Understanding risk: variance, expected value, and why luck is not a strategy

Two punters bet the same selection. One wins. One loses. Both made the same bet. This is variance, and understanding it is the difference between a punter who lasts a season and one who blows a bankroll in a weekend.

By Tolu Shotade · Editor, Bets.ng · Updated 5 May 2026

Two punters bet on Manchester City to beat Crystal Palace at 1.40. One wins, one loses. Both made the same bet. This is variance, and understanding it is the difference between a punter who lasts a season and one who blows a bankroll in a weekend.

Variance and EV, in 60 seconds: why a winning bet can still be a bad bet.

Variance, in plain English

Variance is the gap between what should happen on average and what actually happens on the day. Bet on a fair coin to land heads at even money, fifty times in a row. You expect 25 heads. You will sometimes get 22, sometimes 28, occasionally 18 or 32. All of those outcomes are normal. The expected value is unchanged. Variance is the noise around the signal.

Football betting has more variance than coin flipping because the underlying probabilities are harder to estimate. A bookmaker offering Manchester City at 1.40 is implying a 71% probability. The actual probability might be 73% or 68%. Across a season of bets, the small differences average out. Across a single weekend, they look like miracles or disasters.

Expected value, the metric that actually matters

Expected value, or EV, is the average outcome of a bet if you placed it a thousand times. The formula:

      EV = (Probability of winning × Profit on win) - (Probability of losing × Stake)
    

A 1,000 naira stake at 2.00 odds, with a true probability of 55%, has an EV of (0.55 × 1,000) - (0.45 × 1,000), which is +100 naira. The bet wins or loses on the day, but on average each bet returns 100 naira of profit. Place a thousand identical bets and you finish 100,000 naira up.

The catch is the word "average". The actual hundred-bet stretch could go 60-40 in your favour, or 48-52 against you. Both are normal. Variance.

Why a winning bet can still be a bad bet

You backed a 6.00 long-shot. It landed. You collected. The bet was still a bad bet. The bookmaker priced it at 6.00, implying roughly 17% probability. If the true probability was 14%, the EV was negative, and the win was variance, not skill.

This is the hardest concept in betting and the most important. The result of a single bet tells you almost nothing about the quality of the bet. A losing bet at +20% EV is still a good bet. A winning bet at -10% EV is still a bad bet. Skilled bettors evaluate the decision, not the outcome. Recreational bettors do the opposite.

The variance budget

A betting strategy with a 5% edge over the bookmaker, played at 1% unit stakes, has a positive EV but a meaningful variance budget. Across a thousand bets, the strategy's profit ranges from +30,000 to +200,000 on a one-million-naira simulated bankroll, depending on the variance of the run. The expected average is around +50,000.

In the bottom 5% of variance outcomes, the same strategy loses money over a thousand bets. The edge is real but the sample size is too small to overcome a bad run. Across ten thousand bets, the bottom 5% becomes the bottom 1%, and the variance washes out almost entirely. This is why long-term bettors track everything.

The ledger

If you bet for fun, the ledger does not matter. If you bet to test whether you have an edge, you must track every bet. Date, market, stake, odds, result. Spreadsheet, notebook, phone notes. Anything. Without the ledger, you have no idea whether your wins are skill or variance.

A hundred bets is not enough to tell. Five hundred starts to look meaningful. A thousand is the point at which the variance has compressed enough that your win rate is roughly your true win rate, plus or minus 3%. Tolu's tipster ledger on bets.ng exists for the same reason. Public, full, every bet.

Risk of ruin

Risk of ruin is the probability that a punter goes bust before their edge plays out. The formula factors in win rate, average odds, stake size, and starting bankroll. The shortcut version: the larger your stake-to-bankroll ratio, the higher your risk of ruin, regardless of edge.

A punter with a real 5% edge but a 10% stake size has a meaningful risk of ruin. Around 15% of simulated runs end in bankruptcy before the edge plays out. The same edge with a 1% stake size brings risk of ruin under 1%. Same picks, same skill, completely different long-term outcome.

What luck is and is not

Luck, in betting, is variance you have not earned. A short hot streak is luck if your underlying edge is small. A long cold streak is luck if your underlying edge is positive. Both feel like skill. Neither is.

The Nigerian betting culture treats short-term wins as proof of skill and short-term losses as proof of bookmaker conspiracy. Both readings are wrong. The variance is doing exactly what variance does. The signal is somewhere underneath, only visible at scale. The behavioural side of this is in avoiding chasing losses.

Bottom line

A bet is a decision, not an outcome. The decision is good if the EV is positive. The outcome is variance. Skilled bettors evaluate decisions, track them, and let the sample size do the work. Recreational bettors evaluate outcomes and learn the wrong lessons. The two paths look identical for the first hundred bets and diverge sharply over the next nine hundred. The mechanical companion is bankroll management and the slip-construction companion is smarter slip building.