^ Exhibit B: This is an assembly of ~12 strategies with about 50 signals on 20+ instruments. The equity curve is much smoother but the backtest is showing Sharpe of 1.62.
I've been relying on my own custom metrics with the custom backtester but would like to understand more about the Sharpe.
The Sharpe Ratio is a dinosaur and a throw-back to old buy and hold equity days. It penalizes you for upside volatility, because back in the day, they though any type of volatility, either up or down was indicative of risk.
I suggest you calculate MAR, Sortino, Calmar Ratios and Frequency. Also, you might want to check for future leaks, other than that it looks good. Hopefully you re-balance between Systems/Methods and are validating cash and margin restrictions.
Sure, I agree. I actually have my own complicated formulas in the custom backtester that combines a lot of metrics. It's a beast of an equation and much fancier than Sortino or Calmar, etc. I believe Amibroker actually automatically calculates MAR as "CAR/MDD."
Regardless of penalizing upside volatility, I would think the Sharpe would still be much higher for the second chart, just because of, well, how low the volatility is. I am wondering how Amibroker is calculating Sharpe when it comes to portfolio backtests versus single-symbol backtests. One of my metrics gives some weight to the Sharpe and works fine when optimizing a single strategy but for optimizing the whole assembly the funky Sharpe is throwing it off.
In the first two instances, there is a relatively enormous spike in equity at the end, so look at that, as well as the obvious return differences. Or simply export the equity and do your own Sharpe calculation outside of AB for comparison.
Thanks that's a good idea. I will export them for my own calculations.
As for the spike, they both have a similar spike. The average drawdown is 20 times less, and max drawdown is 3 times less. That wouldn't be the reason for it.
I'm guessing it has something to do with time-in-market or perhaps it is calculated using arithmetic returns vs geometric returns.
As per the documentation (System test report window), AmiBroker reports the Sharpe Ratio of Trades. This is not the portfolio Sharpe Ratio that most people are familiar with.
Sharpe Ratio of trades - Measure of risk adjusted return of investment. Above 1.0 is good, more than 2.0 is very good. More information The Sharpe Ratio . Calculation: first average percentage return and standard deviation of returns is calculated. Then these two figures are annualized by multipling them by ratio (NumberOfBarsPerYear)/(AvgNumberOfBarsPerTrade). Then the risk free rate of return is subtracted (currently hard-coded 5) from annualized average return and then divided by annualized standard deviation of returns.
It is calculated from trade returns not from equity line.
@Tomasz this seems something to change in the documentation since now in the Analysys Settings/Report tab there is a field to set the risk free rate of return.