Stats Backtesting and evidence Sharpe uncertainty

A Sharpe of 1.0 needs 2.71 years of daily data to beat zero

Annualised Sharpe ratioYears of daily returns needed
0.510.83
1.02.71
1.51.21
2.00.69
3.00.31

Source: Bailey and Lopez de Prado, minimum track record length, daily returns at 95% confidence against a benchmark of zero. The requirement goes with the inverse square of the Sharpe ratio, so halving the Sharpe quadruples the time. Non-normal returns lengthen every row.

A Sharpe ratio is an estimate from a sample, so it carries an error bar. An annualised Sharpe of 1.0 takes about two and three quarter years of daily returns to separate it from zero. At 0.5 it takes nearly eleven years.

2.71 yearsSharpe 1.0 needs
10.83 yearsSharpe 0.5 needs
0.188Standard error at T = 60
up to 65%Root 12 overstatement

Members only

The rest of this page is for members

Below this point there are 6 sections, 1 chart, 1 table and 4 named sources, roughly 1250 words of it. Every figure carries the source it came from and the date the data is from.

You can keep browsing every statistic in the library for free. The intro and the summary are always open.