How Detection Works
A tracker that names accounts owes its readers its error bars. This page documents every signal behind a ban call, what each is worth, and what the system cannot prove.
The detection mechanism
There is no official ban feed. EA publishes no list of banned accounts, and the fields in their API that sound like ban flags are unreliable — we have verified accounts carrying a "ban reason" value while demonstrably active on the ladder. So this tracker doesn't read a ban flag; it watches the ranked ladder itself.
The system polls the top-750 ranked ladders on every platform around the clock. When an account holding Master+ RP disappears from the ladder — while the RP is still on the account and the player is offline — that is the shape EA's systems produce when an account is banned. Detection fires on that transition, captures the account's RP, ladder position, and level at that instant, and writes a timestamped ban event. The same watch in reverse is how unbans are detected: a banned account reappearing on the ladder.
Inference has error bars. A ranked split reset, a tracking gap, or an account simply going dormant at the wrong moment can mimic parts of the banned shape. Rather than pretend otherwise, every detection carries a graded evidence score — documented below — and reversals are published as a standing reversal rate.
The evidence score — 100 points across seven signals
- 30ptsRemoved from ranked ladderThe primary signal. A top-750 account that vanishes from the ladder while still holding Master+ RP, and is offline, matches the shape EA's systems produce when an account is banned. The tracker also writes a detection marker at the moment this fires, so the observation survives after the live state ages out.
- 25ptsDiscrete ban event recordedDetection fired while the system was watching and wrote a timestamped ban event — with the RP, ladder position, and account level captured at that moment. Event-backed bans are also the only ones whose reversals we can track, which is what feeds the published reversal rate.
- 15ptsPre-ban ladder position capturedThe system recorded exactly where the account sat on the ladder immediately before it vanished. This rules out the main false-positive class: an account that merely looks removed because it never held a ladder position in the current split.
- 10ptsEA ban timer presentEA's API reported an actual ban countdown on the account at detection time. Strong when present, but most real bans are permanent and show no timer — so its absence proves nothing, which is why it's worth 10 points and not 40.
- 0–10ptsTime gone without reappearingEvery day the account stays off the ladder is evidence the removal wasn't a glitch: 1 point per 3 days, capped at 10 (a month gone scores full marks). This is the one component that moves over time — and it moves to zero the moment the account comes back.
- 5ptsPrior ban on recordThe account has been through a ban/unban cycle before. Repeat detection on the same account makes a tracking artifact much less likely.
- 5ptsName-change historyTwo or more recorded names on the account. Name churn correlates with ban evasion; it is deliberately the smallest weight because plenty of legitimate players rename too.
Score → confidence tier
Multiple independent signals agree — typically ladder removal plus a recorded event plus the pre-ban snapshot. The strongest call the system makes.
The core removal evidence exists but corroboration is partial — e.g. detected via ladder removal with a pre-ban snapshot, but no discrete event row, or the account hasn't been gone long.
Bare heuristic territory. The account matches the removed shape but little else corroborates it — this could still be a split-reset artifact or tracking gap, and we say so.
What this system cannot prove
Why an account was banned. Ladder observation shows that an account was removed, not whether it was cheating, boosting, or something else entirely. We never claim a reason.
Anything below the top 750. Detection covers the ranked ladders we watch. Bans outside them are invisible to us — which is why our counts on the reconciliation page are always far below Respawn's game-wide totals.
A perfectly calibrated reversal rate. Reversals can only be tracked for event-backed detections (the account has to come back onto a ladder we watch), so the published rate is measured on that subset. We state this rather than hide it.
If a detection about you is wrong, you can contest the listing. Confirmed errors feed the reversal rate — they are not quietly deleted.