Expected Value: The Foundation of Profitable Trading

Expected Value (EV) is the single most important concept in prediction-market trading. It's the difference between gamblers who randomly win and lose and traders who compound bankrolls year after year. If you only learn one piece of math, learn this one.

The Definition

Expected Value is the average outcome of a trade if you could repeat it infinitely many times. The formula:

EV = (Probability of Winning × Profit if You Win) − (Probability of Losing × Loss if You Lose)

If EV is positive, you make money on average. If it's negative, you lose money on average. It doesn't matter what happens on a single trade — what matters is that you only take positive-EV trades, and you take enough of them for the law of large numbers to play out.

A Concrete Example

Polymarket has a market: "Will the Fed cut rates in June 2026?" The Yes share trades at $0.40. That means the market thinks there's a 40% chance.

Your research, news monitoring, and analytical models tell you the real probability is closer to 55%. You're confident enough to act.

If you buy one Yes share at $0.40:

EV per share = (0.55 × $0.60) − (0.45 × $0.40) = $0.33 − $0.18 = $0.15 of expected profit per $0.40 risked.

That's a 37.5% expected return per trade. If your probability estimate is accurate, you'll make money about 55% of the time, lose 45% of the time, and the wins more than offset the losses.

EV vs. Edge

In Predite's UI, we usually display "edge" rather than raw EV, because edge is easier to compare across markets. Edge is the gap between your estimated probability and the market's implied probability, measured in percentage points.

In the example above: estimated 55% − market 40% = 15pp edge.

The relationship: bigger edge generally means bigger EV, but the EV also depends on the absolute price. A 5pp edge at the $0.05 / $0.95 ends of the market is mathematically larger in EV terms than a 5pp edge at $0.50 / $0.50, because the payout ratio is more favorable.

A quick mental rule: edge above 5pp is worth investigating; above 10pp is a strong signal; above 15pp is rare and either a real opportunity or a sign the resolution criteria are tricky.

Why Most Traders Lose Money

Most traders never compute EV. They trade based on:

None of these are positive-EV strategies. They're how you lose money slowly. EV-driven trading is boring — you take a bunch of small +EV bets, you lose roughly the fraction you predicted you'd lose, and the wins compound.

The Two Components You Need to Estimate

To compute EV, you need two things:

1. The market price. That's free — it's right there on the screen.
2. Your estimate of the true probability. This is the hard part. It's where 95% of the work is.

Predite's AI Probability Engine helps with #2. Our 3-model consensus (a large language model analyzing the situation, a heuristic model checking liquidity and momentum signals, and a market microstructure model looking at order flow) produces an estimate calibrated against thousands of resolved markets. You can see the AI estimate, the market price, and the resulting edge in the EV Scanner. But you should also build your own intuitions and adjust the AI estimate based on information the model might not have.

Variance: The Thing That Will Test You

Even with a +15pp edge on every trade, you will lose individual trades. You'll lose three in a row sometimes. The math says you should win 55% of the time, which means 45% you don't. Variance is brutal over small sample sizes.

A useful mental model: imagine flipping a biased coin that lands heads 55% of the time. Over 10 flips, you might see anywhere from 3 to 8 heads. Over 100 flips, you'll see 45-65 heads with very high probability. Over 1000 flips, you'll see ~550 heads, very close to the true rate.

Your bankroll is the same. The first 10 trades will be random. By trade 100, the math starts to show. By trade 500, your win rate should be very close to whatever edge your model actually has.

This is why we keep saying: stick to small position sizes when you start. The Kelly Criterion guide explains exactly how small. For now: never bet more than 1-2% of your bankroll per trade until you have at least 50-100 resolved trades to evaluate your true edge.

Practical Checklist Before Every Trade

Before clicking buy, mentally run through this:

If you can't answer all six, don't trade. That trade isn't ready.

Where the AI Edge Comes From

Predite's models exploit four specific inefficiencies, all of which produce measurable EV:

Combined, these produce a measurable monthly edge if you take a large enough sample. Predite's job is to find the candidates; your job is to verify, size correctly, and execute.