How PicksIQ predictions work
The model, the markets, the probabilities, the value picks — explained without jargon. Read this once and the rest of the site clicks into place.
By Bela B. · Last reviewed 5 May 2026
PicksIQ uses a data-driven model to estimate the probability of different outcomes across football fixtures. The aim is not to guarantee results — it is to give users a clearer view of how likely certain outcomes may be, based on available data.
What the model looks at
A range of football performance signals get folded into every prediction. Different markets weight them differently — a 1X2 prediction leans on outcome priors, while goal-totals markets care more about scoring patterns and defensive weakness.
Markets covered
PicksIQ provides predictions across six popular football markets. Each market is calculated separately — a strong winner does not automatically mean a strong over/under or BTTS angle.
Model probabilities
Every prediction starts with a probability. If PicksIQ gives Over 2.5 Goals a 64% probability, that means the model estimates that outcome would happen around 64 times out of 100 in similar conditions.
These probabilities are not certainties. They are estimates based on data.
≈ 64 times out of 100 in similar conditions. Not a guarantee — a frequency.
The market may be underestimating this outcome. PicksIQ flags it as potential value. It does not mean the pick will definitely win.
Value picks
A value pick appears when the model probability suggests an outcome may be more likely than the implied probability of the available odds.
If the model estimates 60% but the odds imply 50%, PicksIQ flags it. This is the signal we're hunting — not certainty, just a structural edge that, played consistently with sane stakes, builds a positive expected value over time.
Confidence
Confidence reflects how strongly the available data supports a prediction. Higher confidence picks usually have stronger agreement across the model's signals. Lower confidence picks may still be interesting — but they're more uncertain.
Football is unpredictable. A high-confidence prediction can still lose. A low-confidence prediction can still win. Confidence helps you understand the strength of the signal, not the outcome.
Why predictions can change
Predictions and value signals may shift as new information arrives — odds movement, team news, fixture updates, market changes, fresh data.
Always check the latest available information on the event page before making any decision.
Under the hood
Models expected goals per team, calibrated on a rolling window of recent fixtures.
Balances bankroll growth against variance — Kelly without the rollercoaster.
Powered by Claude, Ollama and ChatGPT models, constrained to the model's probability output — the prose can never recommend a market the underlying numbers don't support.
Fixtures, lineups, results, season stats. Refreshed daily throughout each match day.
What PicksIQ is not
The platform is designed to provide football analysis, probabilities and tools for users who want a more transparent way to assess fixtures.
Use this as analysis, not advice
PicksIQ should be used as an analysis tool, not as financial advice. Betting always carries risk, and users should never stake more than they can afford to lose. The best way to use PicksIQ is to compare model probabilities, understand the reasoning behind a pick, and make your own informed decision.
If gambling is a problem, please contact BeGambleAware.org or call the National Gambling Helpline on 0808 8020 133. 18+.