I used to take pity on the punter who walked into a betting shop and put a tenner on a 250-to-1 NFL outright. He looked happy with his ticket and he had a story to tell his mates for six months. What I have come to understand after nine years of pricing these markets is that he was almost certainly losing more than he realised — not because his team was definitely not going to win, but because the price he took was systematically shorter than the team’s true championship probability warranted. That mathematical reality is what bookmakers and academics call longshot bias, and it is the single most reliable structural quirk in the NFL futures market.
Longshot bias is the reason 100-to-1 tickets feel exciting and look like value but rarely produce positive expected returns. It is also the reason a disciplined UK punter who understands the phenomenon can occasionally find genuine longshot value when the market gets the probability assessment wrong. This piece is about both sides of that coin.
What longshot bias is and where it came from
Longshot bias is the observation that bookmakers systematically price longshot outcomes shorter than the underlying probability warrants. The first rigorous studies of the phenomenon came from horse racing in the 1940s, when economists noticed that betting on favourites in pari-mutuel pools generated a smaller average loss than betting on longshots, even though both groups of bets should have produced equal losses if the prices were unbiased. The pattern has been replicated in countless studies across decades and across sports, and NFL futures markets show it as clearly as any market I have analysed.
The technical explanation is that longshot prices include a larger margin slice than favourite prices. When a bookmaker prices a 32-team Super Bowl board with a 35 per cent overround, the margin is not evenly distributed across all teams. The favourites carry roughly 5 to 10 per cent margin within their implied probability. The longshots carry 100 per cent or more. A team whose true championship probability is 0.3 per cent might be priced at 0.1 per cent implied, which means the bookmaker is taking three times the team’s actual probability as margin.
The behavioural explanation runs parallel. Recreational punters love long-priced tickets because the potential payout is large relative to the stake, and the psychological appeal of a “lottery ticket” outcome makes them willing to accept worse pricing on longshots than they would accept on favourites. Bookmakers know this and price accordingly. The longshot bias is therefore both a structural margin mechanic and a behavioural exploitation of recreational psychology.
Evidence of longshot bias in 25 years of Super Bowl boards
The cleanest evidence of NFL longshot bias comes from comparing preseason futures prices with actual championship probabilities computed from underlying team strength. Take the 25-year history of Super Bowl outrights and compare the implied probability of each team’s preseason price with their actual championship rate. Teams priced at 50-to-1 or longer have won the Super Bowl at a rate below what their preseason implied probability suggested. Teams priced between 5-to-1 and 25-to-1 have won at rates roughly matching their implied probability. Teams priced at 4-to-1 or shorter have won at rates slightly higher than their implied probability suggested.
That pattern is exactly what longshot bias predicts. The favourites are mildly underpriced relative to their true probability — they offer marginal positive expected value in the long run. The middle of the board is roughly fair. The longshots are systematically overpriced and produce negative expected value over time. The pattern has held with remarkable consistency across the 25-year window, including the Rams’ 1999 longshot championship at 150-to-1, which was a clear case of a team being so badly underpriced that the bookmaker’s longshot bias did not catch the underlying probability shift quickly enough.
The relevant size of the bias is not subtle. On the longest 25 per cent of the Super Bowl board, the average implied probability sits roughly 30 to 40 per cent above true probability. A team priced at 100-to-1, implying 0.99 per cent, may have a true championship probability closer to 0.7 per cent. A team priced at 250-to-1, implying 0.40 per cent, may have a true championship probability closer to 0.28 per cent. Multiply those gaps across hundreds of recreational tickets and you understand why bookmakers are happy to take longshot action all day.
Why UK punters love a 100-to-1 shot
The behavioural side of longshot bias is worth understanding because it is the reason the bias persists despite being well-documented for 80 years. UK NFL punters are particularly susceptible to longshot bias for a specific reason: the NFL season is long and the Super Bowl is a single dramatic event in February, which makes the lottery-ticket framing especially powerful. A £10 preseason ticket on a 100-to-1 team gives the punter six months of anticipation and a small but real chance at a £1,010 return. The psychological math is irresistible even when the financial math is poor.
Compounding the bias is the structure of UK NFL coverage. Sky Sports and other broadcasters give significant airtime to underdog narratives across the season, which keeps longshot stories visible and reinforces the punter’s belief that “anything can happen” in the NFL. The 217 per cent increase in UK female NFL betting between 2020 and 2024 has not changed the structural bias — if anything, new market entrants are more susceptible to longshot bias than experienced punters, which has likely deepened the structural margin advantage for UK bookmakers on longshot markets.
The recency factor matters too. Every time a genuine longshot wins the Super Bowl — the 2007 Giants, the 2017 Eagles, the 1999 Rams — the next year’s longshot ticket sales spike. UK books know this and tighten longshot margins in years following longshot championships, which means the post-championship longshot board is even worse value than the steady-state board. The Eagles’ price the year after their 2017 win was meaningfully shorter than the team’s actual championship probability deserved.
Exploiting longshot bias responsibly
The natural question is whether the bias can be reversed — whether there are situations where longshot prices offer genuine value despite the structural pattern. The answer is yes, but rarely, and only with a clear methodological approach. The first situation is when a team’s roster has undergone a significant talent upgrade that the market has been slow to recognise, particularly at quarterback. The Rams in 1999 are the textbook example. Their roster overhaul was real, the new quarterback was elite, and the bookmaker’s 150-to-1 longshot price reflected the team’s history rather than its current capability.
The second situation is the late-season longshot on a team that has emerged as a genuine contender from a previously unexpected start. When a team that was 50-to-1 in August reaches 8-2 by mid-November, their futures price will have contracted meaningfully but often not enough to reflect their now-elite probability. The bookmaker is slow to fully reprice in part because the longshot bias structurally biases their thinking against medium-priced teams emerging into the favourite tier.
The third situation is the divisional or conference longshot rather than the Super Bowl longshot. The longshot bias is most pronounced on the longest end of the longest market, which is the 32-team Super Bowl board. Smaller markets — eight-team divisions, two-conference championships — have less structural longshot bias because there is less room for the bookmaker to overprice the longest end. UK punters looking for genuine longshot value should focus on these smaller markets rather than chasing 250-to-1 Super Bowl tickets.
For punters who want to understand how this all interacts with bookmaker overround mechanics specifically, the practical work in my piece on the Super Bowl prop bets vs Super Bowl futures difference covers the related question of how one-game wagering markets handle the same probability distortions.
When 100-to-1 is genuinely 200-to-1
The most important lesson from 25 years of longshot data is that the recreational punter’s instinct — “100-to-1 feels long enough that it must be value” — is almost always wrong. The headline price you see on a longshot futures market is shorter than the team’s true probability, often by a wide margin, and the only honest response to that fact is to either stake longshots in markets where the bias is mild, or to stake favourites and second-tier contenders where the bias actually favours the punter slightly. The 100-to-1 ticket that feels like a lottery ticket is, on the maths, almost always closer to a 200-to-1 ticket on probability. Knowing that does not stop me from occasionally backing a longshot when the structural circumstances genuinely justify it, but it keeps my stake size small and my expectations honest.
