It would seem to me that if you are trading any type of strategy that is so saturated and efficient that a competitive edge of a millisecond or so makes any real difference, there couldn't be much there anyway. Also, I would think that the thread originator who has been system trading for five years (back in the good days even) ought to be financially capable of doing it on his own(?), his own way, instead of relying on unnamed brokers.
Spoofing would be my guess - need to pull that size as soon as a bid ask strike triggers the pull. Latency of data coming from the broker would be an issue in that case. Just a guess.
The CME is about to aggregate its tick data. Unless you own a seat at the exchange, this talk about milliseconds is pointless.
What do you mean by aggregating? Are they going to collect and only dissiminate it every 1/2 second or so? That pretty much sucks. I understand where they are coming from though.. having real capacity issues. I've heard stories of some players sending hundreds of thousands of orders, canceling, and actually having only 1 out of that actually being executed. A simple way to solve that is add another fee that would only kick in after several hundred/thousand orders per day.
I doubt it. Spoofing is badly overrated as a tactic. Who would they be 'fooling', a few one lot traders? Only think I can think of is some kind of micro-arb, or maybe a trade initiated based on short term momentum, needing to take out the current price, not the next level.
Regarding programmers fixing latency: The risk/reward for the programmer is totally wrong to expect any kind of optimal solution. Usually there are only a handful of people who even complain about the latency issue unless it is REALLY BAD, and most of those are complaining about things that are out of the broker's control ( ISP speed, Internet delays, etc. ). The reward to the programmer for "fixing" any minimal latency at the brokerage is usually NOTHING. Most people won't notice any changes for the better. A few customers might notice an improvement, but these will be the same anal twits that will continue to complain about any remaining latency until there is ZERO latency whatever, which of course is impossible. So, basically they are never happy. Fast is never fast enough for them. The author of this thread is a clear example of this. The risk to the programmer is substantial, as normally fixing any kind of latency is not a matter of making a few tweaks, but involves substantial design changes. This is because all of the "obvious fixes" for latency are usually done in the beginning. Once a product is mature, there is no "low-hanging fruit" remaining. Then, a comprehensive testing cycle. And all for what - there isn't any added revenue to be made by reducing a 3 ms latency at the brokerage to 2 ms, when the ISP is adding maybe 50-100 ms anyways. Any improvement will be lost in the noise. When the "improvements" are rolled out, many times it breaks something else that was working well before. Then, you have THOUSANDS of real customers complaining and deciding to take business elsewhere all because of a few anal twits who are never satisfied. The programmer is now at risk of being demoted or fired if the problem is bad enough. If you were a programmer, tell me WHY the hell you would ever want to bother with something like this. I certainly can't find a good reason.
You pay for what you get with Latency = $$$$. I'm sure most of the players involved want fees for almost free?? If you want zero or almost none then you will have to cough up dough. Pretty simple proverb "You get what you pay for!". I think I heard it in china. Most likely your fee structure is low?? Most likely your system sends a bunch of cancels?? Those cancels take up network speed which the BD don't make anything on. So if you want better latency you will have to pay for it in higher commissions. BD's will make your latency go down if they make them money via commissions. Very simple.
The current CME messaging policy already restricts a message / fill ratio (ratio varies based on product), exceeding the limit would mean extra charges (more like a fine). The new CME Trade notification message aggregation only means that the fill messages *could* be grouped together, for instance, 5 x 1 lot fill in the past would come as 5 seperate messages, now, it may come as 1 x 1 and then one message notifiying the 4 lot fill. That's all.