I think the main issue is that you are speaking in very vague terms without citing explicit sources of latency or firm numbers.
I agree with Stephen's note. I didn't want to get into any technical discussion because all the description of latency has been incredibly vague. I believe one poster tried to point out the potential areas of latency (BP check, etc), but no specifics has been outlined. It is clear that quite a few of us are at least familiar with the common problems faced with running low latency trading systems, so for this to be anything more than just talking about generalities about "slowness", we need some specifics, only then specific debates can occur. For instance, even just saying Market Data is slow is not sufficient, how is the market data distributed (and how it is sourced?), is it distributed over TIB, Wombat, PtoP, Spread, ACE? How does the protocol handle subscriptions? Is it because the data was source from an internal ticker plant, directly from the exchange, from a market data provider, any format translations? Even if we are talking about multicast, in sparse mode, any bridging? The details are incredibly diverse, just for market data sourcing and distribution by itself. Only then can we have some useful (and hopefully lively!) debate. Rufus
Ask Susquehanna, Citadel, Autodesk if they have latency? Doubt it. It all comes to how much cash you are willing to pay your broker and their willness to build you a system that isn't crap. Crap in is crap out. That is the definition of latatency on the board now. Or old ancient chinese proverd "You get what you pay for."
I lack the technical knowledge to form an opinion as to whether these latency issues have any validity, but I think it is a step in the right direction that IB is at least discussing and investigating the issue with the author of this thread.
"Ask Susquehanna, Citadel, Autodesk if they have latency? " Sometimes, the bigger you are, the more you overlook the smallest things..........and those make the biggest difference sometimes especially when you find 10 of them. Added together and you have something out of nowhere.
Quote from JMartinez: Ask Susquehanna, Citadel, Autodesk if they have latency? I'd be willing to bet they have overall latency far worse than the originator of this thread is seeking. That doesn't mean they're ignorant or not willing to spend the cash if needed. It means that their trading does not rely on beating another autotrader by less than 10ms. When I wrote my first "keep me at the inside" bot many years ago, it ran from a DDE Excel spreadsheet, and was fairly effective for what it attempted to do. As soon as the hordes of other bots started showing up, instead of just mindlessly improving my own speed (there was certainly plenty to be done there), I took a step back and thought about the road ahead. I realized that it would become a fight that either: A) Gets a bunch of people to the same performance, splitting up the profit to the point where it's not worth doing; or B) So inundates the market with orders that it becomes untradeable. I don't know if A is true, but B certainly is. Some markets have already started to react (e.g. GLOBEX cancel fees, aggregation). When I'm trading manually on a stock that has any kind of activity (like ERTS today), it kills me that I have to trade so sloppily because it's impossible to actually set up an order and take a price and have it still be there by the time you send the order, even with hotkeys. Hell, my machine gets so swamped by the quote flow, I can't even rely on the quotes. Personally, I found other ways of trading profitably that didn't end with the MAD scenario (at least not yet ).
Quote from alanm: ...When I'm trading manually on a stock that has any kind of activity (like ERTS today), it kills me that I have to trade so sloppily... Worse, I'd bet that the originator of this thread, and others like him, are on the other side of those "sloppy" trades, picking me off for 0.01-0.03. It could be said that, because they make the market untradeable, it has the same effect as a denial-of-service attack. Watching some of these bots in the pre-market, where you can actually see every print, it's clear that, as another poster suggested, the fill-to-order cancel ratio must be very low, further giving weight to my DOS assertion. While I'm not a fan of having to pay for cancels, hopefully as more markets start to threaten this, brokers will implement pacing restrictions to slow this madness down.
alanm, trading a fast stock today is pure bedlam. If you remember the days of 1/8-1/4 spreads (and more) and actual MM's , you know it was nothing like this. It's virtually impossible to game the actual inside price on most fast stocks. Too fast, and price is moved by the sweepers, who you will never be ahead of. 0 latency, and the sweep is still ahead of you. Hey you could always trade one of those $2 wonders that trade in penny moves. At least you can see whats going on.
Lets see: They can: 1. short without checking inventory (they are market makers). 2. unlimited buying power so no checking for buying power cuz they can. Latency, well they eliminated that latency and have only. 1. quotes- depending?? uncontrollable* but just say 40 milscecs 2. executions: ecn's, market makers-(most likely using fiber) , and specialist) about 0-40 milsecs ave probably lower on most ecn's, nothing if market maker themselves. Listed is higher if you use true acks. 3. back up databases if that-depending? controllable* In all, they can come in around 0-80 milsecs. Depending on the quotes. Not bad.