Analytics Changed Sports Betting. It Mostly Helped the Sportsbooks.

There is a comfortable story about Moneyball and gambling that goes roughly like this. Billy Beane used data to beat richer baseball teams, therefore you can use data to beat the bookmaker.

The first half is true. The second half runs into a problem, which is that sportsbooks got to the analytics first, spent more on it, and are still spending more on it. American operators kept 9.3% of $149.8 billion wagered in 2024, up from 7% in 2019. The house edge grew during the exact period that betting analytics became widely available to the public.

That is worth understanding properly before anyone reaches for the Moneyball comparison, so this piece covers both halves: what analytics genuinely did to sport, and what it genuinely did to betting on it.

What Moneyball was actually about

Michael Lewis published Moneyball in 2003. The setup is familiar. Beane ran the Oakland Athletics on roughly a third of the Yankees’ payroll and could not compete on spending, so he competed on pricing.

The foundational work came from Bill James, who spent night shifts as a security guard at a Kansas cannery working out why baseball’s standard statistics failed to measure what mattered. Batting average ignores walks entirely. RBIs depend heavily on whoever happens to bat ahead of you. Both were central to how players got valued and paid, which meant the market was systematically mispricing certain skills.

On-base percentage was the underpriced skill of that moment. Oakland bought it cheaply and won 103 games in 2002.

They never won a World Series doing it. A long regular season lets small edges accumulate; a five-game series does not, and Beane said as much himself. That caveat is the single most relevant thing in the book for anyone thinking about wagering, because betting is almost always a bet on a short sample.

Analytics is now universal, which means it is no longer an advantage

Every Major League Baseball team runs an analytics department. Statcast has recorded exit velocity, launch angle and spin rate league-wide since 2015. The information gap Oakland exploited has been fully closed for more than a decade.

The same happened everywhere else. Player Efficiency Rating and the broader shift toward three-point efficiency reshaped basketball. The NFL built Next Gen Stats for speed, separation and coverage. Soccer adopted expected goals, now a standard broadcast graphic. Hockey uses Corsi and Fenwick as possession proxies. Tennis and golf run motion analysis on technique and shot selection.

This matters directly for anyone betting. When every team, every broadcaster and every analyst has the same models, those models are not private information. They are already reflected in the price you are offered.

What the books built while everyone was reading Moneyball

Sportsbooks did not sit out the analytics era. They industrialised it.

Modern operators employ quantitative traders, ingest the same tracking feeds the leagues generate, and run automated systems that adjust prices the moment informed money appears. Line movement that once took hours now takes seconds. A publicly available statistical insight is priced in long before a recreational bettor can act on it.

Odds comparison services like Betbrain exist because prices still differ across books, and shopping for the best available line is a real and measurable improvement over accepting the first one you see. That is a legitimate use of information. It is also a marginal one, and it is not the same thing as having an edge.

The arithmetic nobody puts in the headline

At standard -110 pricing you must win 52.38% of your bets simply to break even. Not to profit. To finish level.

Coin-flip accuracy loses you about 4.5% of everything you stake over time. The gap between 50% and 52.38% sounds trivial written down and is the entire business model of the gambling industry.

Long-run profitability estimates cluster between 3 and 5% of bettors. Across a single year, somewhere between 78 and 85% lose money. The hold rate has climbed because parlays, which carry much higher margins and much lower win probabilities, now make up 35 to 45% of mobile betting volume in some markets.

Reporting on operator data has made the revenue structure explicit. Before Fanatics acquired PointsBet, the company’s VIP bettors made up around 0.5% of its customer base and generated more than 70% of its revenue. Sportsbook economics do not depend on many people losing a little. They depend on a small number of people losing a great deal.

So what does data actually get you

Something, but less than advertised, and in narrower places than the marketing suggests.

Line shopping is genuinely worth doing. Understanding that parlays carry punishing margins is worth knowing. Recognising that markets for less-covered competitions are priced less efficiently than NFL spreads is a real observation. Professional bettors sustain win rates of roughly 53 to 55%, which is enough to profit and is achieved through specialisation, volume, discipline and accounts at many books, not through a clever model applied on a Sunday afternoon.

What data does not do is convert betting into a skill game with a positive expected return for an ordinary participant. The vig is charged on every wager regardless of how well informed it is. The claim that analytics has removed luck from betting inverts what actually happened: the books used analytics to reduce their own exposure to luck, and the cost of that was passed to the people betting into them.

The part of Moneyball that translates

There is one lesson that does carry across, though it points somewhere unexpected.

Beane traded Jeremy Giambi in May 2002 for reasons his own statistics did not support, because he judged the clubhouse cost higher than the on-base value. The Moneyball approach worked not because Beane trusted models absolutely, but because he understood exactly what his models could and could not tell him.

That is the transferable discipline. Know the boundary of what your information covers. In betting, the boundary is unusually close, the opponent pricing the market has better information than you do, and a 9.3% hold sits between you and break-even on every ticket.

Beane’s edge came from finding a market that had not yet done the analysis. Sports betting is not that market. It is one where the analysis was completed, professionalised and automated by the party taking the other side of your wager.