We don’t predict certainty.
We show probability.
A rolling strength rating (shown 0–99) computed from seasons of real J1 League results, updated after every matchweek. The method is a proprietary statistical composition; the record is public.
A fixed home-advantage term is applied to the ratings gap, then converted into each side's expected goals (e.g. 0.96 vs 2.10).
Expected goals are expanded into the probability of every scoreline, giving the 1X2 split, likely scores and half-time leans. “Higher confidence” needs ≥65%; otherwise “evenly matched”.
Forecast integrity
Every forecast is written to the record 24 hours before kickoff and never edited or recomputed.
Post-match pages always show the locked pre-match values; unlocked fixtures are never back-filled.
Wrong calls are kept in full, with the probability the model gave that result pre-match.
The model reads results only — injuries, transfers and squad changes are not inputs, and pages say so.
This season's record
Over/Under 2.5 direction (separate record)
For this competition the model clearly beats a long-run-average baseline, but has not reached parity with the bookmaker market — so we make no market comparison here. All probabilities are model outputs; the record is public, match by match.
Sukina Football is not a betting service. All analytics are for reference only — statistical estimates, not betting advice; past accuracy does not indicate future results.