Signals aren't set in stone - so don't be afraid to fine-tune them

The first forex signal I was ever introduced to was the Commodity Channel Index, or CCI. The CCI is the heart of Woodie's CCI system (yes, that's Woodie himself there with Barbara Bush), and my first forex broker recommended I try this straightforward trading method. So I did, just taking my account rep's word for it that it would work. He told me to set up the CCI with 5-minute periods and a range of 14 periods. So I did, again taking his word for it. At this point I really had no idea why 14 periods would be better than 10 or 20 periods, and I didn't know enough to ask, or to test out the theory that 14 was the magic number with historical data. And with that, I started trading Woodie's system, blindly accepting that this off-the-shelf strategy would work for me. Well, it didn't - or rather, I didn't know how to make it work.

At various other times, I've used similar standard set-ups of other signals. For instance, 5 and 20 period MAs, or 20 period Bollinger Bands. If asked why I was using these particular set-ups, the most honest answer I could've given was that "I read about them somewhere" or "Someone told me to do it this way." I had no personal experience or evidence to base my signal choices on, so I simply had to take other peoples' word that they would work. I don't think it was a coincidence that my trading results were poor. Very poor.

It was only later, when I'd begun using extensive historical data to guide my trading, that I really started fine-tuning and customizing my signals. For example, after trying to design a signal around 20 period Bollinger Bands with daily (24-hour) periods , I began testing out various other ranges of periods: 15 period bands, 12 period bands, 10 period bands, and so on. To my surprise, the most powerful signals were derived from 10-period daily Bollinger Bands - far more powerful than bands based on 20 periods, which I'd always understood to be the standard. Lesson learned! Likewise with moving averages - after extensive testing of many different periods, I found that a combination of 5 and 10 period MAs yielded the most significant results. Perhaps just as importantly, I also discovered that moving medians generated some very interesting signals as well. Now I've never seen a moving median mentioned anywhere outside my own spreadsheets, and had it not been for my new-found willingness to test, tweak, recalibrate and recalculate every potential signal, I don't think I would've ever stumbled on them.

Fine-tuning my signals in this manner also taught me that certain isolated events frequently emphasized in trading tutorials and chart-based trading systems were often less important that I'd originally assumed. For instance, a moving average cross, which certainly worth noticing, predicted a lot fewer trades than the ongoing relationship between two moving averages: how far one was from the other, whether the distance was increasing or decreasing, whether their more recent movements were in conflict or agreement with their longer term movements. I think focusing exclusively on chart-based trading led me to overemphasize single events at the expense of the total set of interactions between signals.

The lesson I hope I'm conveying amid all this rambling is that you shouldn't take anyone's word for it that a given signal will work at all, or will work better than other signals. All signals are just equations that someone else invented, and they can all be recalculated, tested, and retested with new parameters and different values. Think of signals as your employees. You wouldn't just take a new hire and put them straight to work - you'd spend some time getting to know them, training them for the tasks you expect them to carry out, and seeing how they perform in different situations. Signals work for you, and it's up to you to make the best use of their abilities.