When To Change a Losing Game

When To Change a Losing Game

September 6, 2017

When things aren’t going as planned and the breaks are beating the boys, what do we do?

Do we bail on a losing strategy? Or do we battle on?

It’s an important question–maybe the most important question–and it’s an easy one to get wrong.

On one hand, if we’re losing, surely we’re doing something poorly. Losing is bad, and we need to change. Immediately.

On the other hand, if we’re constantly changing, we’re constantly going to get the worst of all worlds, catching all the losing streaks and never staying around long enough to reap any benefits.

So what’s the solution? Let’s first look at sports.

The key to knowing whether to change a losing strategy in a tennis match depends on the type of losing that’s occurring.

For example, let’s say we lose the first set. That’s not good. We failed. We need to change it up, right?


It depends on how we lost. If the match went exactly the way we thought it would but we made way too many mistakes, then there’s no need to change. It’s not the strategy that’s the problem, it’s us. In this case, keep the strategy and just play better. Stop making stupid mistakes.

If we keep the strategy and start playing our best, we’ll win.

If, however, we came out with a plan, didn’t make any mistakes, and lost anyway, that’s a problem. In this situation, something unexpected happened. We didn’t do dumb things and yet we’re still losing.

If we keep on this path, even playing our best won’t win.

We need to make a change if we’re going to pull it out.

A third possibility is a combination of both. Maybe our game plan was to serve everything to the forehand but it turns out that the backhand is the weaker side.

We didn’t make any mistakes and still lost the set. That calls for a change. But not a big change. All we need to do is keep everything the same but make one small change: serve the backhand.

Many times one tiny little tweak is all that’s needed to turn it around.

Finally, we need to consider the overall game plan and how viable it is. If the game plan isn’t as sound as one we’ve used in the past, then maybe losing with a new plan isn’t as smart as going to battle with a plan that’s been successful for a long time. If we’re not making mistakes but aren’t comfortable deep down with the strategy, then losing is a good hint to go back to basics.

In trading, it’s very similar.

If we’re trading a system but making mistakes all over the place, then we’re to blame for the losing. Just stop making mistakes and let the system do its thing.

At the same time, if we’re trading a system perfectly (following all the rules and taking every trade) and we’re in a drawdown that’s historically expected, then it’s time to simply shut up and keep trading.

Sometimes, though, it’s somewhere in-between.

Let’s take a look at the Heron. I have huge hopes for the Heron, as I’ve said, and it looks really good in historical backtesting. This robot should be able to help a lot of people.

However, looking at my myfxbook account, I see that I’m currently down 4.79%. I’ve only been trading the Heron on that account since May 28, 2017, but it’s still not profitable during those few months.

What should I do?

Step one, of course, is to look to see if it’s been traded properly. The answer is no. I’ve made some mistakes and the trading platform has made some mistakes.

On my end, I’ve taken two trades out early (after some especially volatile market conditions), which is poor on my part and not usual behavior. Nonetheless, those are unforced errors.

On the platform side, I lost connection on my VPS early on and missed a full winner. That’s been remedied, but it happened. Furthermore, I’ve had a few winning trades turn into break-even or losing trades due to a substantial widening of spreads at busy times. Those types of things don’t show up on a backtest and can be considered unforced errors by the platform.

And then we have the most recent Heron trade.

On August 31st, my Heron robot took a loser that should never have been a trade in the first place.

The Heron was designed to start trading at the top of the hour during afternoons in the Eastern Time Zone. In other robots I’ve built, to get a trade at the top of the hour, you need to set the “Time of Day” input to one minute before to get the desired execution.

For example, if you want to start trading at 12:00, the other robots needed to be set to start trading at 11:59.

However, the Heron robot took this recent loser at exactly 00:59 minutes after. Instead of starting AFTER the time it was set to, it started trading AT the time it was set to.

In short, it took a trade it never should have taken because it was set to start at :59 after. The code for the Heron is different, so its rules were different.

And, of course, I didn’t know that until that trade on August 31.

Guess what? If the settings are changed to a round number (exactly on the hour) that rogue trade isn’t taken and the whole system is more profitable over the long-term.

The system itself is fine–but it needed one small adjustment.

One last thing.

I mentioned last week that my real-life results have been better than my historical backtesting. That goes for both Tradestation and MT4 testing. That’s a good thing.

However, over the years, I’ve noticed that my Tradestation testing is the most accurate.

In my live myfxbook trading of the Heron, I’ve used settings that were generated only by MT4 research.

In the past few months, though, I’ve tested the Heron methodology on Tradestation and come up with different inputs.

And when I put the Tradestation inputs into MT4, the testing looks even better than the MT4-only testing.

Using Tradestation data makes me more comfortable (because of the good historical performance), so that is an adjustment that looks prudent to make.

In summary, the Heron is losing the past few months.

  1. Mistakes have been made by the human and also by the platform. Those need to be cleaned up (and they have been).
  2. A small tweak was needed due to an unexpected occurrence.
  3. It would be smart to go back to basics. Tradestation data has proven to be very reliable for me for a long time. It’s time to give my most comfortable research a try.

If you’re a member of the Heron Course, I will be adding an attachment to the Course lessons showing the new Inputs. If you’re not a Course member and are interested, you’ll have the option of using either set of Inputs.

It’s always tough to know if we should change a losing game. But if we stick to sound guidelines on what to change and when, we can turn the losing around and maybe win just one for the Gipper.

Or just make a whole bunch of money.