Thanks to our friends from Shark Indicators – www.SharkIndicators.com for the following guest post:
What to look out for when tweaking your trade system:
Last week we covered the how to use a the NinjaTrader Strategy analyzer to backtest a simple classic daily signal composed of the MACD, Stochastic and simple moving averages.
This week we will show you the workflow for making tweaks and changes to your system – an inevitable process when sculpting a system to your personality and risk profile.
Generally you begin by selecting “parameters” to tweak. For example, you can alter your stop values and profit targets (placing your stops further or closer from entry) or change the simple moving average period in your system (speeding up and slowing down the moving average).
A common pitfall for traders looking to design a system is tweaking it so much it that looks great in historically but falls flat on its face in real trading. When tweaking parameters in particular, many falsely think they can hit the “optimize” button and presto – out comes a profitable system that will let you quit your day job and buy that vacation home.
But as soon as they go live with it, they blow up their trading account.
Why is this so? The most common culprit is what we call “curve fitting”. It’s a term that describes “overfitting” a system to past results so closely, that it’s giving you only an optimal performance for exactly what happen in the past. In other words – the system is inflexible, and if you applied it to any other instrument or time frame you’d get horrible results.
When tweaking your system, it is important to keep in mind which changes make the system “fragile”, that is if small changes in certain parameters lead to huge drops in performance. If you find that there is only a narrow window where your system does well with one particular parameter, that is a sign your system is fragile and indicate the need to go back to the drawing board.
The above chart is what a fragile system looks like. On the X-Axis you can see a parameter setting changing very little, but the resulting profit factor (higher the better) swings wildly.
What you want to see instead is more or less consistent performance (a little up, and a little down) when you tweak a particular parameter. This characterizes a robust system, and robustness is far more desirable than optimization.
In contrast to the results of the fragile system, a robust system will show a more consistent profit factor across the a spectrum of parameter tweaks. While the best setting may yield a profit factor lower than the fragile system, the robust system will have a better chance of success in different market conditions.
A robust system is much more preferable to an optimized system since it likely to have better consistency over different market conditions (and even different instruments). And we all know that the market doesn’t stay static right?
Check out the video below, we show you the mechanics of tweaking parameters in your system using SharkIndicator’s BloodHound software, and how to interpret the results in NinjaTrader’s Strategy Analyzer.
Video Link below:
Comments