Alex MorganHow suspicious betting is detected — and why odds moves are not proof
A football team’s odds shorten sharply before kick-off. A screenshot circulates with the claim that someone knows the result. The price move is real; the explanation may be speculation.
Understanding suspicious betting starts with separating three things: a market changing its price, a monitoring service identifying an anomaly, and an investigation establishing misconduct. They are different stages, with different standards of evidence.
What the 2025 figures actually show
In its February 2026 release accompanying Integrity in Action 2025, Sportradar reported monitoring more than one million events across 70 sports and identifying 1,116 suspicious matches, down 1% from 2024. Football accounted for 618 of those matches, followed by basketball with 233. These are the company’s monitoring findings, not a count of court-proven fixed matches. Source: Sportradar’s 2025 findings.
The distinction matters when reading a headline about match-fixing “declining.” A change in detected cases is not a direct measurement of all manipulation worldwide. It also should not be used to estimate the safety of a particular match or competition.
What monitoring services can see
A public odds chart is only one part of the picture. Sportradar describes a system combining bookmaker prices, account-level staking data from participating operators, operator alerts and intelligence. Its published description also makes clear that betting integrity specialists assess patterns and report concerns to partners. Source: Sportradar’s detection methodology.
That helps explain why an observer with screenshots cannot reproduce an integrity investigation. The screenshot may show a price and a time. It usually cannot establish who placed the bets, how the activity compares with those accounts’ histories, or what additional information an operator holds.
It is useful to ask separate questions: Did the price really change? Was the activity unusual in context? Is there evidence connecting it to misconduct? Answering the first does not answer the other two.
Where AI fits — and where it does not
Sportradar reported a 56% year-on-year increase in suspicious matches flagged through AI analysis in 2025. That figure describes AI-assisted detection in its system. It does not mean that match-fixing increased by 56%, especially when the same release reports a small decline in total suspicious matches. Source: Sportradar’s AI monitoring findings.
The practical distinction is between identifying a pattern worth examining and establishing what caused it. An anomaly detector can help direct attention; its output does not, by itself, explain a person’s intent. Our reading of the published methodology is that automation expands the scope of monitoring while expert assessment remains part of reporting concerns.
An odds move with an ordinary explanation
Consider this hypothetical example, not a reported incident. A team’s decimal odds move from 2.10 to 1.80 after the opposition announces that several starters will miss the match.
The quoted implied probability changes from approximately 47.6% to 55.6%, using 1 divided by decimal odds. Those figures include the effect of bookmaker pricing and are not independently verified chances of winning. You can reproduce the calculation with our odds probability calculator.
Someone seeing only the before-and-after prices might describe the move as suspicious. Someone who also has the team announcement has an obvious alternative explanation to examine. Even without an announcement, missing context is not evidence of manipulation.
For a useful comparison, record the exact market and line, bookmaker, timestamp and whether the quote was pre-match or live. Comparing a pre-match total with a different live total can produce an apparent discrepancy that says little about the underlying event.
A real case: an alert followed by an investigation
The NBA’s April 17, 2024 announcement about Jontay Porter illustrates the difference between a betting signal and an established league finding. According to the NBA, licensed operators and a monitoring organization brought suspicious bets on his performance to its attention.
The league subsequently reported findings that included disclosure of confidential health information and limiting participation for betting purposes. It also said an $80,000 proposition parlay associated with the March 20 game was frozen and not paid out. Porter was banned from the NBA. These are findings attributed to the league’s announcement, rather than conclusions drawn from an odds chart. Source: NBA’s original announcement.
The lesson is about the evidence chain: unusual activity prompted scrutiny, and the investigation considered conduct and information beyond the price itself.
How to read the next “suspicious betting” story
Before treating a headline as a finding, check what the original source actually says:
- Identify the stage: a rumour, an operator alert, an investigation or a published decision.
- Check the denominator: detected cases out of which monitored events, during which period?
- Separate the measures: AI flags, suspicious matches and sanctions count different things.
- Look for context: the exact market, timing and any relevant public information.
- Keep uncertainty visible: if the underlying data is unavailable, the cause may remain unknown.
An odds move tells you that the quoted price changed. Calling a match manipulated requires additional evidence. For anyone working with betting data, preserving that distinction is more useful than treating every dramatic chart as a hidden message.