Players running Holdem Manager 3 or PokerTracker 4 on Windows 11 have reported a repeatable pattern: stats stop updating somewhere between 38 and 42 minutes into a session, while the hand history import itself keeps running. The table stays live, the client stays connected, and the HUD panels simply freeze on numbers from the last hand they processed. Restarting the tracker clears it, which is exactly why the bug has been slow to get taken seriously — the fix feels like a workaround, so nobody files the ticket.
What Actually Freezes
The failure isn't a crash. It's a stalled refresh loop. In both HM3 and PT4, HUD panels poll the database on a timer — typically every 1–3 seconds depending on configuration — and redraw only when the returned stat set differs from what's displayed. When that poll thread blocks, the panels hold their last rendered values indefinitely.
Two things distinguish the freeze from a normal lag spike:
- The import counter keeps climbing. Hands are being written to the database; the HUD just isn't reading them back.
- New tables behave normally. Open a fresh table after the freeze and its HUD populates correctly for roughly the same 40-minute window before stalling in turn.
That second behavior points away from a memory leak in the display layer and toward something session-scoped — a timer, a connection pool entry, or a cache invalidation that never fires.
The 40-Minute Marker and Windows 11
The timing is too consistent to be coincidence. Community threads on the HM3 support forums have logged the freeze at a median of 40 minutes 12 seconds across 61 reports since early 2024, with a tight cluster between 38 and 42 minutes. That narrow a band usually means a hard-coded interval or an operating-system-level timeout.
Windows 11's power management is the leading suspect. The "Connected Standby" and modern standby behaviors introduced with the 22H2 update can throttle background threads in applications the OS doesn't classify as active foreground processes. A poker tracker polling a local PostgreSQL database looks, to the scheduler, like background I/O. Forty minutes is a plausible window for a power-plan heuristic to decide the process is idle-adjacent and deprioritize it.
The counter-evidence: users on Windows 10 report the same freeze, though at lower rates — roughly 1 in 5 versus 3 in 5 on Windows 11. So the OS may be an accelerant rather than the root cause.
What to Check Before You Blame the Tracker
Before reinstalling anything, rule out the cheap explanations:
- Database size. Freezes cluster heavily above 4 million hands in a single database. Vacuum and reindex, or split by site and year.
- NoteCaddy. If you're running it, disable it for a session. Its badge processing competes for the same database connection pool.
- Antivirus real-time scanning. Windows Defender's default scan of the PostgreSQL data directory causes exactly this symptom on some machines. Add an exclusion for the data folder and the tracker's executables.
- HUD update interval. If you've set it below 1 second, you're hammering a single connection. Drop it to 2 seconds and see whether the freeze moves.
None of these fully explains the 40-minute regularity, but each one changes the timing, which is diagnostic in itself.
The Unanswered Question
What's missing is a vendor statement confirming the mechanism. Both HM3 and PT4 have shipped point releases since the reports began, and neither changelog mentions the HUD refresh thread. Until someone instruments the poll loop and captures a stack trace at the moment of stall, the 40-minute figure stays a correlation.
For now, the practical read for US players multitabling on a single machine: restart the tracker between sessions rather than leaving it running for eight hours, and don't assume a frozen HUD means your hands aren't importing. Check the import count before you panic-fold your way through a session on stale stats — and if you're playing long sessions, that's one more argument for a hard stop well before the two-hour mark anyway.