Marker Score

The Marker Score API OpenAPI specification is available for download here. Explore the API endpoints here.

The Marker Score API is part of Sportradar's Insight Tech Services and provides a machine-learning based assessment of whether an account behaves like a "marker" – a high-performing bettor who consistently places winning bets before markets close. This profiling helps sportsbooks identify accounts that systematically outperform the closing odds and pose elevated financial risk.

By using historic ticket data, the model outputs a score between 0.01 and 1.00, indicating the likelihood that a customer exhibits marker-like betting behaviour. The goal is to detect accounts that should be monitored or limited due to their strong predictive edge.

What it does

The Marker Score model profiles active accounts to estimate the probability of them being "marker" players. These are users who place pre-match singles with regularity, beat the market closing line, and demonstrate high win ratios over time.

The model incorporates a range of behavioural and betting indicators, including how early users place bets relative to the market close, how often they bet on flagged markets or outcomes, and how their bet performance compares to market movement. The score is used both for monitoring and for automated segmentation and intervention.

When it’s triggered

The model evaluates active accounts daily, focusing on users with recent betting activity. Once accounts reach a threshold of historical data, they are included in daily scoring routines. Based on the resulting score, the MTS system may automatically classify the account with a Marker tag. If the score later drops below a configurable threshold, the flag is reset.

Input requirements

To access the Bot Score through the API, the following parameters are required:

• bookmaker – the assigned Insight Tech or MTS bookmaker ID

• sub-bookmaker – the sub-entity of the bookmaker (use * to retrieve all sub-entities)

• customer - customer's uniqueID

Some endpoints also support range filters (lower, higher) or pagination using startKey. A full breakdown of all accepted parameters is available in the Endpoints section.

Output schema

A typical response includes the list of items, and each object under responseItem contains:

• srcId: compound source identifier for internal tracking

• mscore: A decimal value between 0.00 and 1.00 representing the Marker Score

• id: entity identifier

Interpretation guidance

• A mscore closer to 1.00 suggests high confidence that the account is a consistently sharp bettor who beats the closing odds.

• Operators may apply additional limits or enhanced monitoring to accounts with high Marker Scores.

• Scores are dynamic and update daily. Players can lose their marker classification if their score drops below the threshold.

• Absence of a score typically means the account hasn’t placed enough bets to be profiled yet.

The Marker Score can be used alongside other models such as Bot Score, Late Bet Score, and sCCF to gain a complete picture of account behaviour and risk.

For implementation instructions, schema references, and rate limits, consult the Endpoints page.