How to Turn Your Trading Robot into Your Servant, Rather Than Your Master

One of the most entertaining and thought-provoking interviews I listened to recently was a Two Blokes Trading podcast that featured Will Hunting who is really a kindred spirit of mine.

Will, who is a discretionary trader, rips apart all the conventional data driven platitudes that pass for “modern trading advice” for the retail trader. Namely, he takes issue with the idea that you need thousands and thousands of data points in order to prove your strategy “right.” Specifically, Will makes the counterintuitive point that the more data you have -- the less valuable your signals will be. Something that worked in 2011-2013 is very unlikely to work today even if the overall equity curve of the strategy is positive.

When I was a young trader I remember that like every newbie, I was enamored with rising equity curves -- the longer the better. Until one day I took a closer look at a system that was wildly positive over the past decade only to realize that it made equity highs 18 months ago and was actually slowly losing money ever since.

This death by a thousand cuts, or a lobster slow boil is the most common problem that trips up systematic traders. They do all the right things only to wind up with all the wrong results, or as Will put it in the interview, tongue firmly planted in cheek, “I have a lot of respect for professional system traders who keep going until they go broke.”

The point being that all systematic trading is the application of a static model to dynamic price action and while the model is important -- critical even -- to consistent trading success, it needs human oversight. Discretionary trading in the true professional sense is not just random placing of trades by “feel”, but the rather judicious use of your model under live market conditions. In short, good discretionary trading looks to minimize the selection of “bad” trades in your model.

Now I know that this is much harder to do than it sounds -- and it certainly requires experience and judgment, but in the end, I think it is the best way to trade.

Which got me thinking. In retail FX, we have the great benefit of encoding our trading models into MT4 EAs which do all the clerical drudgery of culling through price data to find the trades but then take every signal indiscriminately.

So one way to improve that is to run the EA on a demo account and then have the signals sent directly to your smartphone. If you like the setup you can place the trade on your real account. If not, you can pass it up. You are still using your trading model but you act as the human filter and this “pause” provides you with more control and more accuracy. It’s no panacea, but it is an intelligent way of introducing discretion into a formal trade model.

Boris Schlossberg

One comment

  1. Kyle says:

    Indeed, I do something similar but with TradingView alerts. Sends a signal and then I evaluate if the trade actually makes sense before entering.

Leave a Comment

Your email address will not be published. Required fields are marked *