Moving averages on ticks
Posted: Sat Oct 31, 2026 6:52 pm
Moving averages on ticks
A moving average of the last 20 bars is standard. A moving average of the last 200 ticks is something different. It reacts to activity, not to the clock.
During a burst, 200 ticks may span three seconds, and the average tracks price closely. During a lull, 200 ticks may span several minutes, and the average lags far behind.
The adaptive behavior has a benefit: the indicator speeds up when the market does. It also has a flaw: tick counts are feed-dependent and include flicker, so the speed of the average varies by broker.
A tick-based average also makes backtesting harder, since it requires real tick data, and the result depends on how that data was recorded.
An alternative is to build the average on time-weighted prices, sampled every second, which removes the dependence on tick frequency.
Choosing between the two is partly philosophical. Do events or time define the market's pace? For scalping, event-based often fits better, as it adapts to volatility.
Either way, the number of ticks or seconds is a parameter that should sit on a plateau of good results, not on a peak.
Have tick-based averages been compared with time-based ones on the same strategy?
A moving average of the last 20 bars is standard. A moving average of the last 200 ticks is something different. It reacts to activity, not to the clock.
During a burst, 200 ticks may span three seconds, and the average tracks price closely. During a lull, 200 ticks may span several minutes, and the average lags far behind.
The adaptive behavior has a benefit: the indicator speeds up when the market does. It also has a flaw: tick counts are feed-dependent and include flicker, so the speed of the average varies by broker.
A tick-based average also makes backtesting harder, since it requires real tick data, and the result depends on how that data was recorded.
An alternative is to build the average on time-weighted prices, sampled every second, which removes the dependence on tick frequency.
Choosing between the two is partly philosophical. Do events or time define the market's pace? For scalping, event-based often fits better, as it adapts to volatility.
Either way, the number of ticks or seconds is a parameter that should sit on a plateau of good results, not on a peak.
Have tick-based averages been compared with time-based ones on the same strategy?