Not a model of ours at all: the bookmakers' collective opinion, read straight from the odds. It is the benchmark every other column should be judged against, and the gap between it and the Poisson model is the Value flag.
How it works
Take the average decimal odds across bookmakers that football-data.co.uk publishes for home, draw and away (and for over/under 2.5 goals).
Convert each to an implied probability: 1 / odds. Odds of 2.00 imply 50%.
Those three add up to more than 100% - the bookmaker's margin, or overround (typically 104-108%). Divide each by the total so they sum to 100%.
Value: for each of home, draw and away, subtract the market percentage from the Poisson percentage. If the biggest gap is 5 points or more, that outcome is flagged, e.g. A +8.
Worked example
Average odds
home 1.53, draw 4.12, away 6.02
Raw implied
65.4% + 24.3% + 16.6% = 106.3% (6.3% overround)
Normalised
home 62%, draw 23%, away 16% → 62:23:16
Poisson said
42:28:29 → away gap +13 → Value: A +13
How BetChair calls it
The market's favourite is its vote in the H:D:A column. Odds appear on football-data about a week before kick-off, so fixtures further out show a dash and the market abstains (seven voters instead of eight). Value only ever compares Poisson with the market - it does not consult the form models.
Where it shines
The single hardest benchmark to beat: it already contains team news, motivation and everything the models cannot see.
Bookmaker odds are well calibrated - a 60% market favourite wins about 60% of the time over a season.
Where it falls down
It is not a prediction you could not have made yourself by looking at the odds.
Value is only as good as the model on the other side of it: early-season Value flags mostly reveal what Poisson does not know (promoted sides, new signings) rather than mispriced markets.