Expectancy

Expectancy denotes the expected outcome of a trade within a specified probability model. Expected value weights possible outcomes by their probabilities. By contrast, the sum of recorded outcomes divided by the number of trades is a sample mean. It describes that dataset exactly, but estimates future expected value only under additional assumptions. State whether the calculation uses a model, actual executions or a fictional illustration. These concepts must not be confused merely because their calculations look similar. [1] [5]

CME combines win frequency with average profit and subtracts loss frequency multiplied by the average loss magnitude. For a completed dataset, this is another expression of its mean if the units match and all outcomes are covered. The loss in the subtracted term is a positive magnitude, not another negative number. Trades with zero net outcome remain in the total count even though they add zero to the sum. If a category is absent, use the sum of all outcomes directly rather than the undefined average of an empty group. [2]

Partial entries and exits can create multiple fills for a single trade. Coinbase describes partial execution and retention or cancellation of the unfilled remainder. The journal must therefore define in advance how it groups fills and allocates costs. An open remainder cannot be presented as a completed win, and a rejected order is not automatically a zero-result trade. Record unexecuted signals and reasons for exclusion separately. Without this definition, two apparently identical averages may use different denominators. [4]

Subtract fees from each trade's outcome before classifying it as a win, loss or zero result. A gross win can change category after costs. Coinbase distinguishes maker and taker fees by fill; the illustration's fixed cost is therefore not a fee schedule. Include financing where applicable. Do not subtract spread and slippage twice if they are already included in execution prices. A monetary outcome cannot be compared directly with a percentage or otherwise normalized result without explaining the conversion. [3]

Reordering the same completed outcomes changes neither their sum nor their arithmetic mean, but changes the path of cumulative results. The average alone therefore does not specify drawdown from a previous peak, the length of a losing streak or the required cash reserve. This is a mathematical limitation of the measure, illustrated by our example below. In risk management, CME distinguishes an individual trade from simultaneous positions and whole-account exposure. With changing position sizes or reinvestment, the average of fixed-size trades cannot simply be transferred to account growth. [7]

NIST explains that the sample mean varies between samples and an interval estimate expresses uncertainty. In the stated statistical model, both sample size and data dispersion affect it. The number of trades alone is not universal confirmation that a trading strategy is reliable. Before applying a particular interval, its assumptions must be checked, including the treatment of dependent observations. We derive no interval for a real strategy from our fictional dataset; its calculated average is not a probability of future profit. [5]

For hypothetical performance, NFA highlights hindsight, imperfect representation of liquidity and slippage, and differences from actually bearing losses. An average of selected successful trades is therefore insufficient. Record the rules, the entire observation period, costs and the distinction between proposed and executed trades. Changing rules after seeing results changes the meaning of the number subsequently reported. Hypothetical and actual outcomes must remain distinguishable; a numerical illustration is not evidence of a trading edge. [6]

Comparing new executions with an older estimate requires the same trade definition, units and cost scope. A change in position size or execution method can change the average without representing the same strategy under the same conditions. Records should therefore connect outcomes to fills and risk-management rules. A historical average promises neither the next win nor a substitute for reviewing total exposure. Expectancy is intended to sharpen questions about outcomes and their assumptions, not turn one number into certainty. [4] [7]