Soccer Analytics Tools & Calculators
Run the models yourself — xG to outcome, Elo to win expectancy, shootout odds, and a World Cup 2026 tournament simulator.
These are the actual models behind the explainers, made interactive. Type in numbers, get a result, see the method that produced it, and read what it means. Everything runs in your browser from your inputs — there are no betting tips, no scraped feeds, and (for the tournament tool) no real World Cup results: just transparent maths you can poke at.
xG → match outcome & xPoints
Treat each side's total xG as a Poisson mean, build the scoreline grid, and read off win / draw / loss probabilities and expected points.
Open calculator →Elo / SPI expected score
Turn a rating difference into a win expectancy with the logistic Elo curve — the same formula clubelo and SPI build on.
Open calculator →Penalty shootout probability
Given a per-kick conversion rate, how often is a best-of-five shootout (plus sudden death) won, and how often does it stay level?
Open calculator →World Cup 2026 simulator
A Monte-Carlo group-stage + knockout simulator over the real 48-team format. You set the strengths; it reports advance and title odds. Model only — no real results.
Open the simulator →xG → match outcome & expected points
Add up each team's shot xG and enter the two totals. The tool models each side's goals as an independent Poisson count with that mean, multiplies the two goal distributions into a full scoreline matrix, and sums the cells into win, draw and loss probabilities. Expected points then follow as 3·P(win) + 1·P(draw). This is the engine behind expected points (xPts).
This calculator needs JavaScript. With it off, the method is simple: model each team's goals as Poisson with mean equal to its total xG, build the scoreline grid, and sum into win / draw / loss; expected points = 3·P(win) + 1·P(draw).
Elo / SPI expected score
Every Elo-style rating — clubelo for clubs, SPI and World Football Elo for nations — turns a rating gap into an expected result through one logistic curve. Enter your team's rating and the opponent's; the tool returns the expected score, a number between 0 and 1 that blends wins and draws. See soccer power ratings: Elo, SPI and why they disagree for what the systems do differently around this shared core.
This calculator needs JavaScript. The formula is E = 1 / (1 + 10^(−ΔR/400)), where ΔR is your rating minus the opponent's. A 400-point gap implies a 10-to-1 expected-score ratio.
Penalty shootout probability
The penalty is football's most studied shot, converted about three times in four. Enter a per-kick conversion rate and the tool computes — exactly, from the binomial — how a standard best-of-five shootout plus sudden death plays out between two equally-matched sides. The counter-intuitive result it makes visible: a higher conversion rate makes the shootout more likely to stay level after five kicks. Background in the penalty conversion rate.
This calculator needs JavaScript. The method: each of five kicks is an independent Bernoulli trial with probability p, so made kicks follow Binomial(5, p); compare two such sides and break any remaining tie 50/50 for sudden death.
World Cup 2026 simulator
The flagship tool is large enough to deserve its own page. It runs thousands of Monte-Carlo tournaments over the real 48-team, 12-group format — top two of each group plus the eight best third-placed teams into a 32-team knockout — using the team strengths you supply, and reports advance and title odds with a sample bracket. It is framed strictly as a model: it contains no actual World Cup 2026 scores, standings or results, and the numbers move entirely with your inputs.
Open the World Cup 2026 simulator →
How to read these well
Three honest cautions carry across all four tools. First, one match is a single sample from a noisy process — these probabilities describe the long run, not the next ninety minutes. Second, the output is only as good as the input: garbage xG or made-up ratings produce confident nonsense. Third, none of this is betting advice; the probabilities here are analytical tools for understanding why results happen, nothing more. For the assumptions behind each model, follow the linked explainer next to it.
Last updated 14 June 2026.