The maths most punters
never get told.

1 · Expected Value (EV)

Every bet has an expected value — what it pays on average over many repetitions. A simple formula:

EV = (probability of winning × profit if you win) − (probability of losing × stake)

Example: you bet ₦1,000 at decimal odds 2.50 on a team you think has a 45% chance of winning.

profit if win  = ₦1,000 × (2.50 − 1) = ₦1,500
EV  = (0.45 × ₦1,500) − (0.55 × ₦1,000)
    = ₦675 − ₦550
    = +₦125 per bet (positive — value bet)

Flip it. Same odds, but you only think there's a 35% chance:

EV = (0.35 × ₦1,500) − (0.65 × ₦1,000)
   = ₦525 − ₦650
   = −₦125 per bet (negative — sucker bet)

Bookmakers' edge is built into the odds. To win long-term you need bets where your estimate of probability is more accurate than the bookmaker's — that's the only edge that exists. When you see the green VALUE tag on Mintscore, it means our model thinks the bookmaker has mispriced the line.

2 · Bankroll & stake sizing

Your bankroll is the total money you've decided to risk on betting. It must be money you can afford to lose entirely without affecting rent, food, school fees, or savings. If you can't afford ₦20,000 to disappear, your bankroll is not ₦20,000.

The Kelly Criterion is the mathematically optimal stake size given your edge:

stake fraction = (edge as decimal) ÷ (decimal odds − 1)

Full Kelly is theoretically optimal but emotionally brutal — losing streaks chew through your bankroll fast and your model is never as accurate as you think. Use quarter-Kelly (the stake we recommend on each match): same expected growth direction, far less variance. Even quarter-Kelly will rarely tell you to stake more than 2-3% of bankroll on a single bet.

Rule of thumb: if you find yourself wanting to stake 10%+ of your bankroll on a single bet, it's not edge talking — it's tilt.

3 · Why accumulators are the trap

The bookmaker's margin compounds. If a single match has 7% margin against you, a 5-leg accumulator stacks that 5 times: roughly 1 − 0.93⁵ ≈ 30% margin against you. A 10-leg acca? Roughly 50%. The huge potential payouts disguise that the house edge has gone from "you'll bleed slowly" to "you're funding the building."

Singles or short doubles on value bets beat 10-leg accumulators long-term — every time, by a lot. If you must build accumulators, keep them to 2-3 legs and only on selections flagged as value.

4 · How the Mintscore model works

For each match we estimate two numbers: the home team's expected goals and the away team's expected goals. These come from each side's recent attack and defence ratings combined with the league's baseline scoring rate.

From those two numbers we build a full distribution of possible scorelines using the Poisson distribution — the standard statistical model for goal counts. We then apply the Dixon-Coles correction (Dixon & Coles, 1997, JRSS) to fix a known weakness in plain Poisson: real football has more 0-0, 1-0 and 0-1 results than independent Poisson predicts.

Sum the joint probabilities of all home-win scorelines, all draws, and all away-win scorelines, and you get the 1X2 probabilities you see on every match page. That's the entire method. No insider tips, no AI black box, no "guaranteed wins" — published openly so you can challenge it if you spot a flaw.

The model has known limits: it doesn't yet incorporate injuries, suspensions, motivation/form swings, or specific head-to-head context. That's the next iteration.