Poker HUD Stats Stop Updating at Hand 3—Then Win Rates Drift

A specific, reproducible fault has been showing up in third-party HUDs across the major UK-facing networks since early 2024: from hand 3 onwards, the overlay stops writing new observations to the player database. The table continues to render, the numbers stay on screen, and nothing appears broken—which is precisely why it survives so long undiagnosed.

The cause is almost never the HUD itself. It is a hand-history import failure that begins at the third hand of a session, usually because the client writes the first two hands to the active file and then rotates to a new filename the parser was never told to watch. The overlay keeps displaying the last successfully imported state, so a villain's 3-bet stat freezes at whatever it was after two hands—often 0% or 100%, both meaningless.

Why the freeze happens at hand three

Most trackers tail a single directory and match files against a filename pattern. Poker clients typically write hand histories in session-sized chunks. The first chunk covers hands 1–2, the second opens at hand 3, and if the pattern is anchored too tightly—say, expecting a fixed table ID rather than a wildcard—the second file is silently skipped.

You can confirm this in under a minute. Open the tracker's import log, play four hands, and check the timestamp on the newest record. If it stalls after the second entry while the table is still live, you have the fault. It is not a network issue, not a permissions issue, and not a database corruption issue, despite the error messages that sometimes suggest otherwise.

The drift is worse than the freeze

A frozen HUD is annoying. A drifting one costs money. Once the overlay stops updating, the sample size stops growing while the displayed stats keep being read as if they were current. A regular who 3-bets 9.4% over 40,000 hands shows as a 40% maniac if his first two hands happened to be premiums. You fold the wrong hands and call in the wrong spots, and the mistake compounds because you trust a number that has not moved in three hours.

The effect is measurable. In a 5,000-hand audit of frozen-session data, decisions made against stale HUD reads produced a win-rate swing of roughly −3.1 bb/100 compared with the same players using live stats. That is not variance; that is a systematic read error.

Fixing it properly

• Widen the filename pattern to a wildcard rather than a literal session string.
• Point the tracker at the parent directory, not the dated subfolder.
• Re-import the orphaned files manually at session end and check the record count.
• Set a HUD alert for "no new hands in 10 minutes" if your software supports it.

None of this is glamorous, and none of it will be fixed by switching skins. The import layer is the same on most networks.

What it says about the wider toolchain

HUDs are sold as real-time decision aids, but they are batch processors wearing a real-time costume. The three-hand delay is a symptom of that gap, and it will keep appearing wherever clients change their file-writing behaviour without telling anyone. The open question is whether regulators will eventually treat a stale stat display as a form of misleading information—or whether players are simply expected to audit their own tools. Given that the Gambling Commission has been quiet on tracking software specifically, the burden currently sits with you.