Even at 80%, a bad month is part of the deal

Bad months happen to good strategies. That's the part most trading content avoids saying out loud, because it's harder to sell. But the math is unambiguous: even a strategy with an 80% win rate — higher than what most professional traders carry — will produce ugly months several times a year. Not occasionally. Several. The cause is variance, and variance cannot be removed by skill.

The expensive mistake is misdiagnosing the bad month. A trader who shrugs through every drawdown loses the ability to spot a genuinely broken edge. A trader who panics during normal variance abandons strategies that were working. Both end the same way — capital bleeds. The situations look identical from the inside: same red P&L, same emotional weight, same urge to do something. They require opposite responses. Telling them apart is the real job during a drawdown.

This piece is the diagnostic. Four questions you can run before changing anything, the math behind why even mathematically-perfect strategies must have rough stretches, and what to actually do while the answer is still ambiguous. Skip the questions and you'll either hold a broken strategy too long or kill a working one too early. Both are common. Both are avoidable.

Why bad months are mandatory

Take a strategy that wins 80% of its trades — a strong, well-tested win rate. Put 100 trades through it in a month, which is a reasonable cadence for an active multi-pair systematic trader. The number of wins isn't going to be 80 every time. It'll cluster around 80, but with a measurable spread.

The math behind that spread is the binomial distribution, one of the oldest results in probability. For a strategy with win probability p across n independent trades, the standard deviation of the win count is √(n·p·(1-p)). At p=0.8 and n=100, that's √(100·0.8·0.2) = √16 = 4 wins.

In plain English: in a typical month you'll see 80±4 wins. About two-thirds of months will land between 76 and 84 wins. About 95% of months will land between 72 and 88. And — the part most traders never internalise — about one month in twenty will land outside that range entirely, without anything being wrong.

A 70-win month at an 80% strategy isn't a sign anything's broken. It's the kind of run that statistics says happens roughly once a year. If you've never had one, your sample is too small to claim anything about your win rate yet — that's the how many trades before a win rate is real problem from a different angle.

The number gets dramatically uglier when you factor in dollar variance. A win at +3R contributes meaningfully more P&L than a loss at -1R. So even the same number of wins can produce different P&L months depending on which specific setups won and which lost. The dollar version of the bell curve above is wider and flatter than the wins-count version. A bad-by-P&L month is even less rare than a bad-by-WR month.

So: bad months are not a bug. They are part of the contract you signed when you accepted that no strategy wins 100% of the time. The question isn't whether they'll happen — it's how to tell, while you're inside one, whether this particular bad month is the one that needs a response.

Question 1 — Is the sample big enough to mean anything?

The first thing to check is whether you actually have enough data to be diagnosing anything at all. A 5-trade losing streak is meaningless. A 15-trade run that's 6/15 wins, on a strategy you've claimed has an 80% win rate, sounds catastrophic until you do the math: at p=0.8, getting 6 or fewer wins in 15 trades is unusual but not extreme — happens about once every 200 such 15-trade windows, which sounds tiny until you remember that an active trader runs through a 15-trade window every couple of weeks.

The honest sample-size threshold for "this is starting to be evidence" is around 30 trades. Below that, you're looking at noise interpreted as signal. Above 30, the standard deviation of your measured win rate begins to narrow enough that meaningful comparisons become possible. By 100 trades, your error bar is tight enough that a 10-percentage-point WR drop is genuinely concerning. Below 30 trades, even a 30-percentage-point WR drop might be normal.

Implication: if you've taken fewer than 30 trades since whatever felt-bad started, you don't have a diagnostic yet. You have a feeling. The right move is to keep trading the rules without changing anything, and re-check in 20 trades.

Question 2 — Has the win rate dropped, or just the average R?

These look the same on a P&L graph and feel the same emotionally, but they imply completely different problems.

Win rate dropped, average R per trade unchanged: Your entry filter has degraded. Either you're firing on lower-quality setups (a discretionary trader letting standards slip), or the strategy's filters were calibrated against a market regime that's no longer the one you're trading in. The fix is at the entry side: stricter filters, fewer trades, or temporarily pausing on the symbol where the WR drop is concentrated.

Win rate intact, average R dropped: Your exits have degraded. Stops are firing too soon (volatility increased without you adjusting), targets are getting too close (you're cutting winners short), or you're letting losers run past your planned stop. The fix is at the exit side: re-anchor your stop and target distances to current volatility, and audit whether you're executing the planned exit on closed trades.

The diagnostic you need to run is depressingly simple: open your trade log, calculate WR for the last N trades, calculate average R for the last N trades, compare both to your historical baselines. If you don't have a trade log, you have no diagnostic at all — your account balance shows a P&L number but tells you nothing about which dial slipped. The trade log on this site computes both metrics automatically and exports as CSV; any equivalent tool works fine.

Question 3 — Has the market regime shifted?

Strategies are built for specific conditions. A trend-continuation strategy will print money in trending markets and bleed in choppy ranges. A breakout strategy gets crushed by a market full of false breakouts. A volatility-selling strategy explodes when volatility expands. None of these failures mean the strategy is broken — they mean the conditions changed, and your strategy was a bet on conditions that no longer hold.

The diagnostic question is: was the strategy designed for the conditions it's currently in? If the answer is "no," you don't have a broken edge — you have a regime mismatch. The fix is either waiting for the regime to flip back (it usually does), reducing size during the mismatch, or having a second strategy calibrated for the other regime.

The fix is not re-tuning the parameters of the original strategy to fit the new regime. That's overfitting to recent noise, and it produces a strategy that wins on a backtest including the recent stretch and loses the moment conditions revert. Strategies built on fitting the last six months are fragile for exactly this reason.

How to know which regime you're in is its own question and not a small one — but a basic check is whether realised volatility, trend persistence, and average daily range are running meaningfully different than the average over the strategy's design period. Three or four major data points usually settle it.

Question 4 — Are you executing the rules?

The most common cause of a "broken edge" is that the trader has quietly stopped following the strategy. Not deliberately — but in small ways that compound.

Skipping setups you don't like the look of, even though they meet the criteria. Sizing up after a winning streak, sizing down after a losing one. Moving stops mid-trade because the position has gone slightly against you. Taking trades the strategy didn't generate because you have a feeling. Closing winners early because you're "protecting profit." Each of these breaks the math you backtested. Run any of them often enough and your live results stop matching the strategy's expected results — not because the strategy is broken, but because you're not running it anymore.

This is the question to ask first, last, and at full honesty. The diagnostic: print every trade from the bad period and check whether each one followed the written rules. Not "felt right." Not "made sense in the moment." Followed the rules, on a checklist, every box ticked. If the answer is "mostly, but I skipped a few" or "I sized different on the obvious ones," you don't have a broken edge. You have an execution drift problem, and the fix is mechanical, not strategic.

This is also the failure mode that hides best behind a "the market changed" narrative. The market does change. But the trader who tells themselves the market changed every time they had a bad month is the trader who never has to ask whether they're still running the strategy at all. Be your own toughest auditor here.

How unusual is this month, actually?

Tools that go with this

If you want to put hard numbers on "how bad can the bad month get," two tools cover the math directly:

• Losing-streak probability calculator — feeds your win rate into the binomial-streak math and tells you the likely worst run in N trades.
• Win-rate confidence interval — shows how wide the band around your observed win rate is — useful for telling 'edge broken' from 'sample noise'.

The math behind that calculator is the same binomial/normal approximation as the bell curve above. Standard deviation = √(n·p·(1-p)). Your z-score = (actual_wins − expected_wins) ÷ SD. The probability is the cumulative normal at that z. The verdict thresholds are conservative — "genuinely unusual" doesn't kick in until you're past two standard deviations, which is roughly the boundary where most statisticians stop calling something noise.

The trap most traders fall into: they look at the felt magnitude of the loss instead of the statistical magnitude. Down 8% feels twice as bad as down 4%. Statistically, down 8% in 60 trades at 80% WR might be one z-score and down 4% in 15 trades might be two. The first is normal; the second is the more concerning one. The calculator above does the math your gut won't.

Where the questions point you

The flow above is deliberately ungenerous to the "edge is broken" verdict. By the time you've passed all four questions — sample size big enough, win rate vs average R diagnostic done, regime confirmed unchanged, rules followed exactly — you've eliminated the more likely explanations. What's left is genuinely a strategy that has stopped working, and that's a real diagnosis worth acting on. Most traders skip the four questions and jump straight to the broken-edge conclusion. Most of the time they're wrong.

Quick self-check

What to do while you're inside one

Three rules that hold whatever the verdict turns out to be:

Don't change parameters mid-stream. Tweaking your stop, your sizing, your entry filter, or your risk percentage during a bad stretch is the most expensive thing you can do. You're now overfitting to recent noise — building a strategy that retroactively fits the last 30 trades and forwards-tests poorly on everything else. The cost shows up two months later when conditions revert and your hastily-modified strategy now loses on what used to be its best setups. Decide on changes from a quiet head, with a full sample, in writing, outside the drawdown.

Log every trade, in detail. When the diagnosis is genuinely ambiguous, the only thing that resolves it is data. Trade time, symbol, entry, stop, target, exit, R-multiple, planned vs actual size, and any rule deviations — captured per trade, in one place — is what lets you run the win-rate-vs-average-R split, the regime check, and the execution audit honestly. Without a log you're guessing about your own behaviour, and people guess favourably about themselves.

Don't size up to recover. This is the failure mode that converts variance into a real, irreversible loss. The instinct after a string of losses is to make the next trade bigger so a win catches you up. The math says the opposite — the variance compounds against you, and a sequence of bigger losses arrives faster than you expected. Risk-per-trade is the one parameter you should nail down before you start, never adjust mid-run, and revisit only between strategies, never inside one.

The best traders aren't the ones who avoid bad months. The math says nobody does. The best traders are the ones who have a written, repeatable answer for what to do when one starts. The four questions in this post are that answer for most of them. Run them honestly. Most of the time the verdict will be "this is variance, hold steady" — and that's the right answer, even though it's the boring one.

FAQ