Forex Forays: My Signal Flipper

I had a pretty ugly run of failed EUR/USD short signals in the past month or so, so ugly that I stopped my manual trading entirely for a while and switched over to automated trades while I rethought my original strategy and tested out some new ones. One of the new strategies I considered seemed a little strange, since it called into question a lot of my assumptions about trading signals, but maybe strange enough to work. It's a signal flipper, which turns my short signals into long signals during periods when they're failing miserably.

The flipper is triggered when the success rate of my short signals falls below 40% over the previous 20 trades. So over the past several weeks, it would've converted almost all my short signals to long signals, and a steep loss to a profit (theoretically at least). The backtested results on a graph look promising, and most importantly they're consistent over a few years of price data: the flipper does what it's supposed to, even in periods outside the original timeframe I was worried about.

There are of course pros and cons to the signal flipper. The fact is, it doesn't really seem to add to the total profits - it just smooths out the earnings curve and dramatically reduces drawdowns, making it less likely I'll blow my account in the midst of erratic market behavior. For me this is a really big deal, since not only are drawdowns dangerous to the financial health of my trading career, but they make me call my whole trading system into question, resulting in low confidence and poor execution.

And as I mentioned, it calls into question certain assumptions about my trading signals. Frankly, when I create a short signal I want it to stay a short signal, not act as a long signal in certain market conditions. But that's what the principle of the trade flipper implies. It doesn't care what a signal's supposed to do, it just looks at what it's actually done for me lately and flips it short or long accordingly. Is that a good or bad thing? You tell me...but if it smooths out my trading performance and saves me from some nasty drawdowns, I certainly won't complain.

Related topics:
Creating Self-Tuning Signals
Your Worst Short Signals May Be Great Long Signals (and Vice Versa)

Labels: Signals

2 Comments:

At 6:26 PM, Jason said...: Hello, how long would the flip stay active? When would you determine that your signals can now go back to normal? Thank you.
At 6:34 PM, The Riverman said...: Good question - here's how it works: my standard set of signals run the same way they always have. However, I've now added a tracking equation that constantly monitors the outcome of those signals - basically, a moving average of their success rate over the past 20 trades. When that average is 40% or higher, the short signals continue to be short signals. When the tracking average drops below 40%, they switch over to being long signals. So there's basically a meta-signal running on top of my standard signals that interprets those signals as short or long based on their past performance. Hope that makes sense!