I’ve been working on bigger projects till about a week ago then I got real curious about a trading idea: what happens to S&P 500 next week when commodity currencies are aligned up or down the same day? What happens if I want to reduce the drawdown periods and increase the run-up periods? what happens if I apply Relative Strength to a strategy’s risk management by using it as a Equity Curve “Surfing” method ? (thanks to Market Rewind for introducing me to inter-market analysis and Relative Strength).
So I just ran with it.
In the process I learned several new things:
- This type of specific correlation condition doesn’t work during bull markets most of the time.
- It does work during bear markets. Probably because of seasonal increase in asset correlations.
- How equity curve filters that reduce drawdown periods and increase run-up periods affect the strategy in a total lookback period of 250 days: Run-up periods need more time to run and drawdown periods need to be cut much more quickly.
- How Relative Strength looks like within the context of equity curve risk reduction.
- How to apply newly (to me) used programming methods like arrays, an In-Sample optimization method and even a nice Autofill by user’s referenced column macro in Excel that saves a lot of time in the algo-building process.
I’m aware of several non-kosher statistical factors to the testing method that reduce expected returns in this test:
- Filters have parameters with lookback periods that reduce the degrees of freedom.
- There’s a significant increase in complexity from these filters’ conditions that also reduce Expected Returns.
- The back testing window is 8 years or 1/2 way through to the threshold in the number of years it takes for a strategy’s Expected Returns to flatten out – about 15 years of daily data. Until that threshold there’s a convex slope of % points that one needs to reduce from the arithmetic mean of returns in the back test to adjust for a realistic figure of Expected Returns.