Wait for some time for reconstructing large order, so that orders are correctly put back togther

[Jan123]

I noticed that Bookmap does not correctly reconstruct the tape, even though it claims to have 100% accuracy. I did a thorough comparison with other software and the actual Rithmic data feed that has the aggressor ID. In fact large trades often get fragmented because part of the order arrives a tiny bit late, because the computer can't catch up. That is especially frustrating given that the Aggressor ID is submitted with Rithmic. In addition the exchange execution timestamp are NOT used it seems like for order reconstruction.

So even if the exchange execution times for the large order line up to the millisecond, Bookmap will think those are two different order, if one part arrives even 100 msec late at the computer. 

I have talked about this issue with support, and they say that they do this in order to not "rewrite" history. Yet, I rather have my large orders 100 msec lagging instead of having them fragmented into two.

Please consider having a parameter that allows people to define an "integration window" for summing up executions that belong together as one trade!

The pictures below show the Rithmic tape on the left, which has execution timestamps + Aggressor ID, which is enough to prefectly reconstruct large orders. In the middle is teh bookmap times and sales filtered for orders larger than 30 contracts. To the right is the Motivewave times and sales that combines trades correctly (Except for market-limit orders where part of the order is left as limit order in the book. Market limit orders are for example buy limit orders placed at the ask (so on the "wrong side of the spread"), with part of it being executed immediately, taking the full liquidity, and the rest staying as passive limit order in the orderbook at the ask. Please see "market-limit orders" on the CME website for more details). 

 

[helminiak247]

Why has Bookmap support never addressed this?  Seems like a no brainer. I've been using reconstructed tape with IRT and IQ_Feed. If other data feeds can provide the info needed it "seems like" an easy fix
-jjh