In winter, one analyst noticed something strange: in high-altitude matches, teams “run out of steam” faster, while the home side unexpectedly “exceeds” expectations more often. He didn’t chase miracles — he simply checked the figures for venues, referees and the calendar, and then adjusted the model for specific locations and dates. Over time, his curve grew not because of luck, but because of systematic imbalances that the market had long ignored. This strategy does not seek a “secret formula”; it catches consistent deviations from the norm and monetises them until the environment changes.
What to Consider an Anomaly and When It “Works”
An anomaly is a recurring shift in statistics relative to the basic model of the game that cannot be explained by the strength of the teams. In sports, such shifts can be structural (altitude, empty stands, competition format) and behavioural (how the market reacts to a series of victories or media hype). An important criterion is explainability and verifiability: an anomaly must be supported by data, not just stories.
Structural Anomalies: Venue, Altitude, Spectators
Altitude above sea level is a classic source of advantage for the home team. In the NBA, Denver has for years created the most “tough” home environment precisely because of the thin air; visiting teams’ players tire more quickly, and the Nuggets’ “home advantage” metric is consistently above the league average.
Empty stands during the COVID seasons also reduced the usual home advantage: the pressure from spectators disappeared, the behaviour of referees changed, and some leagues recorded a decrease in the “home effect.” This natural experiment showed how sensitive the game is to external conditions — and how quickly the odds can change.
Behavioural Anomalies: Market Reaction and “Hot Hand”
The market tends to overpay for narratives. In the NFL, for example, the “hot hand” effect leads to an increase in the volume of bets on teams with winning streaks — even when the spread remains unchanged. For a contrarian, this is a signal: the price may be inflated not by the strength of the team, but by public expectations.
Cases: Where and How Anomalies Turn Into Price
This chapter shows how to formalise an anomaly in a specific type of bet, based on historical patterns and the context of the game.
Home Underdogs in American Football
A classic example is the long-standing “home underdogs” in the NFL. A number of studies based on data from previous seasons have found that the total return on such positions exceeded the commission, i.e. the market systematically underestimated the “home + underdog” combination. Even if this premium melts away over time, the very fact of historical overpayment for favourites is an illustration of a persistent anomaly.
Altitude and Score: Totals in Denver and Beyond
In baseball, Coors Field in Denver is a laboratory for the influence of altitude: the thin air changes the movement of the ball and widens the “window” for a hit, which historically pushes the score up. For strategy, this means that the baseline totals at this arena are not equal to the “league average” and require separate calibration. Public stories and data from recent years regularly highlight how different the game becomes at a mile above sea level.
Umpiring Patterns in Baseball
The human-controlled strike zone inevitably carries minor systemic biases: it “shrinks” on two strikes, expands on a 3-0 count, and consecutive calls influence the next one. These are small but recurring effects; knowing them is important for models on inning totals and pitcher/batter prop markets.
How to Find and Validate Anomalies
Any statistical finding must go through the same “conveyor belt”: hypothesis formulation, overfitting test, stress test for regime shifts, and liquidity assessment:
- Formulate an explanation. “Denver has higher totals” is a weak formulation; “thin air changes the movement of the pitch → more damage to fastballs → xwOBA increases in the upper zone” is already a model of causality.
- Divide the story into periods. Check the stability of the effect before/after changes in rules, calendar, and audience environment (example — “empty stands”).
- Separate the signal from the media noise. If the price increase is explained only by news of a winning streak, check whether there is an overpayment for a “hot hand.”
- Introduce applicability filters. For height, these are specific venues; for referees, specific referees and scores; for home underdogs, narrow spread corridors.
- Assess costs and limits. An anomaly is useless if it is realised in markets with negligible limits or delays where you are overtaken.
- Watch for degradation. As soon as the market has overestimated the effect and the line has begun to factor it in, reduce your stake or “turn off” the module.
Risks: Why Anomalies Die and How to “Resuscitate” Them
It is important to recognise that any anomaly exists in a specific environment. As soon as the rules, schedule or audience behaviour change, the premium itself changes.
Market Adaptation and “Burnout” of the Effect
As more and more players and analytical teams begin to target the same signal, the line shifts in its direction — and excess returns disappear. This has happened with a number of “calendar” effects that have become part of basic assessments.
Regime Shifts in the Environment
The return of spectators after “empty stands” partially restored the home advantage; changes in the calendar structure cut out some of the “easy” spots; referee health issues and the introduction of technological assistants are changing the profile of errors. Therefore, the model must have “mode switches” that track the scale of external shifts.
Controlling False Positives
If you sift through hundreds of signals, some will “work” by chance. Combat this with simple methods: set aside a control period outside of training, apply corrections for multiple checks, and evaluate the stability of the effect on neighbouring data windows. Most importantly, demand a causal explanation, not just a nice curve.