The way I read it for loading historic data it averages the data across the second (the resolution of the database). If this is the case they have missed an opportunity. If I have mis understood and that is nit the case i will be delighted, i have been on at them about this for years!
As far as I can tell the historic database is still one second resolution. That is the problem and always gas been. You need precise time stamps and/or properly sequenced data across multiple data streams. Correctly dealing with live data just doesn't cut it you must be able to refresh from historic data in the database.
Have you been testing it? Are they changing the database resolution (that would be big hurrah for all sorts of reasons)
Edit: this is the worrying bit below......(though if they have changed the database resolution to milliseconds or better it would probably be an OK approximation. If it still one second resolution it's no better than what we have now, which just doesn't cut it)
Important information: The logic described above is for data collected in real-time. The following logic is used for historical plots. Example: Assume there are 3 historical trades, 9 historical bid quotes and 12 historical ask quotes. All of them have the same timestamp. To find out which historical asks and bids were used to generate the trades, they are all divided into 3 equal parts (because there are 3 trades). the first “batch” there are 3 asks, 1 trade and 4 bids. The most recent ask/bid matching the trade price is considered to be the price where the trade was made. If ask price is considered for the fill, then the trade’s volume goes to BUY volume. If bid is considered for the fill, then the trade’s volume goes to SELL volume..
Last edited by NickA; November 11th, 2012 at 02:45 PM.
Reason: Added quote from MC wiki
I haven't compared yet the differences between historical data and real-time collected data. I'll see if this week I'm able to make a couple of comparisons. With volume delta bars it will be easy to compare real-time collected data and historical data.
Again, this version is still in an "alpha" stage, where many of the new features are still being developed. You're a bit limited regarding some tests.
Let's wait for the "beta" version for a proper testing environment.
If I become half a percent smarter each year, I'll be a genius by the time I die
Last edited by arnie; November 11th, 2012 at 05:39 PM.