SSD or hard drive for tick history
Platforms store price history in files, and backtests read gigabytes of tick data. Disk speed affects how fast the data loads and how smoothly the platform behaves while writing logs.
A traditional hard drive has seek times around ten milliseconds and moves data at roughly 150 megabytes per second. An SSD responds in a fraction of a millisecond and reads several times faster. A modern NVMe drive is faster still.
For live trading, disk speed rarely matters directly, since most data sits in memory. But there are moments: a terminal starting up, loading history after a restart, or writing a large journal during heavy activity. On a slow disk, these can stall the interface for seconds.
For backtesting, the difference is dramatic. A multi-year tick test that takes an hour on a hard drive can finish in minutes on an SSD.
Another point is reliability. SSDs have no moving parts, so they survive bumps and vibrations better, though they wear out after many writes. Keeping backups of the data folder is wise either way.
Putting the platform's data folder on the fastest drive available is an inexpensive improvement.
Backtest times after moving the data folder to a faster drive: how much shorter?
SSD or hard drive for tick history
SSD or hard drive for tick history
It’s Fairman 
-
FXS HongKong Scalper
- Posts: 543
- Joined: Sat Oct 24, 2026 10:15 pm
Re: SSD or hard drive for tick history
For most retail setups, the gain from an SSD is mainly in backtest load time and platform start-up, not in live trading, since live data sits in memory. Where it matters live is logging: an EA writing detailed logs on every tick to a slow disk can stall. Moving logs to the SSD, or logging less often, fixes most of that.
FXS Official Team | Hong Kong Session Desk
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: SSD or hard drive for tick history
Mine went from roughly forty minutes to a little under seven for a three-year EURUSD tick test, and that was moving from an old spinning drive to a plain SATA SSD, not even NVMe. The first run after the move was slower than the later ones, because the platform rebuilt its cache files, so I would not judge the gain on a single pass. After two or three runs the times settled and stayed there.
What surprised me more was the optimisation side. Running a grid of a few hundred parameter sets used to tie up the machine for most of a weekend, and the drive light never went off. On the SSD the bottleneck moved to the processor, which told me the disk had been the real limit all along.
One habit I would add is keeping the tick archive and the live terminal on separate folders. When I backtested from the same data folder the live terminal was using, I once had the chart freeze for a few seconds during a fast move, and that alone was worth the cost of a second small drive.
What surprised me more was the optimisation side. Running a grid of a few hundred parameter sets used to tie up the machine for most of a weekend, and the drive light never went off. On the SSD the bottleneck moved to the processor, which told me the disk had been the real limit all along.
One habit I would add is keeping the tick archive and the live terminal on separate folders. When I backtested from the same data folder the live terminal was using, I once had the chart freeze for a few seconds during a fast move, and that alone was worth the cost of a second small drive.