Price Action Lab Blog

Premium Market Analysis

Quantitative trading, Trader education

Quant Funds Must Have Realistic Targets

Unrealistic expectations in the quant world often lead to disaster. The recent market history after quantitative trading was introduced in the 80s is filled with stories of failures of prominent but also less known funds that had unrealistic expectations.

I will only allude to a few funds known to almost everyone that outsource analysis to a large number of aspiring quants around the world in an attempt to find gold in data-mining bias. Some of these funds have more realistic targets but at least one of them has unreal expectations. I will not get into the details because my objective here is not to attack some otherwise good people that apparently have unrealistic expectation of what the market can offer. But every quant, especially the new ones, must understand that on top of the increase in data-mining bias as a function of time due to reuse of old data, there is also a fundamental regime change: the inflow of dumb money willing to provide profits to quant funds has decreased substantially.

I will not go into beta requirements for strategies developed by quants imposed by some funds because in most cases they are ridiculously strict. I will just concentrate on Sharpe ratio, which is the most popular measure of quantitative performance.

To make a long story short, the buy and hold Sharpe of SPY ETF total return since inception is around 0.51 while the annualized return is a respectable 9.24%. Of course, this performance came at high maximum drawdown and volatility of about 18.24%. How unrealistic is a target of a Sharpe of 2?

Suppose you have a magic forecasting tool that could forecast all daily returns in SPY greater than 4% with 100% hit rate. Is it possible? I say no, probably some think it’s possible. There have been 32 such occasions in the history of the ETF and below is the equity curve after buying at the close before the forecast and selling at the next close. Commission is $0.01 per share and equity is fully invested.

Even after assuming a super-forecasting ability, Sharpe is just above 1 but annualized return is much lower than buy and hold. Some could argue that this performance can be leveraged but this is known only in hindsight. What about if the extraordinary forecasting promise did not materialize? (Does this sound familiar?) Besides, this is not a strategy but an example to make a point about unrealistic quant fund targets.

Fine then, the 4% will not do it, why not hiring a clairvoyant to guess all daily returns greater than 3% in SPY at a 100% hit rate. This is easy in hindsight. Below is the equity curve.

Now we get outperformance but Sharpe is still below 2 at 1.70.

What to think of those very high Sharpe values reported in simulations in some online platforms that outsource quant work to anyone with an Internet connection?

I have seen Sharpe of 2, 3, 4 or even higher. These are better than the clairvoyant who correctly guessed the exact timing all daily returns in SPY greater than 3%. Most high Sharpe results are flukes generated by data-mining. But those who are not familiar with the problematic aspects of constant strategy testing are always happy due to high expectations. The fact is the more you backtest, the higher the probability that you will be fooled at the end. See this article for more details.


If you found this article interesting, I invite you follow this blog via any of the methods below.

Subscribe via RSS or Email, or follow us on Twitter

If you have any questions or comments, happy to connect on Twitter: @mikeharrisNY

Disclaimer

Leave a Reply

Theme by Anders Norén