The chart is not a record of every tick
Posted: Fri Oct 30, 2026 8:00 am
The chart is not a record of every tick
MetaTrader and similar platforms build candles from the ticks they receive. They do not necessarily receive every tick the broker generates. Under load, the server may send a reduced stream. The platform may also drop updates when it is busy.
This leads to charts that differ across platforms and across time. Two terminals on the same account, started at different times, might show slightly different highs and lows for the same candle.
For a scalper on a one-minute chart, it matters. A level that looks like a clean double top on one terminal might be a slightly different pair of highs on another.
Backtests built from the history inherit the same gaps. Historical data downloaded from the broker may be a compressed version of the actual price path, with the highs and lows right but the intermediate ticks missing or interpolated.
Tick-level strategies suffer most. A model that relies on the sequence of ticks within a candle has no real data to test against.
Better to record live ticks directly using a script and build a private dataset. After a few months, the difference from the platform's history becomes a revealing exercise.
How different is the self-recorded tick data from the platform's historical data on the pair being traded?
MetaTrader and similar platforms build candles from the ticks they receive. They do not necessarily receive every tick the broker generates. Under load, the server may send a reduced stream. The platform may also drop updates when it is busy.
This leads to charts that differ across platforms and across time. Two terminals on the same account, started at different times, might show slightly different highs and lows for the same candle.
For a scalper on a one-minute chart, it matters. A level that looks like a clean double top on one terminal might be a slightly different pair of highs on another.
Backtests built from the history inherit the same gaps. Historical data downloaded from the broker may be a compressed version of the actual price path, with the highs and lows right but the intermediate ticks missing or interpolated.
Tick-level strategies suffer most. A model that relies on the sequence of ticks within a candle has no real data to test against.
Better to record live ticks directly using a script and build a private dataset. After a few months, the difference from the platform's history becomes a revealing exercise.
How different is the self-recorded tick data from the platform's historical data on the pair being traded?