000 or above 15,000 by the time this is all played out. I do know that the reverberations and panic are creating opportunities each day. At least good fund managers are taking action and reducing risk during this storm. I find this stay in for the long haul, ride out the volatility "advice" ludicrous. Ships do their best to get to the nearest port and aeroplanes avoid hurricanes but sell-side strategists are mostly recommending staying invested and they claim stocks are "cheap"!

If you can't or won't invest in good hedge funds or go short, the safe haven is CASH. It is not as though "common sense" traditional stock and bond funds are immune from all this. There has been little selling by long only and long biased funds so far but if it comes the effects will be worse. The quant funds were selling AND buying last week and the market ended UP. Dig your well before you are thirsty and DEFINITELY before everyone else gets thirsty for the safety of cash.
by Veryan Allen. Copywrite 2007.

No comments:

Lunch is for wimps

Lunch is for wimps
It's not a question of enough, pal. It's a zero sum game, somebody wins, somebody loses. Money itself isn't lost or made, it's simply transferred from one perception to another.