The underlying rates are near-identical — attack 1.16 against 1.17, defence 1.00 against 1.02. On rating it is 1563 against 1541, which the model resolves to 49.7% / 25.0% / 25.3%.
The home side is rated higher. The quoted ratings already absorb this match. The single-scoreline output is 1-1. The tie closed out 0-0. Most likely single result is Fluminense on 49.7%. They have met once on record — 2026, 1-1 in comp-fd-BSA, which is too little history to weigh. Outcomes span 24.7 points. In absolute terms: Fluminense 1.16/1.00, Bahia 1.17/1.02 on attack and defence. The model's expected-goals figures are 1.76 and 1.57. Before kickoff our model had this at 50.3% / 24.8% / 24.9%. That one did not come off. The model gives 75.0% to someone winning. Sample sizes behind the two numbers are 27 and 26 matches. 2 scorelines clear 8% individually. The leading outcome clears the second by 24.4 points. Distance between the ratings: 22 points, narrow on this scale. Flat distribution — 1-1 tops it at just 9.9%. On the Elo curve, expected shares run 62.3% to 37.7%. Relative to the 1500 seed that every rating begins at: +63 and +41.
Source: our elo-poisson-v1 model, run across the complete result record. Not betting guidance.