Probability and expected value are the two mathematical concepts that determine whether a trading strategy makes or loses money over time. This article defines probability in the context of uncertain trade outcomes, explains how to calculate expected value, demonstrates why win rate alone is misleading, shows how to estimate these values from your own trading data, and addresses the role of variance in creating losing streaks even within profitable systems.
What Is Probability in the Context of Trading
Probability in trading is the likelihood, expressed as a percentage or decimal, that a specific outcome will occur on a given trade. A strategy with a 60% win rate has a 0.60 probability of producing a winning trade and a 0.40 probability of producing a losing trade on any single execution.
Trading probabilities are empirical, not theoretical. A coin flip has a known theoretical probability of 50% heads. A trading strategy’s probability must be estimated from historical data — past trade results, backtests, or both. This estimation is always approximate and subject to change as market conditions evolve.
The critical distinction is between the probability of a single trade and the probability of an aggregate result over many trades. No trader can predict whether the next trade will win or lose. But a well-tested strategy can predict the distribution of outcomes across 500 or 1,000 trades — the same principle that allows casinos to profit despite losing on individual bets.
Why Individual Trade Outcomes Are Uncertain but Aggregate Results Are Predictable
Individual trade outcomes are uncertain because markets are influenced by an effectively infinite number of variables — economic data, institutional order flow, geopolitical events, algorithmic activity, and human emotion — that no model can fully capture. Even a strategy with a strong statistical edge faces randomness on every execution.
The law of large numbers explains why aggregate results become predictable. A strategy with a true 55% win rate might show 40% or 70% wins over 20 trades. Over 500 trades, the observed rate will almost certainly fall between 52% and 58%.
This reality has a direct practical implication: traders must think in samples, not individual trades. Evaluating a strategy based on the last 10 trades is statistically meaningless. Evaluating the last 200 trades, categorized by market regime, provides actionable information. The shift from “did this trade work?” to “is this strategy working across a meaningful sample?” is a defining transition in quantitative analysis.
What Is Expected Value (EV) and How to Calculate It
Expected value is the average amount a trader expects to gain or lose per trade over a large number of repetitions, calculated by multiplying each possible outcome by its probability and summing the results. The formula is: EV = (Win Rate x Average Win) – (Loss Rate x Average Loss). A positive expected value means the strategy is profitable over time. A negative expected value means it loses money regardless of short-term results.
| Scenario | Win Rate | Avg Win | Loss Rate | Avg Loss | Expected Value |
|---|---|---|---|---|---|
| High win rate, small gains, large losses | 70% | $100 | 30% | $300 | ($20) per trade |
| Moderate win rate, balanced risk/reward | 50% | $200 | 50% | $150 | +$25 per trade |
| Low win rate, large gains, small losses | 35% | $500 | 65% | $100 | +$110 per trade |
The table reveals a counterintuitive but essential truth: the first scenario, despite winning 70% of the time, loses money. The third scenario, despite winning only 35% of the time, is the most profitable per trade. Expected value, not win rate, determines profitability.
Calculating EV requires four inputs: the probability of winning (estimated from historical trades or backtests), the average size of a winning trade, the probability of losing (1 minus the win rate), and the average size of a losing trade. These inputs must come from a large enough sample to be statistically reliable — a minimum of 100 trades for a rough estimate and 300 or more for reasonable confidence.
A High Win Rate Does Not Guarantee Profitability
A high win rate does not guarantee profitability because win rate measures only the frequency of winning trades, ignoring their magnitude relative to losses. A trader who wins on 80% of trades but averages $50 per win and $250 per loss has a negative expected value: (0.80 x $50) – (0.20 x $250) = $40 – $50 = -$10 per trade.
This pattern is common among traders who take profits quickly but hold losing trades hoping for a recovery. The equity curve shows a slow, steady rise punctuated by sharp drops that erase weeks or months of progress.
Many retail traders are drawn to high-win-rate strategies because frequent wins feel good. This preference is an emotional bias, not a rational one. A strategy’s value is determined entirely by its expected value, and EV depends on the interaction between win rate and win/loss size ratio. Professional traders and quantitative analysts evaluate both dimensions simultaneously.
How to Achieve Positive Expected Value Even with a Low Win Rate
Positive expected value with a low win rate is achieved by ensuring that the average winning trade is sufficiently larger than the average losing trade to compensate for the lower frequency of wins. Trend-following strategies are the classic example. They typically win on only 30-45% of trades but capture outsized gains on the trades that work, because they hold winners through extended trends.
The required ratio can be calculated directly. If your win rate is 35%, your average win must be at least 1.86 times your average loss to break even: W/L = 0.65/0.35 = 1.86. To generate meaningful positive EV, the actual ratio must exceed this threshold substantially.
Practical techniques include trailing stops that let winners run while capping losses, ATR-based position sizing for consistent risk per trade, and filtering entries to require a minimum reward-to-risk multiple — commonly 2:1 or 3:1.
The psychological challenge is significant. Losing on 60-70% of trades means enduring extended losing streaks. Understanding probability theory and trusting a tested positive-EV system is essential — the math works, but only if the trader executes consistently through the losing periods.
How to Estimate the Probability and Expected Value of Your Trading Setups
Estimating probability and expected value for your own setups requires a systematic review of historical trade data, either from your live trading journal or from backtested results. The process starts with categorizing trades by setup type, then calculating the win rate and average win/loss for each category independently.
Step one: define your setups precisely. “Long entry on a daily close above the 20-day high with volume exceeding the 20-day average, exited on a close below the 10-day low” is specific enough to categorize consistently. Every trade should be tagged with its setup type.
Step two: calculate the basic statistics for each setup — total trades, wins, losses, win rate, average win (in dollars and R-multiples), average loss, largest win, largest loss, and expected value per trade.
Step three: segment the data by market regime. A breakout strategy might have a 55% win rate in trending markets and 30% in range-bound markets. The combined 42% obscures this critical distinction. Regime-specific probability estimates are more accurate for deciding when to deploy each strategy.
The Minimum Sample Size for Reliable Probability Estimates
The minimum sample size for a statistically meaningful probability estimate is approximately 30 trades for a very rough directional estimate and 100 trades for reasonable confidence in the win rate. For precise expected value calculations that account for the variability of win and loss sizes, 200-400 trades provide substantially more reliable estimates.
Small samples are dominated by random variation. A strategy with a true 50% win rate has a 17% chance of showing 60% or higher over just 30 trades. Over 200 trades, that probability drops below 1%.
A practical rule: do not make major strategy changes based on fewer than 50 trades. Do not commit significant capital based on fewer than 100 trades. And always recalculate estimates as new data accumulates, since market conditions shift and historical estimates degrade over time.
Using Expected Value to Compare and Select Trading Strategies
Expected value per trade is the primary metric for comparing strategies because it directly measures the average profit generated per execution. However, raw EV alone is insufficient — it must be evaluated alongside trade frequency, capital requirements, and risk-adjusted return.
A strategy with an EV of $50 per trade that generates 200 trades per year produces $10,000 annually. A strategy with an EV of $200 per trade that generates only 20 trades per year produces $4,000. The lower-EV strategy is more valuable because of higher frequency. Expected value multiplied by trade frequency gives the expected annual return — a more complete comparison metric.
Capital efficiency also matters. A strategy that ties up $100,000 to generate $10,000 (10%) is less efficient than one that ties up $25,000 to generate $5,000 (20%), even though the first produces more absolute dollars. The Sharpe ratio incorporates return and risk into a single metric. When comparing strategies with similar expected values, select the one with the higher Sharpe ratio — it achieves similar returns with smaller drawdowns.
The Role of Variance and Drawdown — Why Positive EV Strategies Still Lose Money Short-Term
Variance is the statistical measure of how widely individual trade results deviate from the expected value, and it explains why a strategy that is mathematically profitable can experience extended periods of losses. A positive expected value guarantees profitability only over a theoretically infinite number of trades. In the short and medium term, variance creates drawdowns that test the trader’s capital and psychology.
A strategy with a 55% win rate and a 1.5:1 reward-to-risk ratio has a positive expected value. Yet this same strategy will, with mathematical certainty, experience losing streaks of 5 or more consecutive trades, drawdowns of 15-25% of the account, and extended periods where the equity curve is flat or declining. These are not signs that the strategy is broken — they are the statistical cost of operating in a probabilistic environment.
Understanding Losing Streaks as a Statistical Certainty
Losing streaks are a statistical certainty for any strategy with a win rate below 100%. The probability of a consecutive losing streak of length N equals (1 – win rate)^N. For a strategy with a 55% win rate, the probability of 5 consecutive losses is 0.45^5 = 1.8%. Over 500 trades, this near-certainty means multiple 5-trade losing streaks will occur.
The probability calculations become sobering at longer streak lengths. The same 55% win-rate strategy has a 0.37% chance of 7 consecutive losses on any given sequence. Over 1,000 trades, the expected number of 7-trade losing streaks is approximately 3-4. These streaks feel catastrophic in real time even though they are mathematically normal.
Traders who do not understand these probabilities often abandon profitable strategies during statistically expected drawdowns. They interpret a normal losing streak as evidence that the strategy has stopped working. This abandonment behavior — exiting a positive-EV system during a temporary but inevitable variance event — is one of the most expensive errors in trading. Understanding probability prevents this mistake.
How Position Sizing Protects Against Variance
Position sizing is the primary defense against variance because it determines how much account equity is at risk on each trade, and therefore how severe a losing streak’s impact will be on the account. A trader risking 1% of equity per trade can survive a 10-trade losing streak with a 9.6% account drawdown. A trader risking 5% per trade faces a 40.1% drawdown from the same streak.
The Kelly Criterion provides a mathematically optimal position size based on win rate and win/loss ratio: Kelly % = W – [(1-W) / R], where W is the win rate and R is the ratio of average win to average loss. For a strategy with a 55% win rate and a 1.5:1 win/loss ratio, Kelly suggests risking 25% of capital per trade. In practice, most traders use a fraction of the Kelly amount — typically one-quarter to one-half — because full Kelly sizing produces drawdowns that are psychologically intolerable.
Conservative position sizing ensures that variance does not destroy the account before the law of large numbers has time to work. The expected value is a long-run average. Position sizing is the mechanism that guarantees the trader survives long enough to reach the long run. Combined with proper risk management, it transforms a mathematically profitable strategy into a practically survivable one.
Probability Theory Concepts That Enhance Trading Analysis
Probability theory offers several concepts beyond basic win rates that sharpen trading analysis. Bayesian updating adjusts probability estimates as new information arrives — if the options market prices a large implied move before earnings and short interest spikes, Bayesian reasoning updates the decline probability to reflect this new evidence.
Conditional probability — the probability of event A given that event B has occurred — is directly applicable to trading. The probability that a stock rises on any given day might be 52%. The probability that it rises on a day when the broader market is up more than 1% might be 68%. Conditioning on relevant variables produces more accurate estimates than unconditional averages.
The gambler’s fallacy — the belief that a losing streak makes a win more likely on the next trade — has no basis in probability theory when trades are independent events. Each trade’s probability is determined by the strategy’s edge and current market conditions, not by the sequence of prior results.
Monte Carlo simulation generates thousands of hypothetical equity curves by randomly resampling actual trade results. The output shows the range of possible outcomes, including worst-case drawdown scenarios, providing a probabilistic risk assessment that a single backtest cannot.
How to Apply Expected Value Analysis to Your Current Trading Journal
Applying expected value analysis to your trading journal begins with exporting your trade history into a spreadsheet with columns for setup type, entry date, exit date, position size, P&L in dollars, and P&L in R-multiples (where 1R equals your initial risk on the trade).
Calculate EV separately for each setup type. If you trade breakouts, pullbacks, and reversals, each has its own probability profile. Combining them into a single EV calculation obscures which setups carry your edge and which drag on performance.
Next, calculate EV by market regime. Tag each trade with whether it occurred during an uptrend, downtrend, or sideways market. The same setup in different regimes is effectively a different strategy with different probabilities.
Finally, track how your EV estimates change over time using a rolling calculation across your last 100 or 200 trades. Declining EV warns that market conditions may be shifting away from your strategy’s strengths. This rolling analysis, combined with regular backtesting validation, creates the ongoing feedback loop that sustains a quantitative trading approach over the long term.