Designing a Trading System, Tip #1: The Data is Your Friend

When I first started playing around in the forex markets I was using an off-the-shelf system handed me by the account exec at the broker I was then working with (Global Forex Trading). It was Woodie's CCI system, a well-known short-term trading system that uses the CCI signal on a 5-minute chart. While Woodie's system is very straightforward and I have no doubt it works very well in the right hands, mine were definitely not the right hands. The 5 minute chart is a jumpy, jerky, volatile animal, and staring at it for hours on end, far into the night, can make you more than a little crazy. The timeframe just wasn't healthy for a somewhat high-strung and obsessive newbie trader like me.

It was also very hard to determine how well Woodie's system would perform if executed consistently over months and years. How many trades would it get right, on average, and how would the good and bad trades stack up over time? (Assuming the trader knew what they were doing -- another problem, since I didn't.) I just didn't have that data and didn't know where to find it, with the result that I was operating in the dark, not knowing how many trades I could expect to get wrong or right, not knowing what types of gains I could expect from a good trade, not knowing how long it took to get good at this system...not knowing anything, really, except the basic signals and the timeframe.

What I needed, and what I eventually decided was the key preprequisite to any forex system, was data, as much of it as possible. Without data you're left guessing at too many different factors, and the more you're left guessing, the more likely you are to panic and do something stupid: a surefire way to zero out your funds in a few weeks.

After flailing around hopelessly with the CCI system, I set about creating a data-driven system that would operate on a timeframe that was less likely to drive me crazy (24 hour intervals rather than 5 minutes) and that could be thoroughly tested via the best statistical tool I had available (MS Excel). The key element was historical price data that I could use to generate testable signals. But where could I get it, preferable for free? Initially I downloaded a large table of daily data from Oanda's historical price tool. But this table only offered the daily average prices, and didn't have opening, closing, high, and low prices. I wanted more data.

Fortunately I stumbled on a mention of the Metatrader platform in the Oanda discussion forums, so I downloaded it and quickly discovered it offered a huge amount of free historical price data on all the major currency pairs. (If you have the software installed, look under Tools > History Center.) With this data downloaded into a spreadsheet, I was able to begin trying out different formulas and seeing how they performed over almost two years of daily market activity. (Obviously more than two years would be even better but I haven't found this data available for free...yet.) I could determine the total gains from each signal, the average gain per trade, the maximum drawdown, and all sorts of other data points that reassured me I had a system that could work. I could also begin combining signals into a more complex formula that would work under a variety of market conditions.

The result? I now have enough confidence in my system that I can trade it consistently, without panicking, and without staring obsessively at charts trying to second or third-guess or fourth-guess the next market trend. Oh yeah, and it makes money. That's good too. (But who knows, this post may have jinxed the whole thing and I'll have to start over from scratch. Like many traders, I've got a superstitious streak. And no amount of data can get rid of that ;-)

So, to summarize: the data is your friend. Find the data and you'll find the money.