I manual trade on 5 seconds TF and I was experiencing lag in candle formation last week also!

Price / the candle going from barely any body to 2 bars formed / showing up at once, both with bodies of several points!

Very much Off Topic 🙂 but FYI :

inverse wrote:

Shouting out to @PeterSt here because I know he mentioned early days v12 when it started to get noisy he too would see interop problems that related to ram and leaks knocking on.

It is only two weeks ago that I gave up on it all, and minimized my number of charts in my main account, so that it would reasonably boot sufficiently quick (but always two times slower than V11.1). Otherwise “learn to live with it”. So I am now at V12 everywhere.

The backtesting is okay these days, and I can last a whole day (of 18 hours) of that. I must admit : I expanded my memory from 32GB to 64GB.

Another FYI, and somewhat more On Topic : With 4 accounts on the lowest setting of the Push etc. data, I’m in my own way with sending out data and imply at least ~200ms of lag towards Chicago. I would gain that 200ms when I would minimize my total setup to 1 account only and close the other platform instances (that one left still holding back, obviously). I regard this a general PC bottleneck and therefore would take it that towards me (receiving data in PRT) I deal with the same lag which I estimate 50ms.
This can’t be a big deal.

Of far more importance, and now totally On Topic, I use 1 second systems and only two days ago had to solve the issue of lacking 1 second bars. Please notice that this is IB – I did not check for that in IG. These 1 second bars lack – out of all ! – not long after the opening, say 90 minutes later. At ETH this is way way worse and I can’t tell why this is. Crucial regarding this fact is that I make use of parallel (payed) data streams, and there is no lack of data at all. Of course, right after the opening (or when News) it is 100 times more, but there’s always a stream of data on-going, although at quiet hours this comprises of say a 20 fold of quotes data compared to (changed) price data (I talk Nasdaq now, S&P is say 10 times less in the first place).
One could say that I coincidentally ran into this (issue ?) only the other day, because I was changing a Market Order system into a Limit Order system and miraculously found trades all over the place, PRT not even showing the 1 second bars. This is easy to show – I could do that later.

Because I needed the (1 second) bar count, I even needed to make something “odd” for that, because normally it is throughout most of the day not possible to count the “seconds” reliably.

Yes, I only found this a couple of days ago, but this does not tell a thing because before that I didn’t look at matters this way. It probably always has been so ?
To me this comes across as if PRT first collects say 10 price changes before she hands out a new 1 second bar and otherwise just skips it.

Lastly for now, I feel this is not related to what you see or saw, (Trader)Glyn, or else I would have responded similar right away. Still, when you’d observe the building of the bars live and you’d notice a change compared to before, then why would it not be a coincidental very same. I actually never look at the building of 1 seconds bars (my charts comprise of 10 seconds, no matter the AutoTrading uses 1 second).

Regards,
Peter

Because V13 is around the corner plus by now I learned to live with V12 via the other accounts.

Net, I think V12 has the edge by quite some distance. It’s only real downside is that it is more slow / sluggish. But that solves itself by lesser charts (still more sluggish, but one gets used to that – because doable).

I also consider the advantages. Example : what if they would take out the shown profit per trade in the charts, so it will be as fast as V11 ? … I would choose for V12 …

image_2024-10-27_123613270.png image_2024-10-27_123659457.png