Perpignan v Racing 92
Set a scoreline and see where it sits among the 5,000 simulated runs behind this fixture. Ratings as at 2026-10-19.
PerpignanRacing 92The steppers skip 1, 2 and 4. Points come in fives, threes and twos with a conversion needing a try in front of it, so no match has ever finished on any of them.
19.8% of runs finished in this bin — within 2 points either way.
51 points is around the 56th percentile of the total — the middle run scored 48.
Those two figures are separate readings and are not multiplied together. The simulation draws the gap and the amount of scoring independently, so one bar of the histogram covers many scorelines and the payload holds nothing about which of them came up with which gap.
Your pick names a different winner from the model, which makes Perpignan more likely. The model’s two answers point different ways here: it makes Perpignan the more likely winner while its average scoreline has the two level. That is not a contradiction — the average gap is taken over both winners, so a side that wins narrowly more often can still be behind on it. Nothing here scores any of it. The model’s gap is an average over both winners and yours is one afternoon, so the two are different kinds of number rather than one being right. How the model works.
Why Perpignan are favoured
How the model arrives at the chance of Perpignan winning. It starts at an even 50%, and each line below adds to that or takes from it, in percentage points (pp): a bar to the right helps Perpignan, a bar to the left helps Racing 92.
Perpignan 1424, Racing 92 1632 — 208 points, favouring Racing 92
Perpignan are at home
Recent results, past meetings between the two sides, and days of rest. It has a limit, so it can nudge the rating but never overrule it.
Perpignan and Racing 92 have no fixture on the published schedule, so there is no ground, no kick-off forecast and no team sheet to put beside this. The prediction itself does not need any of them.
14 meetings — Perpignan 4, Racing 92 10. Last 2026: Racing 92 55–25 Perpignan.
None of this reaches the model. Weather is not a feature, team selection is not a feature and the record is already inside both ratings, so the probability above does not move because any of it changed.
Instagram has no way for a website to post on your behalf and does not unfurl links, so a post there is the saved image with a caption. The copied link unfurls into the same card on X, WhatsApp, Slack and Facebook.