Strategy Fatigue: Why Your Winning Algo Might Be Losing Now
Strategy fatigue is when an algo strategy that used to make money slowly stops working, not because of a bug, but because the conditions it was built for have changed. Common causes are a shift in market behaviour, an edge that was partly overfitting, too many traders running the same idea, rule changes such as new expiry days or lot sizes, and rising execution costs.
Fatigue or a normal drawdown?
This is the first question, and getting it wrong is expensive either way. Stop a healthy strategy in a normal drawdown and you lock in the loss. Keep running a tired one and the losses keep coming.
| Signal | Normal drawdown | Strategy fatigue |
|---|---|---|
| Depth | Within the backtest's worst drawdown | Beyond it, or a long slow slide |
| Pattern | A few large losses, then recovery | Win rate and average win both drift down over weeks |
| Trades | The strategy trades as it always did | Fewer valid signals, or signals that behave differently |
| Market | Ordinary volatility | A clear change in conditions or rules |
| Live vs backtest on the same days | Similar | Diverging |
Why strategies get tired
1. The market regime changed
A trend-following strategy that did well in a strong rally will struggle when the market goes sideways. An option-selling strategy built in a quiet period will suffer when volatility rises. Nothing is broken; the strategy is simply in conditions it wasn't built for.
2. The edge was partly overfitting
If the strategy's settings were tuned hard on past data, part of its backtest profit was noise. That part disappears live, and what looks like fatigue is really the gap between the fitted result and the real edge showing up over time.
3. Crowded trades
When many traders run the same idea, such as selling the same strikes at the same time, the edge can shrink: premiums at popular times may get thinner and fills worse. This is hard to measure directly, but it is a common explanation for widely shared strategies fading.
4. The rules of the market changed
This one is often overlooked. Indian index derivatives have changed a lot recently, and strategies built on the old setup can fail even if their logic was sound:
Each change altered which contracts exist, when they expire, or how big a position is
- A weekly BANK NIFTY expiry-day strategy has nothing to trade; BANK NIFTY now has monthly expiries only.
- A strategy timed around a Thursday NIFTY expiry is now on the wrong day; NSE expiries are on Tuesday.
- A strategy sized for an old lot size now trades a different exposure: NIFTY's lot is 65 and BANK NIFTY's is 30.
See what to check on live strategies after the lot size change.
5. Execution got worse
Slippage can creep up as you increase size, trade thinner strikes, or as market conditions change. A strategy whose average profit per trade is small is the first to suffer. Compare your fill prices with signal prices every month; read understanding slippage.
Warning signs to watch
- Win rate drifting down over several weeks, not just after one bad session.
- Average win shrinking while average loss stays the same or grows.
- Drawdown beyond the tested worst, or recovery taking much longer than before.
- Live results diverging from a fresh backtest of the same period.
- Slippage or missed fills rising.
- An urge to override or tweak it every day. That's often your instincts noticing something the numbers haven't confirmed yet.
What to do about it
- Rule out operational problems first: broker connection, rejected orders, wrong contract. See why did my algo strategy fail.
- Re-run the backtest with recent data and realistic costs. Did the recent period look different from the past?
- Find what changed: volatility, trend, contract rules, execution. Fatigue with a clear cause is easier to fix.
- Test a targeted fix, such as a market filter (trade only when volatility or trend conditions suit the strategy), a different expiry, or a different entry time. Change one thing at a time.
- Run the fixed version Live Offline alongside the original, and compare them on the same days.
- Reduce size while you investigate, rather than stopping blind.
- Retire it if the reason it worked is gone. Some strategies have a natural lifespan.
How to make strategies last longer
| Practice | Why it helps |
|---|---|
| Simple logic with a clear reason to work | Fewer settings to overfit, easier to see when the reason is gone |
| Test across different market conditions | You know in advance where it struggles |
| Run several uncorrelated strategies | One tired strategy doesn't sink the account |
| Review on a schedule | Catch fatigue early without constant tinkering |
| Track live vs backtest every month | The earliest objective warning sign |
| Keep a record of every change | You can tell which version actually worked |
Frequently asked questions
What is strategy fatigue in algo trading?
Strategy fatigue is the gradual decline of a previously profitable strategy because the conditions it relied on have changed: market behaviour, the rules of the instruments it trades, crowding, or execution costs.
What is the biggest cause of strategy fatigue?
Changes in market conditions are the most common, such as a shift from trending to sideways or from calm to volatile. In Indian F&O, rule changes to expiries and lot sizes have also been a major cause recently.
How do I know if my strategy is tired or just in a drawdown?
Compare the current drawdown with the worst one in the backtest, and look for a broad, steady decline in win rate and average win. A drawdown within the tested range with normal trading behaviour is usually not fatigue.
How often should I review my trading strategy?
Check live results against the backtest every few weeks, and do a deeper review each quarter. Review immediately after a rule change that affects the instruments you trade.
Should I delete a strategy that is underperforming?
Not straight away. Reduce size, investigate the cause, and test a fix Live Offline. If the reason the strategy worked has gone, retiring it is the right call.
Can strategy fatigue be avoided?
Not completely, because markets keep changing. Simple logic, testing across different conditions, diversification and regular reviews make it less likely and easier to spot early.
