My opening post mentions data from NinjaTrader Market Replay.
My trading platform is Microsoft Visual Studio. I have my own proprietary software, so my focus is on the data.
I have recorded live data, and its different to the data provided by NinjaTrader/CQG and IC. My problem is I need several years of data and third party data is very thin on volume. Volume in CQG/NinjaTrader and IC are similar but the timings are different, and I'm not sure if their volumes describe trades or lots?
CQG/NinjaTrader is the more granular dataset, but their milliseconds is always 0, which is not believable. The problem with rounding the times is that it bunches trades together. You can replay the rows one-by-one to simulate the milliseconds, but a bigger problem is that both IC and CQG/NinjaTrader can contain gaps of 5 seconds between trades during peek times - on major currencies that's alarm bells!
IC makes this worse because it sums together all lots/trades in the same moment so you can't even simulate the missing milliseconds. Instead the price jumps without trades taking place. That can happen in life, but if you replay a live ladder recorded with a video recorder you will see that it does not happen in the way IC data suggests. When they round the times and block together volumes, we assume no other data is being chopped - and I'm no longer sure that's a safe assumption.
My live recordings contain more trades and more volume than data from IC/CQG/NinjaTrader - how is this possible?
Last edited by London Trader; April 9th, 2016 at 10:51 AM.
It took a while, but my data feed appears to contain x2 volume and x10 the frequency. My live data feed might be wrong: Perhaps the provider is munging together multiple feeds without filtering, causing me to receive duplicates? This working hypothesis is based on data appearing to be both smoothed and amplified.
Any thoughts would be appreciated!
I cannot restore my live feed to IC equivalent because doing so would require my system to see the future. As a quick hack, I've multiplied the volume on historic data by a factor of 10. Good/bad?
Last edited by London Trader; April 10th, 2016 at 12:51 PM.
Hi there.agree it is very good. However have you checked how the data looks during the days index fell in august 2015. When I looked at some of those days it seemed that there was a lag. Since I was not there live during those days...not sure if the data was in spurts....or probably market replay accurately captured the same? Any comments will help as I wanted to see those days replayed...but I would not do any sane analysis as data seemed lagging.