Thursday, February 12, 2009

The limitation of Sharpe ratio

Just as one should not trust VaR completely, one should also beware of high Sharpe ratio strategies. As this Economist article pointed out, a strategy may have a high Sharpe ratio because it has so far been accumulating small gains quite consistently, but it could still be subject to a large loss when black-swan events strike.

Personally, I am more comfortable with strategies that do the opposite: those that seldom generate any returns, but always earn a large profit when financial catastrophes occur.

21 comments:

GS751 said...

strategies like that would have been profitable this past year. I think some of it is still priced with an elevated VIX level.

Anonymous said...

I completely agree that pure kelly's strategy or sharpe ratio maximization will not necessarily generate desirable trading strategies.

I recently openned a blog at: http://www.weizhenstanford.com/blog to discuss issues on quantitative modeling, trading, data mining etc. It covers some issues related to kelly's strategy, market price of risk (sharpe ratio) as well.

Anonymous said...

Mr. Chan, I read your latest book and was very pleased with the quality and the amount of information it contained. In one of the first chapters you mentioned that readers will gain password to the subscription part of your web site in later chapters. Can you please specify where it is, as I was unable to locate it. Thank you.

Best regards,

Michal Kecera

Anonymous said...

Dear Ernest,

I am enjoying your book a lot and promoting it to some friends and colleagues in London.
Concenrning your preference about winning a small bit everyday (or even losing) and get the big bonus when improvable events happens, I have noticed that it is the preference of other big guys Taleb. Man, this is the difficult to bear, the constant pain waiting for the big win. In your portfolio of strategies, what % of them bet on improbable events?
Bye
Ruso

Ernie Chan said...

Michal,
Thanks for your kind words. The password is in last paragraph of page 34.
Best,
Ernie

Ernie Chan said...

Dear Anonymous,
I would say I am currently 100% in this type of strategies personally!

Note that Warren Buffer and other deep-value investors can be regarded as investing with this strategy in mind also.

Ernie

Anonymous said...

Venture capital is also a common form of the lots of small loser with one or two "home runs" making up for the other, smaller losses.

Although, it should be noted, Taleb's hedge fund did much of nothing when he attempted to employ it using options, where he bled theta constantly while buying mostly strandles on biotech stocks...

Anonymous said...

Hi Ernest
From your book and blog I got the impression you are mainly a pairs trader.
To the other Anonymous you have said that you are 100% in these big payoff type trades.
May I ask what sort of percentages you allocate to the different strategies you run - pairs, large payoff, contrarian etc

Ernie Chan said...

Anonymous,
Yes, I traded pairs and various other strategies in my personal account before, and still trade them for others. However, at this moment, my personal account consists of only 1 strategy that seldom fire unless a big market move occurs.
Ernie

Anonymous said...

Mr. Chan, I read through subscription part of your blog. What program do you use to calculate the number of shares which one should trade? (I am referring to your paper on XLE arbitrage against its component stocks) Thank you.

Best regards,
Michal Kecera

Anonymous said...

Hi Ernie,

I'm very confused by your statement:

"Yes, I traded pairs and various other strategies in my personal account before, and still trade them for others. However, at this moment, my personal account consists of only 1 strategy that seldom fire unless a big market move occurs."

From all I've read on your blog and in your book (I own a copy) you are taking the arb route - small safe returns. Now this. This is a very, very radical departure.

Can you shed some light on your switch to your reading audience:

> Why the switch?
> What is new approach?/or at least some meaty examples of that class of appoach.
> Why we should/shouldn't be less interested in or concerned about the start arb models this forum presents given that the cook is no longer eating from his recipe book?

Thanks,
Carl

Ernie Chan said...

Hi Carl,
There wasn't a switch in my strategies. I still trade some version of all the strategies that backtested well and described in my book. However, I trade different strategies for my personal acct vs my fund acct due to their different needs. I no longer have the time to pay too much attention to my personal acct, so I only like to trade infrequently, and when there is deep value. For the fund acct, we are still working hard everyday trying to eke out small gains and prevent big losses.

By the way, stat arb does not preclude a deep value strategy. It just means entering at, e.g., 3 standard deviations instead of 1 standard deviation.

Ernie

Ernie

Anonymous said...

Hi Ernie,

Thanks for the response. I like the large payoff strategy - fewer trades/higher rewards...

Carl

Anonymous said...

OK, I've been playing around with GLD/GDX as covered in the book. The ratio of the pair has drifted significantly prior to and following the snapshot covered in the book. In fact, I am finding this "drift" to be a problem in several pairs that I am looking at. The profits can come dripping in and then whoosh - the relationship drifts and a big loss comes in.

Any thoughts on how to manage drift related losses vs mean reversion profits?

Thanks,
Carl

Ernie Chan said...

Carl,
In trading pairs, we should continuously adjust the hedge ratio based on the most recent N-days look-back.

Of course, if the hedge ratio drifts too much, that simply means that the pair no longer cointegrates. Then surely the mean-reversion strategy will not be profitable for this pair. The key is diversification: one should not put all the capital on just one pair.

Ernie

Anonymous said...

Hi Ernie,

How much difference would it make if, instead of using the most recent N-day value of the hedge ratio, you simply enter each trade with equal (absolute value) dollar amounts?

Ernie Chan said...

Hi Anon,
Equal long-short dollar value will work too, but you still need to use the spread based on the hedge ratio as an entry/exit signal.
Ernie

Anonymous said...

Hi Ernie,

In your book you mention that you use IB's Basket Trader to enter (up to 1,000) pair orders before the market opens. Are these market orders based on the fact that the spread at the previous close had widened sufficiently, or am I missing a way to trigger these orders based on the spread during the day?

Ernie Chan said...

Hi Anon,
I said in my book that I entered 1,000 orders by IB's BasketTrader, but I did not say that these are pairs orders. As far as I know, it is not possible to enter pair orders this way.
Ernie

junhua said...

Mr chan, in you book (page 57, example 3.6: pair trading of GLD and GDX), why you choose use equal dollar long short strategy instead of the spread based on the hedgeRatio? I found the sharpeRatio will be better if using hedgeRatio.

Ernie Chan said...

Junhua,
Equal dollar long short just make the code slightly easier to understand, but of course you can use fixed shares-ratio instead.
Ernie