Hello, As I continue to develop my automated futures trading system, I keep having second thoughts about the way I maintain the orderbook inside of my program. The thing that has been bothering me is what I do when a trade posts. I assume, hopefully correctly, that on GLOBEX futures exchanges a trade is posted before the depth of market book is updated (I am also using NxCore and assuming that they forward data from the exchange in a FIFO fashion). I.e if there are 30 contracts being offered at a price of xxxx, and a 5 contract transaction at xxxx occurs, I assume I will receive a message saying a trade of 5 contracts just occured at xxxx, before I recieve a message saying the the offer at price xxxx is 25 contracts. At this point, should I manually adjust my internal book to say 25 contracts are available? Or should I wait for the exchange to tell me there are 25 available? I understand this is a bit of a judgement call and is application specific, and this becomes a more important consideration for low latency systems. The processing time to do it myself isn't really an issue, its a fear of doing it myself and being wrong. I guess what I am really asking is how long does CME/NYMEX wait before they update their depth of market after a trade takes place? I assume the the GLOBEX DOM updates are event driven and not time based? Is it even safe/reliable to update the book on my own? Thanks in advance, Brad
trades should not affect your order book, your order book is what the term says, it contains orders. You have no idea at that point when the order takes place whether the trade was the only incident that affected your order book or whether others cancelled or modified their offers. So, you should wait until you get the update from your data provider.
This makes sense, no sense trying to infer the state of the market when there are too many degrees of freedom and limited visibility. Thanks for the input.