MARKET WIZARDS SEARCH PART 2: THE PERFORMANCE STATISTICS I USE
DAILY DATA VERSUS MONTHLY
If available, I always prefer to use daily data instead of monthly data since the fact that that the number of data points is 21 times larger makes daily-data-based statistics far more statistically reliable. Also, daily data can uncover meaningful information that is hidden by monthly data. For example, if an account loses 10% in the first half of the month and then rebounds 11.2% in second half of the month, that 10% drawdown will be invisible using monthly data, but fully visible on daily data. There is no place to hide with daily data. There is also artificiality inherent in monthly data. If in the foregoing example, the 10% loss occurred in the second half of a month, and the 11.2% rebound occurred in the first half of following month, the drawdown would be visible in monthly data. The fact that the same exact price action can result in very different statistics depending on the calendar days when that price movement occurred is an undesirable feature.
MY PREFERRED PERFORMANCE STATISTICS
Gain to Pain Ratio—The sum of all returns (daily or monthly, depending on data being used) divided by the absolute value of the sum of all negative returns. In essence, the GPR shows the ratio of net returns to the losses incurred in getting those returns.
Sortino Ratio/√2—The Sortino Ratio is a variation of the Sharpe ratio that uses only downside deviation to measure risk instead of the standard deviation, which is based on all returns. The reason for dividing by √2 is explained in the note below. I much prefer the Sortino ratio to the Sharpe ratio because it only penalizes downside volatility, whereas the risk measure of the Sharpe ratio doesn’t distinguish between upside and downside volatility.
Note: Since the Sortino ratio has the same numerator as the Sharpe ratio but calculates the denominator based on the squared deviations of only losing returns, instead of all returns, it will be biased to be higher than the Sharpe ratio, even for traders whose returns are negatively skewed (i.e., large losses are greater in absolute magnitude than large gains). A common mistake is to assume if that if the Sortino ratio is higher than the Sharpe ratio, it implies returns are positively skewed (i.e., large gains are greater in magnitude than large losses). Since the loss measure in the Sortino ratio will be based on summing a smaller number of deviations (i.e., only the deviations of losing returns), the Sortino ratio will almost invariably be higher than the Sharpe ratio. To allow for comparing the Sortino ratio to the Sharpe ratio, we multiply the risk measure of the Sortino ratio by the square root of 2 (which is the same as dividing the Sortino ratio by the square root of 2). Multiplying the risk measure of the Sortino ratio by the square root of 2 will equalize the risk measures of the Sharpe and Sortino ratios when upside and downside deviations are equal, which seems appropriate. The adjusted version of the Sortino ratio allows for direct comparisons of the Sharpe and Sortino ratios. Generally speaking, a higher adjusted Sortino ratio implies that the distribution of returns is right-skewed (a greater tendency for large gains than large losses). And, similarly, a lower adjusted Sortino ratio implies returns are left-skewed (a greater propensity for large losses than large gains).
FundSeeder Score—This metric uses a complex proprietary formula that is computed based on the time-series of daily returns. The core component of the Score is the Probabilistic Sharpe Ratio.
Average Annual Compounded Return—This value is the return level than when compounded annually will yield the cumulative return. Although I pay more attention to the forgoing return/risk metrics than return, it is possible for a performance record to have superior return/risk values and an unacceptably low return level. Therefore, it is still necessary to check return alone.
THE PERFORMANCE LEVELS I LOOK FOR IN MARKET WIZARDS SEARCH
In searching for potential candidates to interview for a Market Wizards book, I look for the following performance thresholds:
Track Record Length—10 years or longer (7 years or longer if performance statistics are exceptional)
Gain to Pain Ratio (Daily Data)—0.30 or higher
Gain to Pain Ratio (Monthly Data)—2.0 or higher
Sortino Ratio/√2—2.0 or higher
FundSeeder Score—Minimum of 60, but preferably 75 or higher
Average Annual Compounded Return—The minimum value I look for here is highly dependent on the return/risk values. For example, a trader with only a 10% average annual compounded return with a daily GPR of 0.6 and a Sortino Ratio/√2 of 3.0 would be of interest, whereas a trader with a 30% average annual compounded return with a daily GPR of 0.1 and a Sortino Ratio/√2 of 0.80 would not be.
HOW TO CHECK YOUR OWN PERFORMANCE STATISTICS
Disclosure Note: I have an interest in FundSeeder.
You can access all the above performance statistics, as well as a lot of other analytics, for free by uploading your return data or net liquidating value (NLV) data using the provided Excel template available at https://fundseeder.com/home. The template will be customized to accommodate either daily or monthly data, and either return or NLV data, depending on user inputs.