Guide

What the columns mean

The board is dense on purpose: more fixtures per screen means less scrolling. Every column and colour is explained here, including the one that admits we have no signal.

1, X and 2 — the confidence ramp

Three probabilities per match: home win, draw, away win. They always sum to 100%. The fill behind each is a six-step ramp, so a row's shape is readable without parsing digits — one saturated cell means a confident call, three pale ones mean the model has little to say.

ProbabilityFillReading
under 35%no fill33.3% is the three-way coin flip. Below a real lean, colouring it would invent signal
35–44%faintestA lean, not a lead
45–54%lightClear favourite. Roughly where the model's real accuracy lives
55–64%mediumStrong
65–74%deepVery strong
75%+full brand greenRare, and kept rare so it means something

The ramp only ever gets darker as confidence rises. That sounds obvious; it is not universal. At least one well-known prediction site's scale reverses at its top step, making its most confident cell visually quieter than its second-most.

Tip

The single most likely outcome: 1 home, X draw, 2 away. It ignores the odds completely. It is not necessarily what we would bet on — see value betting.

FT

The full-time score, once a match has finished and been graded. Grading runs every two hours, so a score appears within a couple of hours of the final whistle rather than the next morning. A dash means not yet settled, or a competition we have no results feed for.

Safer

Double chance, when the market offers one worth showing: 1X covers home or draw, X2 covers draw or away. Two outcomes out of three, so shorter odds and a higher strike rate. Shown in amber rather than green deliberately — it is an alternative the market offers, not what the model recommends.

Best odds

The best price we found across the bookmakers we survey, for each of the three outcomes. On narrow screens this collapses to the tipped outcome's price only.

Signal

Where a row's numbers come from. This is the most important column on the board and the one most sites do not have:

SignalMeaning
MODELOur statistical model, fitted on that competition's own results
MARKETBookmaker consensus with the margin stripped out. No model for this competition, but real market information
VALUEThe model's edge over the market is large enough to be worth backing
SKIPThe claimed edge is too large. Almost always our error, so we refuse it
No signal. A league-average baseline. Not a prediction

About that last one. Fixtures showing 45% / 27% / 28% are the league-average home-advantage baseline, not model output. We publish them so the board shows the day's full football rather than only the competitions we can model, and we label them rather than dressing a constant as a forecast. If the three numbers are 45/27/28 exactly, there is no signal in that row.

Greyed-out rows

A row fades once its match is over — more than 115 minutes past kick-off, so added time never greys a game out early. It is worked out from your own clock as you read, which is why the board ages during the day without a reload.

Current as of 27 August 2026. Column behaviour changes are noted here when they happen.