How we predict
Every MantisPicks football tip is generated the same way: with maths, not hunches. Here is exactly how.
1. Expected goals (the model)
For each team we take recent scoring and conceding form and convert it into an attack and defence strength relative to the league average. Multiplying these gives each side's expected goals (xG) for the match — how many goals they should score on average against this specific opponent, with a home advantage adjustment.
2. Poisson distribution
Goals in football arrive as rare, independent events, so the Poisson distribution models them well. From each team's expected goals we compute the probability of every scoreline (0-0, 1-0, 2-1…). Summing those gives us clean probabilities for match result (1X2), Over/Under 2.5, both teams to score and the most likely correct scores. You can see this on every football tip page.
λ = expected goals · k = number of goals
3. Value vs the odds
A prediction is only a bet when the price is wrong. We convert bookmaker odds into an implied probability and compare it to our model. When our probability is higher than the odds imply, there is positive value — and only then do we rate it a pick. This is why we show a "% value" badge on qualifying tips.
4. Staking & tracking
We publish results with flat 1-unit stakes so performance is honest and comparable. Every settled tip — win or loss — is logged on our Results page. We never delete losers.
Basketball
Basketball is high-scoring, so goals-based Poisson doesn't fit. Instead we project each team's points from their scoring and defensive form versus league average, then use a normal distribution to get moneyline win probability (from the projected margin) and Over/Under probability (from the projected total vs the line). Same transparent, value-first idea — different maths for a different sport.
Important
No model is a crystal ball. Even a 70% pick loses roughly one time in three. Betting carries risk — treat our tips as informed opinion, stake responsibly, and never chase losses. 18+.