Charts of my Trading System Results

Here's a picture of the EUR/USD system's backtested performance over 23 months of daily trades...I feel like a proud father :-)



On the vertical axis are the total number of pips gained, with 1 equalling 10,000 pips. So this chart shows over 11,000 pips in (theoretical) profit, before the 3 pip spread is factored in. With 503 trades charted, the highest possible spread costs for all these trades would be 1509 pips. However, the real number would've been a lot less, since many of these are successive trades in the same direction that would get rolled into one trade, saving substantially on the spread. I haven't figured it out exactly but I suspect actual spread costs are around 800 pips at most. If I ever get around to switching to Oanda they'll be even less. I also haven't factored in interest, but the way the trades are divided between long and short it wouldn't have added or subtracted much either way.

For comparison and contrast, here's the chart of my GBP/USD system's backtested performance for the same period.



Note that it's significantly choppier than the EUR/USD chart, which makes sense given this pair's higher volatility. Total profit before spread costs comes out to 11,027 pips over 566 trades, with spread costs probably subtracting around 1200 pips.

While these charts may look pretty good, you have to bear this crucial caveat in mind: I haven't traded for this entire period. The vast majority of the data you see here are results from backtests, not actual predictions. In fact, I've only done one actual trade of the GBP from this system, and it didn't go well. The period that I have traded on a regular basis is at the very end of the EUR/USD chart, from about 11/17/05 onwards on the horizontal axis. Note that this hasn't exactly been a period of stunning success. For me, this raises some interesting questions about how to design and test trading strategies that are successful in the real world, not just on a spreadsheet.

For instance, is it possible that I've created a system that very effectively describes the past couple years of market activity, but doesn't actually work well at predicting it in the future? Are the signals I've identified so specific to past activity that they have no value in forecasting new trends? How valuable is a backtest anyway, when the market's dynamics are always in flux and, as one trader recently put it, "it mutates like a virus"? Are there certain types of signals that are simple and true enough that they'll always work, whereas others are transient and illusory? And do I even have enough data here to base predictions on? Would another couple years of data produce a completely different picture? Yes, it probably would - now if only I could get that data cheap (all the data used in creating this system was acquired for free, incidentally).

All good questions, and tough ones to answer...many of the answers will only be revealed through future trades, and hopefully not at an exorbitant cost. Or if you happen to know any of them right off the top of your head, please leave them in the comments!

Related topic:
A Snapshot of My Forex Trading System's Results