Trading Patterns Have an Expiry Date

There’s a version of this story that almost every systematic trader goes through. You find a pattern, validate it properly, run it for months and it performs well. Then gradually — not in one blow-up, just slowly — it starts underperforming. The win rate drifts down. Drawdowns get longer. You wonder if you’re in a bad streak or if something has actually changed.

Most of the time, something has actually changed.

Patterns in financial markets are not physics equations. They don’t hold indefinitely. They work for a period of time because market participants are behaving in a particular way, and they stop working when those participants adapt, when market conditions shift, or when the macro environment changes enough that the old behaviour no longer makes sense.

Why edges exist and why they disappear

An edge in trading exists because a group of participants is reliably doing something predictable. Institutions accumulating positions at specific levels, retail traders placing stops at obvious points that get hunted, momentum chasers piling in after breakouts. These behaviours are exploitable because they repeat.

But markets are adaptive systems. When an edge becomes well-known, participants start to anticipate it and the edge gets arbitraged away. This is why every retail trading strategy that becomes popular on YouTube gradually stops working — not because the people teaching it were lying, but because enough people traded it that the other side of the market adapted.

Even without explicit adaptation, the behaviour that created the edge might simply stop occurring. If your RSI oversold pattern worked because gold was in a consistent uptrend where dips got bought by institutional longs, it will stop working the moment gold enters a different phase — a ranging environment, a downtrend, or a period of macro uncertainty where institutions are neutral rather than consistently positioned.

The pattern didn’t break. The conditions that made it work changed.

Market regimes and what they mean for your patterns

The concept of a market regime is important here. A regime is a persistent state that the market is in — trending, ranging, high volatility, low volatility, risk-on, risk-off. Most patterns work in some regimes and not others.

A momentum-following pattern (price above MA200, RSI cross upward, enter long) works well in trending regimes. In ranging regimes, it will get chopped up as price oscillates around the moving average with no follow-through.

A mean-reversion pattern (RSI oversold, price touches lower Bollinger Band, enter long) works well in ranging or mildly trending regimes where price reverts to mean. In a strong downtrend — where oversold conditions simply stay oversold — the same pattern will get crushed as price continues lower after every entry.

The problem is that regimes don’t announce themselves. You don’t get an alert that says “trending regime ended, ranging regime started.” You infer them from what the market is doing, and by the time the pattern is clearly failing, you’ve already taken the losses.

This is why periodic revalidation is not optional if you’re running systematic strategies seriously.

How to tell if your pattern is expiring

There are a few practical signals that a pattern’s edge is deteriorating:

Win rate drifts consistently lower over 4-6 weeks. Not a bad week — a consistent trend. If you track your live results week by week and the win rate is on a clear downward trajectory over a month or more, that’s a regime signal, not bad luck.

Drawdowns take longer to recover. A healthy pattern in a good regime has drawdowns that recover within a reasonable timeframe. If your pattern is taking 3x longer to recover from losses than it did historically, the environment has changed.

The pattern stops triggering. Sometimes a pattern doesn’t fail — it just stops firing. If your conditions become so rare that the bot goes weeks without a signal, the market has moved into a regime where those conditions don’t arise. This is worth knowing too.

The losses come from a specific type of situation. Review your losing trades and look for a pattern within the losses. If most losses are happening when a specific macro condition is present (e.g. during dollar strengthening periods, or in the week after a central bank decision), that’s useful information about where the edge is breaking down.

The practical approach: set a review schedule

Rather than waiting for failure, build periodic review into how you run your strategies.

Every 4-6 weeks, pull your live results alongside a fresh backtest on recent data. Compare the win rate on the last 6 months versus the win rate on the original validation period. If recent performance is meaningfully lower — say more than 8-10 percentage points below your validated baseline — that’s worth investigating rather than just waiting out.

Re-run your patterns through the backtester on the most recent data. A pattern validated on 2022-2023 data should be re-run on 2024-2025 data periodically to confirm it still holds. If it does, keep running it. If it fails on recent data, you have your answer: the regime changed.

Some traders cycle patterns intentionally — running a set of patterns for a fixed period, then rotating to freshly validated ones. This is more work but it keeps your approach current with the market rather than always fighting the last war.

What to do when a pattern expires

The tempting response when a pattern starts failing is to tweak it — adjust the RSI threshold from 30 to 28, change the MA period, add another condition. Resist this unless you have a logical reason to make the change. Tweaking a failing pattern on live data is a form of the same overfitting that caused the problem in the first place. You’re fitting the pattern to recent losers.

The better response is to go back to the backtester, run fresh discovery on recent data, and find what is actually working now. Treat it the same as the first time: discover on training data, validate on unseen data, only deploy what survives both.

Markets reward people who adapt, not people who hold on to what worked before.

The honest truth about pattern lifespans

There’s no reliable rule for how long a pattern will last. Some edges persist for years because the structural behaviour behind them (liquidity grabs, institutional accumulation at levels) is deeply embedded in how markets work. Others last months because they depended on a specific macro environment that doesn’t repeat.

What you can control is how quickly you notice when something has changed, and how quickly you adapt. A trader who re-validates every six weeks and rotates strategies when they decay will consistently outperform a trader running the same strategy for three years wondering why it stopped working.

The pattern is not the edge. Your process for finding, validating, and retiring patterns on the right schedule — that’s the edge.