Variance in football betting: why EV+ can lose in the short term

Quick summary

What is variance in football betting?

Variance is the annoying but inevitable part of betting with a method. You can make a correct read, get an odd above the fair price, enter with positive EV, and still lose that bet.

In football, this happens all the time. A missed penalty, a red card, a lucky goal, a shot off the post, an injury in the first half, or a team that dominates but does not score can change everything.

This separation distinguishes the isolated result from the process analysis.

Why does football have so much variance?

Football is a low-scoring sport. A game can be decided by a single play, and the team that creates more does not always win.

• Few goals change everything: a single play can decide the outcome.
• The scoreline is misleading: a team can play better and still lose.
• Rare events carry heavy weight: red card, penalty, VAR, injury, or individual mistake.
• Markets move differently: 1x2, goals, BTTS, and corners do not have the same risk.
• Small samples deceive: one week can look great or terrible without representing the strategy.

In the Brasileirão, Série B, Copa do Brasil, or any league, a short sequence can look great or bad without revealing the real quality of the method.

Variance is not the same as a bad strategy

Not every losing streak is variance. Sometimes, the strategy is genuinely poorly built. The difference shows up in the signs.

Variance is part of it. The problem is using variance as an excuse to keep betting without price, without base, and without management.

The mathematical logic of variance

The main formula for understanding value betting is:

EV = (estimated probability × odds) − 1

If the result is greater than zero, the bet has positive EV, that is, positive expected value in the long run according to that estimated probability.

But EV does not eliminate fluctuation. It only says that, repeating similar bets many times, that price tends to be interesting if the probability is well calibrated.

In the first formula, p is the estimated probability of winning and n is the number of consecutive bets.

p = 55% Chance of 5 consecutive losses = 45%^5 Approximate chance = 1.85%

It seems like little, but if you bet frequently, losing streaks appear. Not because of eternal bad luck, but because that is how probability works.

Practical example in football

Imagine two scenarios in football games:

Both bets have positive EV. Only Bet B has much more variance. It can be mathematically interesting, but it will lose more often. Higher EV does not mean a smoother path.

Which markets tend to have more variance?

There is no perfect rule, but you can think about it this way:

The point is not to run from every volatile market. It is to know which volatility you are buying.

Why can positive EV lose?

Why Positive EV is not a prediction for a specific game. It is an expected average when repeating similar bets many times.

A bet with 55% of chance still loses in 45% of similar scenarios, on average. And these losses do not come alternating one by one. They can come in a sequence.

Estimated chance: 55% Odd: 2.05 EV = 0.55 × 2.05 − 1 EV = +12.75%

This is a value bet on paper. Even so, it still has a 45% chance of losing. EV means the price looks good for that probability, not that the bet will win.

How to interpret variance in the long run

In the short term, results rule your emotions. In the long term, process rules your analysis.

A negative week does not define a strategy. Likewise, a very positive week also does not prove you found the holy grail.

1. Market: compare 1x2 with 1x2, goals with goals, and corners with corners.
2. League: evaluate competitions separately when the base allows.
3. Filter: keep the same filter before concluding anything.
4. Sufficient sample: do not treat a few entries as strong truth.
5. Accumulated return: compare how the price relates to the result over time.
6. Factor coherence: identify convergences and divergences before the game.

How to apply this on Predora.si

At Predora.si, variance is not treated as an excuse or a mystery. It is part of the process reading.

Variables to contextualize variance

1. Market: distinguish 1x2, Double Chance, Over 2.5 goals, Both Teams to Score, and Over 9.5 corners.
2. Odds: compare the observed price with the estimated fair odds; price changes alter the EV.
3. EV: distinguish high probability from positive EV.
4. Sample: compare more observations to assess the subset’s stability.
5. Factors: explore which assumptions explain the conflicting data.
6. Stake: changes based on the previous result alter exposure.
7. Exposure: compare the proportional impact of a loss on the simulated bankroll.

Common mistakes about variance

Good variance vs. bad variance

Not all variance is a problem. The problem is buying variance without an edge.

How to investigate a segment more deeply

Use these signals to identify which parts of a segment deserve a closer look:

• the odds moved and the EV got close to zero;
• the factors show significant disagreements;
• the league sample is small;
• exposure changes in response to a previous loss;
• the market shows higher volatility;
• the scenario's criteria cannot be described before the game.

An isolated result doesn't validate the process; sample, period, and comparable criteria contextualize the series.

How variance shows up in historical records

In practice, hit rate and return will never be a straight line.

A red week doesn't destroy the method. A green week doesn't prove brilliance. Sample, league, market, and cumulative return help separate signal from noise.