Edge, expectancy, and variance (intro)
A clear intro to a trading edge, expectancy, and variance, and why variance hides whether an edge is even real. Education only, not advice.
Part of the Styles of Trading and the Honest Reality track on Agenticks. About 10 minutes, written for a intermediate reader.
Three words get thrown around constantly in trading, and most people use them loosely: edge, expectancy, and variance. They are not interchangeable. Understanding how they fit together is the difference between thinking you have something that works and actually knowing.
A trading edge is a measurable reason to expect a positive result over many trades, after you account for costs. The key phrase is over many trades. An edge does not promise anything about the next trade, or the next ten. It is a small statistical lean that only shows up across a large sample. If a method only looked good on one chart, that is an example, not an edge.
An edge is an average, not a promise
A real edge tilts the odds slightly in your favor across hundreds of trades. It says nothing about whether the next trade wins. Any single result is mostly noise.
Expectancy is how you put a number on that lean. It is the average amount you would expect to win or lose per trade over the long run. The simple version: take how often you win and how much you win on average, weigh it against how often you lose and how much you lose on average, and combine them. A positive expectancy means that, on average, each trade adds to the account. A negative one means each trade slowly drains it, even if it wins often.
This is why win rate on its own is a trap. A method that wins 80 percent of the time can still bleed money if the 20 percent of losers are huge. A method that wins only 35 percent of the time can be strongly profitable if its winners dwarf its losers. Expectancy captures both halves at once, which is why it is the more honest number to lean on.
A rough picture helps. Imagine a rule that wins about 4 trades out of 10, and on a winning trade it makes roughly twice what it loses on a losing trade. Four wins of two units each is eight units; six losses of one unit each is six units. Across those ten trades the rule nets two units, which works out to about a fifth of a unit gained per trade on average. That per-trade average is the expectancy, and it is positive here even though the rule loses more often than it wins. Flip the sizes around, so the losers are twice the winners, and the same 40 percent win rate turns sharply negative. The numbers are illustrative, not a result from any real market, but they show why you cannot judge a method by win rate alone.
edge sample expectancy win rate
Strategy A wins 80 percent of trades but its rare losers are very large. Strategy B wins 35 percent of trades but its winners are much bigger than its losers. Which statement is correct? Either one could have positive expectancy. Win rate alone does not decide it. Right. Expectancy depends on both how often you win and the size of wins versus losses. A high win rate can still lose money, and a low win rate can still profit.
Now the part that fools almost everyone: variance. Variance is the random swing in results from one trade to the next, and from one stretch of trades to another. Even a method with genuine positive expectancy does not deliver that average smoothly. It arrives in lumps: winning runs, losing runs, flat patches that feel like the edge has vanished.
Variance can be larger than the edge
Over a small number of trades, the random swing can easily be bigger than the underlying edge. So a real edge can show up as a loss, and a coin-flip method can show up as a hot streak.
Here is why that matters. Suppose a method really does have a slight positive expectancy. Run twenty trades and you might see a loss, purely from bad luck in the order the wins and losses landed. Run a method with zero real edge and you might see a beautiful winning streak, purely from good luck. Twenty trades simply is not enough data to tell those two apart. The signal (the real edge) is buried under the noise (variance), and over a short run the noise usually wins the argument.
This is the heart of luck vs skill. Over a small sample, luck dominates, so a good run can look like skill and a bad run can hide a real edge entirely. The only honest fix is more data. As the number of trades grows, variance averages out and the true expectancy starts to show through. That is exactly why people who test ideas seriously care about sample size before they trust a result, and why a single great-looking chart never settles anything.
- Edge
- A measurable lean in your favor over many trades
- Expectancy
- The average result you would expect per trade
- Variance
- The random swing in results from trade to trade
- Win rate
- The percentage of trades that end in a profit
Put these in the order that actually tells you whether an edge is real.
- Write the idea as a clear, testable rule
- Run it across a large sample of trades, not one chart
- Read expectancy, not just win rate
- Judge whether the result survived the noise
A method shows a small loss after 18 trades. What is the most honest conclusion? Eighteen trades is too small a sample to know whether it has an edge. Correct. Over so few trades, variance can easily be larger than any real edge, so the result tells you almost nothing on its own.
You can now tell edge from luck
Edge is an average advantage over many trades, expectancy is how you measure it, and variance is the noise that hides it until the sample is large enough.
Common questions
- What is a trading edge in simple terms?
- A trading edge is a measurable reason to expect a positive result over many trades after costs. It is not a promise about any single trade, just a small statistical advantage that shows up across a large sample.
- What is expectancy?
- Expectancy is the average amount you would expect to win or lose per trade over the long run. It combines how often you win with how big the wins and losses are, so it is more honest than win rate alone.
- Why does variance make it hard to know if an edge is real?
- Variance is the random swing in results from trade to trade. Over a small number of trades, that swing can be larger than the edge itself, so a real edge can look like a loser and a lucky run can look like skill. Only a large sample starts to separate the two.
Terms defined in this lesson
Continue
Sources