I had a very interesting discussion yesterday with the head of risk management at a major international investment bank. Among the various things we talked about was the foolhardiness of how the new banking regulations are being structured. In one particularly noteworthy example, he was explaining to me the Fed’s plan for regulating ‘Sharpe ratios’ for all registered money managers and bank prop trading units.
For those of you who don’t know, a Sharpe ratio is supposed to describe a portfolio’s market risk in a standardized way. It is essentially the returns of the portfolio over time, minus a risk free rate, divided by the standard deviation of the return series. What you get is a metric which is supposed to describe the ‘unit risk’ of a portfolio, and over time, of the strategy by which it was generated.
That’s all very wonky so let me abbreviate it. If your P&L goes up every single day and doesn’t wiggle around very much, then you’ll have a low standard deviation, and a high Sharpe ratio. If it goes up two days and then down one day in a repeating pattern, the standard deviation will be higher, and the Sharpe ratio lower. If it goes down 2 days of every three, then your return will be low, and your Sharpe even lower.
The conventional wisdom has been that if you have a Sharpe ratio above one, then you are providing a higher return for each unit of risk and are ‘outperforming’ the market on a risk adjusted basis. But if you have a Sharpe ratio below one, then you are underperforming. And the Fed in its infinite wisdom, is planning to limit risk by examining Sharpe ratios. Their thinking is people with high Sharpe ratios are less risky, and therefore justify more leverage.
But there is a serious problem with this. You can artificially increase your Sharpe ratio by selling ‘implied optionally’. In fact, for strategies with a holding period longer than 1 day (it’s a little different for High Frequency strategies) virtually every strategy with a Sharpe ratio above 2.5 or so will almost certainly be using implied optionally. And what’s even worse – since you don’t actually have to trade options to embed implied optionally, it’s very likely that the managers involved don’t even realize it.
The technical term is ‘selling volatility’. When you sell volatility, you’re getting paid a tiny bit every day in exchange for what may amount to a much steeper than average loss on very rare occasions. The classic analogy for this is ‘picking up pennies in front of the steam roller’. Strategies which have very high Sharpe ratios will almost always be engaging in that practice, and it always ends the same way. Things will be going fine - making a lot of money, and then one day the loss will be 150% (or more… maybe MUCH more) of all the money that’s been made to date, in a single catastrophic loss.
So if the Fed is planning to grant more risk and more leverage to those strategies with high Sharpe ratios, they are doing the exact opposite of what they should be. They are using the new ‘rush to regulate’ and a lack of understanding of how the statistics of finance really work to make the next catastrophic market collapse absolutely certain.
It’s as if they’re building a house, and rather than using sand to lay the foundation they’re using something more solid and less likely to shift around…. they’re using TNT. This is the pure pretense of knowledge. Regulators believe that the Sharpe ratio tells them something about risk that it doesn’t actually tell them, and in response they do exactly the wrong thing.
What’s worse, there is only so much information in the market, and all professional investors look at that same information. So you can bet that whether they realize it or not, they’re almost certainly all investing in the same sorts of things. That greatly increases the odds that a catastrophic loss for one will be very likely be a catastrophic loss for all.
The Stat Arb collapse in 2007 was an excellent example of that. Statistical Arbitrage had demonstrated huge returns over the prior decade greatly outperforming the market. The standard deviation of their returns was tiny and their Sharpe ratios very high. Then on three days in September 2007, virtually every Stat Arb unit in the world began taking massive losses – some giving back an entire decade’s worth of gains in a single day.
The problem was that there was an implied correlation between these strategies. Every firm with a Stat Arb unit hired the same geniuses, trained in the same schools and in the same disciplines. They all performed similar analysis, or at least an analysis which led them to the same conclusions. And when it fell apart for one, it fell apart for all of them.
If the Fed is going to regulate Sharpe ratios in the manner that was described to me, then this is what they should expect going forward. Everything will seem fine until the next crisis occurs, and then the dominoes they’ve set up with this new regulation will bring everything down around us. Regulation like this won’t prevent the next crisis, it will be absolutely ensuring it.
0 comments:
Post a Comment