The chance of a specific run of losses is your loss rate raised to the number of trades in the run. The chance of seeing that run at least once over many trades is far higher. At an 80% win rate, three losses in a row in one given sequence is 0.8%. Across 100 trades, a three-loss run is more likely than not. Sonic AI, the gold (XAUUSD) auto-trading strategy we run, reports an 80% win rate with 18 consecutive verified winning months, so this math applies to it as to any strategy: losing streaks are part of how win rates work, and sizing for them is what keeps an account in the game.
TL;DR
- For one specified block of n trades, the chance that every trade loses is q^n, where q is the loss rate (1 minus the win rate).
- Over a whole trading period, the chance of at least one streak is much larger, because every overlapping window is another chance for a run.
- The expected longest streak grows with the logarithm of the trade count: log(N) ÷ log(1 ÷ q).
- A streak does not make the next win "due". Each trade keeps the strategy's baseline win rate if trades are independent.
- Plan the account around a high-percentile streak, not the average one.
Table of Contents
- What a losing streak is
- The basic formula: one specified block
- The question that matters: at least one streak over N trades
- How win rate and trade count change the picture
- Estimating the longest streak you should expect
- Do streaks predict a win?
- Turning streak odds into a risk plan
- Where Sonic AI fits
- Our perspective
- FAQ
- Sources
What a losing streak is
A losing streak is a sequence of consecutive losing trades with no winner in between. It measures consecutive outcomes, while drawdown measures the fall in equity from a previous peak. The two can diverge. A long streak of small losses can leave a shallow drawdown, and a few large losses can produce a deep one.
That difference matters for anyone copying a strategy. The streak tells you how many losses came in a row. The drawdown tells you how much money you lost. You need both numbers to judge your risk. For the drawdown side, see our explainer on max drawdown for gold copy traders.
The basic formula: one specified block
Under the standard independent-trade model, let p be the win rate and q = 1 − p the loss rate. The probability that one specified block of n trades is all losses is q^n. A worked example: with a 40% loss rate, five losses in a specified block is 0.40^5 = 0.01024, about 1 in 98.
Apply it at a 70% win rate. The loss rate is 30%, so the sequence probabilities are 30% for one loss, 9% for two, 2.7% for three, 0.8% for four and 0.2% for five.
Here is the same calculation at an 80% win rate, where q = 0.2. This is plain arithmetic on a hypothetical rate, not a forecast for any account.
| Losses in a row | Formula | Chance for one specified block |
|---|---|---|
| 2 | 0.2^2 | 4% |
| 3 | 0.2^3 | 0.8% |
| 4 | 0.2^4 | 0.16% |
| 5 | 0.2^5 | 0.032% |
These numbers look small. They answer a narrow question: what is the chance that the next n trades, starting now, all lose? Most readers want a different question answered.
The question that matters: at least one streak over N trades
The chance of at least one streak of k or more losses in N trades is much higher than the chance of one specific run. One calculator gives a clear example: at a 40% win rate, the next 8 trades all losing is about 1.7%, but over 100 trades the chance of at least one streak of 8 or more is about 49.0%.
The reason is overlap. Trades 1 to 3 are one window, trades 2 to 4 are another, and so on. A finite-horizon calculator shows the effect: with a 60% win rate and 10 trades, at least one three-loss run has a 31.4% chance, against 6.4% for one specified three-trade block.
A quick approximation you can do by hand
For a streak of n losses within N trades, a commonly used approximation is 1 − (1 − q^n)^(N − n + 1). It is an approximation, and the exact result needs a recurrence. A university handout gives a similar one: when q^r is small, the chance of at least one losing run of length r in n trials is about [1 + (n − r)p] × q^r.
Using the first formula with an 80% win rate and 100 trades, we get these rough figures:
| Streak length | Approximate chance of at least one in 100 trades |
|---|---|
| 3 or more | about 54% |
| 4 or more | about 14% |
| 5 or more | about 3% |
So a three-loss run inside 100 trades is ordinary at this win rate, a four-loss run is possible, and a five-loss run is uncommon. Compare that with what published tables show for lower win rates.
How win rate and trade count change the picture
A lower win rate and more trades both lengthen the expected worst run. Published figures for 100 trades:
| Win rate | At least one streak | Chance | Source |
|---|---|---|---|
| 70% | 4 or more losses | 43.3% | Clear Tape |
| 70% | 6 or more losses | 4.8% | Clear Tape |
| 60% | 5 or more losses | 62.8% | BacktestBase |
| 50% | 5 or more losses | 81.0% | Forexero |
| 50% | 10 or more losses | 4.4% | Forexero |
A 50-trade exact table gives the same direction over a shorter horizon: at least one five-loss run has a 7.7% chance at a 70% win rate, 25.6% at 60%, 55.2% at 50% and 83.5% at 40%.
An important point for any high win rate: the higher it is, the shorter and rarer the streaks. That is the whole reason win rate is one of the first numbers traders check. For how to read an 80% figure, see our guide to what an 80% win rate means and how to check it.

Estimating the longest streak you should expect
A useful rough estimate for the longest losing run in N trades is log(N) ÷ log(1 ÷ q). The same source applies it to 100 trades and gets about nine losses at a 40% win rate and about five at 60%. Over longer histories, at a 50% win rate across 1,000 trades the expected longest run is about 10, at 40% about 13 and at 35% about 15.
With q = 0.2 and 100 trades, the formula gives ln(100) ÷ ln(5), about 2.9 losses. Over 1,000 trades it gives about 4.3. The number grows slowly, but it does grow. Your worst run so far is a floor, not a ceiling, because expected streak length rises with the log of the trade count. A longer record will eventually contain a longer run.
Do streaks predict a win?
No. If trades are independent, a losing streak does not change the next outcome. After five losses, the probability of winning the sixth is unchanged, and expecting a win to be due is the gambler's fallacy. Another guide puts it the same way: each trade outcome is statistically independent.
Two cautions apply in live markets. First, independence is a model. News, volatility spikes, regime changes and correlated positions can cluster losses and make real streaks longer than the baseline. That is why the simple formula is a starting point and historical drawdown analysis is its companion. Second, a streak is a prompt to review, not a verdict. A sound review separates strategy problems, execution problems and market-condition problems instead of reacting to the streak length alone.
Turning streak odds into a risk plan
The formulas are most useful when they set position size. The sequence is simple:
- Take a win rate from a long record, not an aspirational number.
- Estimate a high-percentile streak for your trade count, using the log formula, a table or a simulation.
- Multiply the streak by your risk per trade to see the drawdown it implies.
- Resize until that drawdown is one you can sit through.
One guide recommends using the maximum streak from a broad record or simulation, and dividing tolerated drawdown by the planning streak as a conservative sizing check. Its example: an eight-loss streak at the 95th percentile with 5% risk per trade implies a 34% drawdown.
Risk per trade changes what a streak costs. After nine straight losses, an account keeps 91.4% of its capital at 1% risk per trade, 83.4% at 2% and 63.0% at 5%. Recovery math then matters too: a loss needs a larger gain to undo, as covered in our piece on why a 50% loss takes a 100% gain to recover.
One habit belongs on the do-not list. Raising the stake after each loss to win it back can exhaust an account when a streak continues. Investopedia's explanation of the Martingale method shows a trader with $10 unable to keep doubling after successive losses. Fixed fractional risk avoids this. For the broader math of survival, read our note on risk of ruin in trading, and for a hard stop on bad days see setting and enforcing a daily loss limit.
Where Sonic AI fits
Sonic AI is an auto-trading strategy for gold (XAUUSD), run by an experienced team and executed through COPYX, which copies trades to your account automatically. Our reported figures are an 80% win rate, less than 1% drawdown and 18 consecutive winning months, verified on MyFXBook. You can read the live results on the live performance page.
For a reader who has just worked through the tables above, this is the practical point. Streak math is easy to do on paper and hard to carry out by hand: it asks you to take every signal, keep risk constant and not react after a loss. With copy trading through COPYX, entries and exits are carried out automatically, so that routine does not depend on your mood after the last trade.

If you want gold exposure without trading manually, start at Sonic AI. Accounts are also available as a 12X or 24X amplified account, and how the performance fee works is published on its own page. For the profit-sharing side, see our note on the hidden split in profit sharing trading.
Our perspective
We built Sonic AI around one instrument, gold, and one idea: a strategy should be judged on a long record, not on its last few trades.
That is why we report the verified record on MyFXBook, with an 80% win rate and 18 consecutive winning months. A run of 18 winning months is a statement about a long sample, and long samples are what streak math rewards. The sample is large enough to say something, and the numbers can be checked by anyone.
We also think the right response to streak math is structure, not hope. Keep risk per trade small, size the account for a streak longer than any you have seen, and let the system execute the same way every time. Our strategy aims at low drawdown for this reason, and our reported figure is under 1%.
If you want to compare how win rate and payoff fit together, our posts on win rate versus expectancy and low drawdown strategy walk through it.
Paulo
FAQ
How likely are 3 losses in a row with a 70% win rate?
For one specified three-trade sequence it is 0.30^3, or 2.7%. Over 100 trades, one published estimate puts the chance of at least one three-loss streak at 93%. The first number is for a fixed block, the second for a whole trading period.
How do I calculate the probability of a losing streak?
Subtract your win rate from 1 to get q. For one block of n trades, compute q^n. For at least one streak in N trades, use 1 − (1 − q^n)^(N − n + 1) as an approximation, or an exact recurrence calculator. Use a win rate from a long record.
Does a losing streak mean the strategy is broken?
Not by itself. A probabilistic system can produce clusters of losses even when its long-term expectancy stays positive, so a streak is evidence that deserves investigation, not a diagnosis. Check rule adherence, risk size and market conditions, and use a large sample to judge.
Is a win due after several losses?
No. If trades are independent, the chance of the next win stays at the strategy's baseline win rate. Expecting a win to balance a streak is the gambler's fallacy.
Sources
- Losing Streak, Pillinger Works
- Consecutive Loss Streak Probability, Pomegra
- Edgewonk's Math Cheat Sheet, NewTraderU
- Losing Streak Probability Calculator & Table, Clear Tape
- Risk of Ruin and Losing Streak Math, TradeOlogy Academy
- Winning and Losing Streaks, NYU Stern handout
- Trading Losing Streak Calculator, BacktestBase
- Losing Streak Calculator, Forexero
- Gestión de Riesgos en el Trading
- Losing streaks lesson, Signal Mixer
- Losing Streaks glossary, TapeSheet
- Risk: streaks, expectancy, journal, EVA Terminal
- How to Handle a Losing Streak in Trading, TradeZella
- Consecutive Loss Probability Calculator, QuantStock
- How the Martingale Strategy Impacts Forex Trading, Investopedia
- Losing streaks in trading: the math behind what's normal, Edgeful
