In NT 7 there were some additional features that allowed users to set delays in the live SIM that were not included as part of market replay. I think by default there were a few delay settings that maybe netted out to a 1 second lag, but this could be adjusted of course, so in theory you could get the delay between market replay and live SIM to only have the lag from your internet speed / data feed.
In NT 8 I am not aware of any user configurable setting for delays so my first question pertains to this.
1. Is there any sort of delay setting in Live SIM in NT 8 like there was in NT 7 and users just can't adjust this, but it is in place a permanent default? If so, what is this setting, just so I know the assumption.
Beyond the potential delay setting that may or may not exist, these are the other things I am aware of that may come into play to explain some of the differences between Live SIM and market replay.
2. Market replay files are ran from your computer to the simulator engine fairly instantly, so there is little to no lag between the price change / volume change events and when your code fires from various event handlers. In Live SIM, the data is coming in from a live data feed so depending on how fast your interest speed is, there can be a lag in moving between your strategy and the data. So the main difference here is just internet lag.
3. Beyond internet lag there is a latency between interacting with moving data that will never stop, and the downloaded data that may be read line by line in tandem with your code. I suspect that the similar engine will execute your ninjascript in sequence with the downloaded data so that it almost behaves like pausing the data feed, reading and reacting to your script then unpausing the data feed and not getting behind at any point. With a live data feed there is no option to pause or slow down the data, so I think there will just be a natural lag there even if internet speed is nearly perfect with virtually no latency.
4. There may be some differences between data providers. Kinetic vs. CGQ vs. Rhythmic all may have various differences. But I understand that Kinetic and CQG both have unfiltered data that is virtually identical, so I doubt there is any material differnce really. But maybe this is a bigger difference between other data providers. Specifically I would just like to know if CQG and Kinetic are in the same ball park in terms of unfiltered data, and should in theory be similar.
5. I am under the impression that both market replay and Live SIM use the same simulation engine, so there is no difference in how fills are treated in one vs. the other. I would not expect there to be any differences here that account for performance differences.
6. Settings: I gather that to get a truly strong apples to apples comparison the best setting to use would be a high tick setting such as 25, 50, 75 for the primary bar, with the added tick replay for additional granularity in sequencing the level 1 events. I just want to make sure I have this right. The other option would be to use the high resolution fill rate with 1 tick on each (market replay and live SIM). Let me know which would produce a more consistent - accurate comparison. (Tick Replay, or High Res @ 1 tick)
So this is kind of it, as far as I understand it. Really the only main differences are just lag from the internet and lag from the delay between sequencing data with a live feed running vs a static file that is downloaded. Do I have this about right?
I am in the process of reconciling some Live SIM trades vs. Market Replay and just want to get all the assumptions on the table to try to explain any differences that may arise. Please let me know if I have captured everything or if there is anything else I should take into consideration.
Thanks,
Ian
Comment