A PC clock that runs even half a second off can quietly mess up a scalper's records, signals, and news timing.
The effect shows up in several places. A strategy triggered at "8:30:00" on the local clock fires half a second early or late relative to the broker's server. Execution logs compare the local send time to the server's confirmation time and produce nonsense, such as fills that appear to happen before the order was sent. News-based entries meant to wait a certain number of seconds after the release use the wrong baseline.
Windows syncs its clock only occasionally by default, and virtual machines can drift noticeably, especially under load.
The remedy is easy. Use a proper time synchronization tool that checks against a reliable source frequently, and verify the offset between the PC and the broker's server time. Many platforms display server time in the market watch window or status bar.
Once time is aligned, execution analysis becomes meaningful. Latency numbers computed from synchronized clocks can be trusted, and weird results disappear.
How far apart are the PC clock and the broker's server time right now?
Clock drift: the computer's time is off and the logs prove it
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
-
FXS Prop Desk
- Posts: 1690
- Joined: Wed Oct 07, 2026 9:52 am
Re: Clock drift: the computer's time is off and the logs prove it
Half a second is enough to make an execution log lie. If local send time and server fill time come from clocks that disagree, measured latency can even come out negative, and any "slippage vs. delay" analysis built on it is junk. Syncing to an NTP server and logging the server timestamp from the trade confirmation, not the PC clock, keeps the audit trail defensible if a dispute with the broker or a prop firm ever comes up.
FXS Official Team | The Idea Lab & Strategy
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: Clock drift: the computer's time is off and the logs prove it
I checked mine after reading this and it was off by a little under a second against the broker's server time, which surprised me because Windows claimed it had synced that morning. I only found out because my trade journal showed a couple of entries on the payrolls release that were apparently filled before the number came out, which obviously wasn't true.
What fixed it for me was installing a small time sync tool that polls a public time server every few minutes instead of relying on the built-in schedule. Since then the offset stays within a few milliseconds. I also added a column in my journal for the server time from the trade confirmation, so if the two clocks ever drift again I can see it straight away instead of finding out weeks later.
The virtual machine point is real too. My cheap hosted server drifted far worse than the home PC whenever the provider was busy, so I set the same sync tool up there and check the offset every Monday before the session starts.
What fixed it for me was installing a small time sync tool that polls a public time server every few minutes instead of relying on the built-in schedule. Since then the offset stays within a few milliseconds. I also added a column in my journal for the server time from the trade confirmation, so if the two clocks ever drift again I can see it straight away instead of finding out weeks later.
The virtual machine point is real too. My cheap hosted server drifted far worse than the home PC whenever the provider was busy, so I set the same sync tool up there and check the offset every Monday before the session starts.