How Many Losing Trades in a Row Is Normal? The Exact Odds

It depends on your win rate and on how many trades you take, and it is usually more than it feels like. At a 50% win rate, 5 losses in a row show up somewhere in 81% of samples of 100 trades. The expected longest losing run over those 100 trades is about 6; at a 40% win rate it is about 8.

A streak that long is not evidence that a strategy is broken. It is what the arithmetic of a coin flip produces. Below are the exact losing streak probabilities for win rates from 40% to 60%, computed rather than estimated. Then comes what they mean for how much you risk per trade.

The short answer by win rate

A losing streak is a run of consecutive losing trades with no winner in between. Two numbers describe how long a normal one gets. The first is the expected longest losing run: the average length of the longest streak across many samples of the same size.

The second is the streak to plan for: the shortest run with no more than a 5% chance of being beaten in the sample. It is the number to size risk against, because the average streak is exceeded about as often as not.

Read it like this. At a 45% win rate over 200 trades, your longest losing run has at least a 95% chance of being 12 or fewer. A run of 9 or 10 is unremarkable.

For your own win rate and number of trades, the losing streak calculator computes the same exact odds, the run to plan for, and the drawdown it causes at your risk per trade.

The full table: the chance of a losing streak

Each cell is the probability of at least one run of that many losses or more, anywhere in the sample. A run of 10 also counts as a run of 8, so each table includes the longer streaks of the tables after it.

5 or more losses in a row

8 or more losses in a row

10 or more losses in a row

12 or more losses in a row

Three things stand out:

• Five in a row is the norm, not the exception, even with a winning strategy. Over 200 trades it happens 88.0% of the time at a 55% win rate, and 71.4% of the time at 60%.
• Five points of win rate move the odds a lot. Over 200 trades, 10 losses in a row has a 9.0% chance at a 50% win rate and 19.9% at 45%, more than double.
• More trades means longer streaks. The longest run grows roughly with the logarithm of the number of trades. At a 50% win rate, doubling the sample from 100 to 200 trades adds about one loss to the expected longest run.

Why streaks feel abnormal

The intuition traders use is the chance of losing the next few trades. At a 50% win rate, the chance that the next five trades all lose is 0.5 to the fifth power, about 3%. That sounds rare, and it is, for any five trades you pick in advance.

But a sample of 100 trades contains 96 overlapping stretches of five, and a streak can start in any of them. The question that matters is whether it happens somewhere, and over 100 trades the answer is 81%. That gap between “these five trades” and “any five trades” is why a perfectly normal streak feels like a sign that something is wrong.

Many streak calculators use a shortcut that treats each stretch of five as unrelated to its neighbors. Overlapping stretches share trades, so the shortcut overstates the odds. At a 50% win rate over 100 trades, it gives 95.3% for five losses in a row; the exact figure is 81.0%. At a 45% win rate over 200 trades, it gives 38.4% for ten in a row, against an exact 19.9%.

The opposite mistake is just as common: assuming a win is “due” after a run of losses. If your trades are independent, the chance of the next trade winning is the same after five losses as after five wins. A streak tells you nothing about the next trade on its own.

What it means for risk per trade

A losing streak only matters because of what it does to the account. If you risk a fixed percentage of your current balance on each trade, a run of losses costs:

If you risk a fixed amount based on the starting balance instead, the drawdown is the number of losses times the risk. So 12 losses at 1% cost 12%. In a BacktestFlow replay, you can size each trade by a risk percentage of the starting or the current balance. That lets you test both rules on the same strategy.

A strategy with a 45% win rate, traded for 200 trades, should plan for a run of 12 losses. At 2% risk, that one streak is a drawdown of 21.5%, before any of the losses that fall between winners. At 1% it is 11.4%, enough to fail a challenge with a 10% maximum loss limit. At 0.5% it is 5.8%. The strategy is the same each time; the sizing decides whether a normal streak is survivable.

Our guide to backtesting a prop firm challenge uses the same exact calculation in its own losing streak table. At a 45% win rate, 10 or more losses in a row within 200 trades has a 19.9% chance, roughly one in five. On the Pro and Max plans, BacktestFlow’s challenge mode replays history under a profit target and daily and maximum drawdown limits. It shows whether your sizing survives the streaks your strategy actually produces.

Your win rate is an estimate too

Every number above assumes you know your true win rate. You don’t: you know the win rate of your sample, and a small sample is a loose guide. With 100 trades and an observed 55% win rate, the true rate could plausibly be anywhere from about 45% to 64%. We worked through why in how many trades you need to backtest a strategy. You can check the range for your own numbers with the sample size calculator.

That matters here because streaks are very sensitive to the win rate. If your backtest shows 55% over 100 trades, the safer move is to size for the streaks of a 45% strategy. Over 200 trades, that is a planning streak of 12 rather than 9.

When a losing streak is a warning sign

A streak is worth investigating when it is longer than the run to plan for at your win rate and sample size. Even then it is a reason to look, not proof that the edge is gone:

• Check your execution first. Did you take the trades your rules describe, at the size they describe? Streaks often start with one rule broken after a loss, one of the reasons backtest results don’t match live trading. On the Pro and Max plans, BacktestFlow’s Breakdown section splits results by checklist item and gives a discipline score. It shows whether the streak came from trades that broke a rule.
• Check whether conditions changed. A strategy built on trending markets can lose for longer in a range. That is a real change, and it is different from bad luck. On the same plans, rolling expectancy in the Stability section shows whether the edge faded over the same stretch.
• Count the trades before drawing conclusions. Ten losses inside a sample of 500 tells you much less than the same ten inside a sample of 30.

What you should not do is change the strategy in the middle of a streak that the tables call normal. That swaps a tested strategy for an untested one at the moment your judgment is least reliable.

The limits of this model

The tables assume each trade is independent, with the same win rate every time. Real trading breaks that assumption in both directions. Losses can cluster when market conditions don’t suit a strategy for weeks at a time, which makes long streaks more likely than the tables show. Rules such as stopping for the day after two losses can make them less likely.

So use the tables as a baseline, and compare them with your own record from a manual backtest. On the Pro and Max plans, BacktestFlow measures that record for you. The Stability section of its trading statistics shows your current, longest winning and longest losing streak.

Monte Carlo, on the same plans, resamples your own R multiples into 1,000 seeded paths. It reports the longest losing run in its unlucky, middle and lucky scenarios: the 5th, 50th and 95th percentiles. It unlocks at 30 trades with an R. If your real longest streak is much longer than the tables predict, your losses are probably not independent. That is worth understanding before you size risk.

How we computed the numbers

Every figure here is exact, not simulated or estimated. For a given win rate, we track the probability of each possible current losing run after every trade. A win resets the run to zero, and a loss extends it by one. Any path that reaches the target length is counted once and set aside.

After the last trade, the probability set aside is the chance of at least one streak of that length. The expected longest run is the sum of those probabilities across every length.

We ran two cross checks. The expected longest runs agree with the approximation published by Mark Schilling in 1990 to within 0.03 of a trade. A simulation of 200,000 random samples per cell matched the exact values to within 0.2 percentage points in the three cells we tested.

The Python script that generates every table is below, so you can rerun it with any win rate or sample size.

# Exact losing streak probabilities for independent trades with a fixed win rate.
# State: the length of the current losing run, from 0 up to L minus 1.
# A win resets it, a loss extends it, and reaching L is counted once in "hit".
# Run with: python3 losing_streaks.py

def p_streak(n, L, win):
    """P(at least one run of L or more losses in n trades)."""
    loss = 1.0-win
    dist = [0.0] * L
    dist[0] = 1.0
    hit = 0.0
    for _ in range(n):
        new = [0.0] * L
        for k, pk in enumerate(dist):
            new[0] += pk * win
            if k + 1 == L:
                hit += pk * loss
            else:
                new[k + 1] += pk * loss
        dist = new
    return hit

def expected_longest(n, win):
    """E[longest losing run] = sum over L of P(longest run >= L)."""
    total, L = 0.0, 1
    while L <= n:
        p = p_streak(n, L, win)
        if p < 1e-12:
            break
        total += p
        L += 1
    return total

def run_at_95(n, win):
    """Shortest run length with no more than a 5% chance of being beaten."""
    L = 1
    while p_streak(n, L + 1, win) > 0.05:
        L += 1
    return L

def pct(x):
    v = 100 * x
    if round(v, 1) >= 100.0:
        return "over 99.9%"
    if round(v, 1) < 0.1:
        return "under 0.1%"
    return f"{v:.1f}%"

WINS = [0.40, 0.45, 0.50, 0.55, 0.60]

def head(label):
    cols = " | ".join(f"{round(w * 100)}% win rate" for w in WINS)
    return f"| Trades ({label}) | {cols} |\n" + "|---" * (len(WINS) + 1) + "|"

for L in (5, 8, 10, 12):
    print("\n" + head(f"{L}+ losses in a row"))
    for n in (50, 100, 200, 500):
        print(f"| {n} | " + " | ".join(pct(p_streak(n, L, w)) for w in WINS) + " |")

print("\n" + head("expected longest losing run"))
for n in (100, 200, 500):
    print(f"| {n} | " + " | ".join(f"{expected_longest(n, w):.1f}" for w in WINS) + " |")

print("\n" + head("run to plan for, 95%"))
for n in (100, 200, 500):
    print(f"| {n} | " + " | ".join(str(run_at_95(n, w)) for w in WINS) + " |")

print("\n| Losses in a row | 0.5% risk | 1% risk | 2% risk |\n" + "|---" * 4 + "|")
for L in (5, 8, 10, 12, 14):
    cost = [100 * (1.0-(1.0-r) ** L) for r in (0.005, 0.01, 0.02)]
    print(f"| {L} | " + " | ".join(f"{c:.1f}%" for c in cost) + " |")

print("\n70% win rate, 3 or more losses in a row")
for n in (20, 50, 100):
    print(n, pct(p_streak(n, 3, 0.70)))

The short version

Losing streaks are longer than they feel. At a 50% win rate, 5 losses in a row show up in 81% of samples of 100 trades. The expected longest run there is about 6, and at 40% it is about 8. Size your risk for the run to plan for, not the average one. Use a conservative win rate, because yours is an estimate. Treat a streak as a warning only when it is longer than the arithmetic allows.

Nothing here is financial advice. Backtested performance does not guarantee future results. BacktestFlow is not affiliated with any prop firm. Its challenge mode is a simulation on historical candles.