Time lag executing and closing trades. It appears there could be a reported time lag of up to 43 Seconds, which could result in profit loss or non TP. Is this something that could be looked into or reduced?
This is so important. It basically means the current backtest results are inaccurate for some positions.
We are aware of the lag concerns - however I’d like to point out that a lag of 43 seconds is an extreme case and does not happen all the time.
There is work to be done and we are aware of it - both on our end and on Trading212’s end as well. We are working close with them to reduce the lag to a minimum amount.
ProQuant is growing faster than we anticipated and that is the main reason for these issues.
Rest assured we do not find this situation acceptable and it will be greatly improved soon.
Is there any live metric/statistic we can follow? “Does not happen all the time” is great, but how often does it happen?
All I can share right now is that we’re working hard on this with the Trading 212 team and it will be taken care of. Thank you for your patience and understanding!
is your computing power collocated with t212s? or are you actually running on their servers?
It’s not 43 second Nikolai, the lag is one trading period (plus 43 seconds). I have another post about this with screen shots. Basically for a strategy working on a 1 hour trading frequency the buy/sell signal will occur 1 hour later in real/simulated trading than in the backtest. The backtest time is the correct one when the trade should occur according to the indicators used. This needs addressing quickly
Sorry Nikola autocorrect changed you name to Nikolai in my previous post
Is that for all trading periods??? If it is, strategies trading in more than 1hr have a very serious problem indeed! Maybe that explains Psyco’s behaviour?
@theorib their response to that other post was that it is a labelling error and the only lag is the 43 seconds. Personally I don’t think that explains the difference between the backtest and live trading I see on some of my strategies but others it seems like it could be correct. No idea what is going on.
I don’t quite know what would be the best way to test for evidence of this. Are you analyzing the technical indicators the strategy should be using and seeing for that they should be triggering the strategy at a different moment in time?
@theorib I was looking at the charts and opens seemed to be too late but PQ have explained there are two factors contributing to this. The first, they say, is that the backtest displays the bar open time when trades are actually made at bar close. The second is that I was looking at strategies working with bollinger bands and had mistakenly thought that “bar opens above/below upper/lower band after opening below/above band” functioned the same way as other indicators which use crosses the zero//level line - it doesn’t - it is a period later by design.
And @theorib your other post would explain other instances.
Thanks, I guess we will have to sit and wait a bit on these ones while they work on it. Hopefully we’ll get a more robust PQ soon.