A table that disagrees with the model
Bournemouth host Brentford on Saturday at the Vitality Stadium with the Premier League table and Nutmegly's AI read pulling in opposite directions. Brentford arrive fifth with five points from three games, unbeaten in the young season. Bournemouth sit fifteenth with two points, winless in three league matches. Yet the model's lean is home, at 36.1% for a Bournemouth win against 36.0% for Brentford and 27.9% for a draw. That is not a ringing endorsement of the hosts. It is a statement that the three-way distribution is almost flat, and the draw band is very much alive. The most interesting question is not who is favoured but why the table and the model are so far apart.
What the stored results show
The recent-results samples stretch across seasons, so they are best read as historical evidence, not current form. Bournemouth's stored eight-match sample yields 1.38 points per game; Brentford's gives 1.25. That is the home side's strongest signal: a slightly better points efficiency over a longer window. In the Premier League-only sample of five matches each, the picture flips: Bournemouth take 0.80 points per game, Brentford 1.40. The discrepancy matters. The longer sample includes older matches from May 2026; the league-only sample is more recent but smaller. Neither is a full season, and neither can settle the argument on its own. The head-to-head sample is just two matches, so it offers little predictive weight.
The case for the home lean
The AI read's home lean rests on two specific edges. First, Bournemouth's home goals average is 1.80 in the stored sample, while Brentford's away goals-conceded average is 1.40. That is a matchup signal: a home attack with a reasonable scoring rate against an away defence that has leaked goals. Second, Bournemouth's recent points efficiency over eight matches is 1.38, just above Brentford's 1.25. Add the over-2.5-goals projection of 53.15%, and the model sees a game with a slight home tilt and a modest chance of three or more goals. The lean is not a forecast of dominance. It is a marginal call.
But the model itself acknowledges the margins are thin. Its counterpoints note that the three-way probabilities are not clearly separated and that the draw interval must be retained. That is the right emphasis. A 36.1% home win is not a majority; it is the largest slice of a nearly even split. The lean is a tiebreaker, not a verdict.
The counterevidence and the unknowns
The strongest counterargument is the table itself. Brentford are fifth with five points, unbeaten, and have conceded only twice in three league games. Bournemouth are fifteenth with two points, without a win, and have conceded five. In the Premier League-only sample, Brentford's 1.40 points per game is nearly double Bournemouth's 0.80. An external forecast also points in a different direction from the model's home lean, a reminder that sample scope can change the read. None of this erases Bournemouth's home edge in the stored sample, but it makes the home lean feel fragile.
Team news is the other unknown. The data contains no confirmed lineups and no injury reports. That is a limitation, not evidence that both squads are at full strength; it simply means this preview is built on results and standings rather than confirmed personnel. If the lineups shift the balance, the model's thin home lean could look even thinner.
The watchpoint
Watch the goal line rather than the win probability. Over 2.5 goals is projected at 53.15%, just above a coin flip, and that is where the match's shape is most likely to be decided. If either side scores early, the game could open up and make the over the more interesting test. If the first half stays tight, the draw probability of 27.9% becomes the number to watch. Either way, a Bournemouth side winless in the league against a Brentford side protecting an unbeaten start is a genuine test of whether the table or the underlying sample deserves more trust.