When should you change your trading strategy?

Change a strategy when the evidence says it does not work, not when the last few weeks felt bad. Those two things feel identical from the inside, which is why the decision has to be governed by conditions you wrote down before the bad weeks started. A losing streak is not evidence. A large sample with negative expectancy is.

The default should be strong inertia. Most traders change far too often, and the switching itself is a major reason they never find out whether anything works.

The strategy-hopping cycle

It goes like this. You adopt a method. It has a poor month, which every method does. You conclude it is flawed and find another. That one has a poor month too, because you arrived during the conditions that do not suit it, and by now you have concluded that the first one might have been fine after all.

Two years pass. You have traded six strategies, each for six to ten weeks, and you have never once accumulated a sample large enough to say anything about any of them. You are not unlucky. You have designed a process that guarantees you only ever experience the noisy, uninformative part of every approach.

The underlying error is treating short-term results as information. They are mostly randomness, for the reasons in what a normal losing streak looks like.

What justifies changing a strategy?

1. A large sample shows negative expectancy. Not twenty trades. At least a hundred, ideally more, calculated properly. If the average trade genuinely loses money over that sample, the strategy does not work and no amount of discipline will fix it. The method is in expectancy.

2. Performance has broken outside its tested range. If your testing showed a worst streak of eight and a worst drawdown of twelve percent, and you are at eighteen losses and twenty-five percent, something has changed. This test only exists if you did the backtesting first, which is one of its underrated benefits: it gives you a boundary to compare against instead of a feeling.

3. The market structurally changed in a way that removes the edge. Rare and frequently misdiagnosed. A rule that exploited a specific inefficiency can genuinely stop working when the inefficiency is arbitraged away or a market’s rules change. The bar here is high, because “the market changed” is also the most comfortable explanation for an ordinary drawdown.

4. Your circumstances changed. A new job, a new schedule, a different account size. If your strategy needs you at a screen at 3pm and your life no longer allows that, the strategy has to go, regardless of how good it is. Fit matters as much as edge, which is most of the argument in swing trading vs day trading.

What never justifies a change?

Recent losses.

That is the entire section. A run of losses is the expected behaviour of a working system. If your strategy wins forty percent of the time, a six-loss streak arrives regularly and a ten-loss streak is not remarkable across a year of trading. Changing in response to it means changing in response to the thing you were told to expect.

The tell is timing. If the urge to change arrives during a drawdown and disappears after two winners, it was never analysis. It was discomfort looking for a justification.

Set the review triggers before you need them

You cannot judge this fairly in the moment, so do not try. Write the conditions in advance, into the plan itself, as described in how to write a trading plan:

• A minimum sample before any strategy change is considered at all, for example one hundred trades.
• A scheduled review, monthly or quarterly, as the only time changes may be made.
• Specific numbers that trigger a deeper review: a drawdown level, a streak length, a monthly loss.
• A cooling-off rule: no changes within a set number of days of a significant loss.

Then the decision is not a judgement call under stress. It is a matter of checking whether a condition was met.

Adjusting is not the same as replacing

Not every change is a strategy change, and the distinction is worth keeping clean.

Parameter tweaking is usually a bad sign. Moving a lookback from 20 to 22 because recent trades would have been better is curve fitting applied to live results, and it is how a tested strategy quietly becomes an untested one.

Execution fixes are almost always legitimate. If your journal shows that you exit winners early, the fix is not a new strategy, it is fixing the exits. Most people who think they need a new strategy need to trade their current one properly. That distinction only becomes visible if you tag trades as in-plan or out-of-plan, which is the practice in why your trading journal beats your next strategy.

Risk changes are legitimate at any time, in one direction. Reducing size is always allowed. Increasing it should follow a rule, not a mood.

Adding a filter sits in between. If a hundred trades show that your setup fails badly in one identifiable condition, excluding that condition is reasonable. But test it on data you did not use to find the pattern, or you have simply fitted the noise.

The honest sequence

Before concluding a strategy is broken, work through this in order. Most people never get past the first two.

1. Did I actually follow it? Compare in-plan and out-of-plan trades separately. Very often the in-plan subset is fine and the out-of-plan trades are the entire loss.
2. Is the sample big enough to say anything? Under a hundred trades, be very sceptical of any conclusion.
3. Is the drawdown inside the tested range? If yes, this is expected behaviour, not new information.
4. Are conditions genuinely different, or just unfavourable? Unfavourable conditions pass. Structural change does not.
5. Only now, consider a change.

The uncomfortable part

Sometimes the answer really is that the strategy does not work, and the arithmetic will say so plainly. Accepting that after a hundred well-executed trades is not failure, it is exactly what the process is for.

What you want to avoid is never finding out either way, which is what strategy-hopping guarantees. Trading one adequate method properly for two years will teach you more than trading six promising ones badly.

The review process I run, and the triggers that sit in the plan, are in The 9-to-5 Trader.

Educational only, not financial advice. Trading carries a real risk of loss.