Trading the Distribution, Not Obsessing Over Short-Term Outcomes
You will never know what the market is going to do. Nobody does. The analyst on television doesn't know, and the fund manager speaking confidently doesn't either. Neither do you, however good your chart looks when you take your entry. If you can understand this one idea, you're already ahead of most of the people you're trading against. Most of the people on the other side of your trades will spend all their time chasing the one thing that can't be done, which is predicting the future.
This is Chapter 1 from my book: How to Trade Like a Mathematician: Quantitative Trade Management Strategies, Risk Management, and Mathematical Models to Master the Stock Market.
A person who wants to trade like a mathematician takes a different approach entirely. They assign probabilities to a range of futures, then structure their decisions so that the math works out in their favor over many repetitions. That's the complete shift this chapter asks of you. Instead of asking, "What will this stock do next?" Ask a better question: "What does the distribution of outcomes look like for this trade, and does the math still favor me if I take this exact trade a hundred times?"
The Casino Owner's Mindset
Consider how a casino runs its floor. When a gambler sits down at the roulette table, the casino does not know whether that particular spin will win-or-lose money for the house. The gambler might hit their number and walk out with thirty-five times their bet. The casino shrugs. On an American roulette wheel, there are 38 pockets: the numbers 1 through 36, plus 0 and 00. A straight-up bet on a single number pays 35 to 1. The player therefore has a 1-in-38 probability of winning and a 37-in-38 probability of losing. For every $1 wagered, the casino’s expected profit is:
Expected casino profit = (37/38 × $1) − (1/38 × $35)
This equals approximately $0.0526, or 5.26 cents per dollar wagered. In percentage terms, the American roulette wheel gives the casino a 5.26% house edge. The casino won’t earn exactly 5.26% over every short period, but as the number of wagers increases, its average profit typically converges toward that mathematical expectation.
Notice what the casino never bothers to do. It never forecasts a spin. It owns a small edge and lets the law of large numbers grind that edge into money, night after night, table after table. On any evening, the house can lose. Over ten thousand spins, the outcome is close to certain. That is the model to follow: trade the way the house operates, and evaluate every position the way the casino evaluates a wheel, through expected value and never through opinion.
Expected Value: The Only Question That Matters
Expected value, or EV, is the mathematical heart of probabilistic trading. The formula:
EV = (Probability of Winning × Average Win) minus (Probability of Losing × Average Loss)
A positive number means the trade is worth taking repeatedly. A negative number means the trade is a slow donation to someone else's account, and it stays a donation no matter how strongly you feel about the setup or how many people on social media agree with you. That is why EV is the only question that matters.
Time for real numbers. Suppose you tracked a specific setup in your trading journal, such as a pullback to the 20-day moving average in a stock trending upward. Over 100 occurrences, 45 trades were winners and 55 were losers. Your winning trades averaged a gain of 2.5R, where R represents the amount risked per trade, while your losing trades averaged a loss of 1R because you consistently honored your stop-loss. Your estimated expectancy per trade would be:
EV = (0.45 × 2.5R) − (0.55 × 1R)
EV = 1.125R − 0.55R = +0.575R
This means the setup historically earned an average of 0.575R per trade. If you risked $100 per trade, that would equal an estimated average profit of $57.50 per trade. This does not mean every trade will earn 0.575R. It means that, across a sufficiently large sample of similar trades, the average result would be expected to approach that amount if the setup’s win rate and payoff ratio remain stable.
Across a sufficiently large number of comparable trades, this setup has historically produced an average gain of 0.575R per trade, even though it loses more often than it wins. Read that again. It challenges one of the most damaging beliefs in trading: that a profitable trader must be right most of the time.
A 45 percent win rate combined with an average winner of 2.5R and an average loser of 1R creates strong positive expectancy before commissions, slippage, and other trading costs. Meanwhile, a strategy with a 70 percent win rate can still lose money if its average loss is three times its average win.
EV = (0.70 × 1R) − (0.30 × 3R)
EV = 0.70R − 0.90R = −0.20R per trade
That trader may feel successful on most trading days while steadily losing capital over a large sample of trades. Whether the trader eventually goes broke depends on position sizing, available capital, the sequence of returns, and how long the negative-expectancy strategy is traded.
This is also why unsupported opinions are dangerous. An opinion says, “Apple is going up.” A probability-based trading statement says, “Based on 100 historical occurrences of this setup, approximately 45 percent reached the 2.5R target before hitting the 1R stop. If those results remain representative and execution costs do not eliminate the edge, the setup has an estimated expectancy of +0.575R per trade.”
That statement does not guarantee the outcome of the next trade. It provides a testable hypothesis with numbers that can be used for position sizing, risk management, and ongoing performance review. The shift from certainty and prediction to probability and expectancy is the whole point.
Individual Trades Mean Nothing; Samples Mean Everything
Now for the place where most traders quietly sabotage themselves. They judge their system trade by trade. They take a perfectly good positive expectancy trade, it loses, and they conclude the setup is broken. Or they take a terrible negative expectancy trade, it wins, and they conclude they've discovered a new edge. Both conclusions are mathematically illiterate, and I say that with love, because I want you to feel a small sting every time you catch yourself doing it.
A single outcome tells you very little about the quality of the decision that produced it. Flip a fair coin ten times, and there is about a 17 percent chance of getting seven or more heads. Seven heads in ten flips would not be sound evidence that the coin was biased because ten observations are too few to separate skill, edge, or bias from ordinary randomness.
Traders often make the same mistake when evaluating a system. They abandon a strategy after a short losing streak, even when that streak is consistent with the system’s expected win rate. A strategy with a 45 percent win rate loses 55 percent of its trades. The probability that any sequence of five trades will contain five consecutive losses is:
0.55⁵ ≈ 5.03 percent
That does not mean a five-trade losing streak will occur exactly once every twenty trades. Streaks can overlap, cluster, or remain absent for long periods. However, across hundreds of trades, one or more five-loss streaks should not be surprising. A single losing streak provides limited evidence that the strategy’s underlying edge has disappeared. Random sequences naturally produce clusters of wins and losses.
The practical defense is to establish evaluation rules before trading the system. Do not judge a setup from a handful of outcomes. Thirty trades may provide an initial sign, but 50 to 100 trades offer a more useful sample, and some strategies require far more because of their win rate, payoff distribution, trade frequency, and sensitivity to changing market conditions.
Record every valid occurrence in a trading journal, including the entry, stop, target, trailing stop, market conditions, execution costs, and the result in R multiples. Then compare the observed win rate, average win, average loss, expectancy, and drawdown with the strategy’s historical assumptions.
Even a large sample does not prove that an edge will continue forever. It gives you a stronger statistical basis for deciding whether recent results are consistent with normal variation or whether the strategy may have materially changed. The goal is not to replace judgment entirely, but to ensure judgment is grounded in evidence rather than emotion, isolated outcomes, or short-term noise.
Watch how this plays out under pressure. Say you trade a breakout strategy and you're 12 trades in with a record of 4 wins and 8 losses. Your emotional brain screams that the strategy is failing. Your mathematical brain runs a check instead. It asks how unusual 4 wins in 12 trades would be if the true win rate really were the expected 45 percent. The binomial distribution answers: getting 4 or fewer wins in 12 trials at a 45 percent win rate happens roughly a quarter of the time. Ordinary. It's the equivalent of rolling a die twice without seeing a six, and you'd never call that die broken. So you keep executing your strategy because 12 trades is far too small a sample to override a tested edge.
Position Sizing Is Where Probability Becomes Survival
Thinking in probabilities forces a hard conclusion about position sizing, too. Since any single trade can lose, no single trade can ever be allowed to matter very much. A setup that wins 45 percent of the time will hand you losing streaks of 5, 6, or even 8 trades with mathematical certainty over a long career. So your position sizing has to guarantee you survive the worst losing streak probability will realistically deal you eventually, and risk management has to guarantee your account survival in advance.
Run the numbers on risking 10 percent of your current account equity per trade. After eight consecutive full-risk losses, the account would retain:
0.90⁸ ≈ 43.05 percent
That is a drawdown of approximately 56.95 percent. Recovering from that loss would require a gain of about 132 percent just to return to the starting balance.
Now apply the same eight-trade losing streak while risking 1 percent of current equity per trade:
0.99⁸ ≈ 92.27 percent
The account would experience a drawdown of approximately 7.73 percent. Recovering from that loss would require a gain of about 8.38 percent.
For a strategy with a 45 percent win rate, eight consecutive losses are unlikely in any eight-trade sequence, but they become increasingly plausible as the number of trades grows. A probability-based trader therefore sizes positions for losing streaks that can reasonably occur, not for the winning trade expected next.
Many disciplined traders risk only a small fraction of their equity on each position, often around 0.5 percent to 2 percent, although the amount depends on the strategy’s historical drawdowns, volatility, correlation between positions, stop-loss reliability, and the trader’s risk tolerance. The principle is straightforward: position size should allow the account to survive normal losing streaks without suffering a drawdown so deep that recovery becomes mathematically difficult or psychologically unbearable.
Turning Market Views into Probability Estimates
The market won't hand you probabilities the way a roulette wheel does, so you have to build your own estimates, and three sources do most of the work. Your own journal comes first: log every setup, let your realized win rate and average R multiples become your probability inputs, then update them as the sample grows.
Backtesting comes second. Before risking capital on a rule-based setup, test it across a sufficiently large sample covering different market regimes, volatility environments, and trend conditions. The goal is not just to prove that the strategy will work in the future, but to estimate its historical win rate, payoff ratio, expectancy, drawdowns, and sensitivity to changing market conditions.
The options market provides another useful source of information. Option prices can be used to derive model-based, risk-neutral probability estimates from implied volatility, strike prices, expiration dates, and the shape of the volatility skew. These estimates are not forecasts of the stock’s true future probabilities, but they show how the options market is pricing uncertainty at that moment.
Suppose a stock trades at $100, and one-month options imply an expected range of approximately $92 to $108. That range is often interpreted as roughly one standard deviation, which corresponds to about 68 percent under a simplified normal-distribution assumption. It does not guarantee that the stock has exactly a 68 percent chance of remaining inside that range, but it provides a useful benchmark for judging whether your price target is consistent with the movement currently priced into the options market.
If your trade requires the stock to reach $120 within one month, you can examine option prices to estimate how unlikely the market considers that outcome. If a model assigns it approximately a 5 percent probability, a binary wager on that event would require a net payoff of about 19 to 1 merely to break even before fees:
Break-even payoff = (1 − 0.05) ÷ 0.05 = 19 to 1
A conventional stock trade is more complicated because its expectancy also depends on the stop-loss level, the path of price movement, partial exits, slippage, and whether the target must merely be touched or held through expiration. Still, the comparison is valuable. If your target lies far outside the option-implied distribution and the potential reward does not compensate for the low estimated probability, the trade may not offer favorable expectancy.
Backtesting, observed market data, and option-implied distributions do not eliminate uncertainty. They convert vague predictions into measurable assumptions that can be tested, sized, and reviewed before capital is placed at risk.
One honest caution belongs here. Your probability estimates will always be imperfect. Markets shift, edges decay, and past frequencies are estimates, never guarantees. Perfect inputs are out of reach, and that's fine, because approximately correct probabilities paired with conservative sizing will work for you over time if you stay consistent with on strategy. When you're somewhat wrong about the size of your edge under that arrangement, the error dents your results instead of wrecking your account. A person thinking like a mathematician builds humility directly into their position size.
The Daily Practice
The routine itself is short. Before every trade, write three numbers: your estimated win probability, your reward measured in R, and your risk of 1R. Compute the expected value. If the number cannot come out clearly positive, you skip the trade, no matter what your impulses are shouting. After every trade, record the outcome without judgment, because one outcome is a single data point landing somewhere inside a distribution you already accepted when you clicked the buy button. Once a month, review your samples by setup, recalculate your actual win rates and payoffs, and adjust your estimates. Run yourself like a floor manager walking the casino at closing time, checking that each table is being managed properly to hold its edge.
Something useful happens to your psychology once this becomes a habit. Losses stop registering as failures. A loss taken on a positive expected value trade, at proper size, with the stop honored, counts as a correctly executed trade in every way that matters mathematically. Log it, note the R multiple, and move to the next entry in the queue.
Executing These Principles in Your Trading Plan
Write an expected value requirement into your plan: no trade is taken unless EV = (Win% × Avg Win) minus (Loss% × Avg Loss) comes out positive based on backtested or journaled trading data.
Before each entry, record your three inputs in writing (estimated win probability, reward in R multiples, and risk of 1R), then calculate the EV.
Set a fixed risk per trade of 0.5 to 2 percent of account equity so that mathematically inevitable losing streaks of 5 to 8 trades produce recoverable drawdowns.
Define a minimum sample size rule: no setup gets judged, changed, or abandoned on fewer than 30 recorded trades, preferably 50 to 100.
Journal every trade with entry, stop, target, and R multiple results, tagged by setup, so your probability estimates come from data rather than feelings.
When a losing streak occurs, run the binomial math to check whether the streak is statistically normal for your win rate before changing anything.
Use implied volatility to check trade probabilities: if the market prices your target as a low-probability event, demand a payoff large enough to justify it, or skip the trade.
Schedule a monthly review to recalculate each setup's actual win rate, average win, and average loss, then update your EV inputs accordingly.
Ban prediction language from your process: replace "this will go up" with a written probability and payoff statement for every trade.
You can purchase this book on Amazon worldwide: How to Trade Like a Mathematician: Quantitative Trade Management Strategies, Risk Management, and Mathematical Models to Master the Stock Market.
<<<<<<<<<<<<<<................>>>>>>>>>>>>>>>
#JobsReportRaisesHikeOdds #CPITestsDecemberRateOutlook #BitcoinRebounds$75KAsChipsLead #US Data Center Rules May Constrain AI Expansion #WillAnthropicListOnTime