As you can see, the new version has many of the same performance numbers. Average gain per trade, win rate, and Sharpe are fairly similar. Improvement, however, is significant in two areas: CAGR and drawdown. The CAGR is a little blinding, and there are a number of reasons NOT to look at the overall number (I address these in other posts). Improvement in excess of 20% is real however. This happens mainly because we increased the number of trades, which ever so slightly increases the exposure. In short, we are able to increase the amount of time we have money in the market, which is earning at the same rate. Said another way, version 7 finds us more good trades than it's predecessor.
The other big improvement, and one I am much more excited about, is a 24% reduction in the Max Drawdown. That's big for those who want to sleep at night. How did we do it? As you can see, we are taking a few more trades, but something more must have changed if they continue to earn at the same rate with reduced volatility. This was primarily attained through a new position sizing algorithm which places a premium on risk. It penalizes position size when the edge is lower and rewards it when the edge is greater.
I will gradually introduce v7 trades into the Covestor model, increasing them as I validate the accuracy and quality of the system. This is normal with any system I trade. There is only so much you can do with computerized backtesting, and then the real money comes into play. I step into it very gradually, validating each trade manually, to be sure the system is working as expected. Only after many completed and validated trades do I increase the automation.
Good Trading...
No comments:
Post a Comment