Score and confidence answer different questions
The score describes how the available signals combine under the published rules. Confidence describes the amount, coverage and freshness of that evidence. Always read both: an apparently reassuring score based on a limited sample is not the same as a well-supported assessment.
Open the explanation to see the contributions behind the total. An account’s score can change when providers return new information or previously missing data becomes available. A change does not necessarily mean the player’s behavior changed.
What evidence is considered?
The assessment considers available ban records, account age and activity, performance context, competitive history, inventory evidence and eligible community feedback. The methodology page publishes the categories, caps and limitations of the current algorithm.
Ban records need source and timing context. A Steam game ban is not automatically proof of cheating in CS2, and a platform restriction is not interchangeable with every other ban type. Performance outliers are contextual signals, not a substitute for verified records.
Privacy and inventory value
A private inventory or unavailable account field is missing evidence, not proof of suspicious behavior. The score does not treat a decision to keep data private as a ban or a performance anomaly. Read coverage before comparing profiles with very different amounts of public information.
Inventory value is a limited contextual signal. An expensive inventory cannot establish honest play or cancel a confirmed serious ban. A small or empty inventory does not establish cheating. Estimates depend on the items and pricing information that were available.
Use the explanation, not just the number
Start with the strongest source-backed evidence, then examine its age and relevance. Check the match sample and rank history when reviewing statistical anomalies. Community feedback is separate from verified ban records and has limited influence.
Do not use a score alone to accuse a player. Follow the linked sources, review the published limitations and account for what is missing. The full methodology explains exactly how the current model combines these signals.