Backtests on scalping systems stand or fall on tick data, and free sources vary a lot in quality.
Several kinds exist. Broker history exported from the platform is convenient and matches the trading environment, but it can contain gaps, and some brokers supply only minute bars. Third-party archives offer long histories, often built from a single liquidity source or interpolated from lower resolution data. Data collected live from a server gives the closest picture of real spreads but only goes back as far as collection began.
Pitfalls are worth knowing. Spreads in many historical files are fixed or averaged, so a test never sees the widening around news. Timestamps may be in a different time zone from the platform, which shifts every session-based rule by hours. Duplicate or missing ticks, flash spikes that never happened on the actual broker, and weekend gaps handled oddly can all skew results.
A cheap sanity check helps. Compare a few known volatile moments, such as a major news release, in the dataset against the broker's own chart. If the spike shapes and spreads look nothing alike, the dataset isn't a good stand-in.
Another check: run the same test on two sources. A system that works on one dataset and collapses on another is leaning on data quirks rather than market behaviour.
Even good tick data is only history. Slippage and requotes in live trading are a separate cost that no file reproduces, so some haircut on results is sensible.
Which tick source has held up best for you, and what flaw showed up only after a suspiciously good result?
Good wine after every good trading forex series
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm