Hi all,

I encountered an issue which really scared me.

In order to conduct some other experiment, I wrote a simple strategy which just writes to a file all ticks (so on-tick writes values from the current tick: time, price, volumes). I have run this strategy localy for several hours.
Then I wrote another strategy that simply counts number of ticks in each minute and also reads those ticks from my file, and counts them per minute as well (matching the same minutes).

Now when I run this "reading" strategy on historical tester, I discovered that numbers of ticks in historical data (per minute) are much higher than what I "recorded" myself !!!
The differences are in almost all minutes and reach sometimes even 50% more.

I tried this on few periods (2 or 3 days long). I'm also aware of the tick execution policy (if execution time is longer that 1 second, some ticks are dropped), but I wrote my code with good buffering, so the on-tick execution is very quick.

Please someone explain this, because it is really worrying...