Have you already considered Citrius, Lime's realtively new data feed? If not, you might wish to since you use their brokerage services. If so, can you comment on why Citrius would not work for you? From their website: Citrius⢠Quote Delivery System Lime recently unveiled its new Citrius⢠Quote delivery system making Lime a leader in low-latency, high-performance data delivery. Citriusâ¢, based on our new groundbreaking core technology, greatly exceeds traditional financial technology system performance by applying state-of-the-art data-communications technology where latencies are measured in the tens of microseconds rather than the usual milliseconds one would expect for such systems. Moreover, the system provides "carrier-class high availability" where every system is backed up with a warm standby unit to take over seamlessly in the event of a critical system failure. Citrius⢠is another example of Lime's commitment to provide the best possible tools and technology to empower our clients to excel in today's highly competitive electronic markets. http://limebrokerage.com/index.php?title=Citrius™_Quote_Delivery_System
Tens of microseconds. I love these marketing announcements they just keep getting more and more absurd. Next year all these vendors will be quoting latency in nanoseconds.
I don't see anything funny here - there are some strategies that count every millisecond, and people pay big bucks to co-locate their boxes. Look at BATS for example, it took a lot of volume away from NYSE by providing blazing fast executions and market data, and a lot of alogs park their orders there because of execution speed and favorable rebate fee structure.
What makes me chuckle most about these sorts of numbers is that there is never any detail. Notice tens of microseconds instead of any hard figures and no definition of what they actually mean by latency. I don't know about Lime but I have contacted many companies in the past to try to get more detail and more often that not that there is little or no data to back up the marketing claims. Sometimes you get something like "we dont give out figures because it can vary depending on the configuration but we have a number of big customers using it and they are very happy" I'll believe the hype when they start accurately specifying and guaranteeing the figures.
It's tough for anyone to put numbers out there, as financial rules/regs make it difficult to say anything unless it's verifiable. You can't say anything in a press release or on a webbie that hasn't been verified or is up for debate, effectively. Rules are clear, clear, and are really hard to "bend" to market well. 'tis a fine line.... The other issue is that some places WANT to trumpet their numbers, but can't. Others choose to actively obfuscate them due to the constant sniffing by competitors. Full order acks really is the order stat you should be looking for; some place love to claim things move at the "speed of light" when things really can't. A good fiber or dark fiber line can be totally differnet figures; also, internal latency of any broker/add-on etc is REALLY what needs to be measured...and can be bent in a wide variety of angles. In the case here, it's pretty easy to measure latency add, as it's a direct feed we are talking about, and what is the ADD in latency upon what you're getting direct - once it has been normalized/cleaned (or not)/put in to usable format - JUST the QUOTE - before it can be used. If that's the case, it can be microseconds, and will eventually be lower...really. I agree that the marketing and claims in this space can be far fetched. It's really more a matter of what exactly is being talked about.
as far as i know lime founder also operates "tower research capital," and the fund is located in the same building (right nextdoor) as the brokerage in nyc. i've heard stories where people met w/ lime there to try to open an acct/inquire about their systems,etc., and lime tried to recruit them to join the fund at the same meeting once finding out a little about what the people are doing. as far as i can see, conflicts of interest could be very high.
NxCore is a classic example of "Artificial Stupidity"... I trade about 300 stocks and require << 5% of Consolidated NYSE messages... But these Clowns can only send the whole thing. Adding simple throttles... Like a list of symbols required... Needing updates only X times/second, etc... Would seem mandatory. The saleman I talked to acted like I was an idiot... Because as an Software Engineer... I refused to process the entire NYSE message universe on principle. There is a REASON why I make 7 figures annually trading... And that Dude gets mid-5 figures for answering the phone. "DTN NxCore requires minimal computer resources. Unlike some datafeeds that require hardware costing over $10,000, DTN NxCore can run on a low-end Pentium laptop (1.0 GHz, 256MB of RAM, 80GB IDE hard drive) and receive and database 900,000 quotes/second, over 6 billion/day; and CPU usage will remain under 5%." And this is just total, complete BS... Anyone trading anything with a 1.0 GHz machine is not even serious... And just the kind of Super Mook... That would try to process the ENTIRE NYSE message feed in his pajamas... Using DTN NxCore, of course. This is a classic example of a product... That an enterprise class trading firm... Wouldn't touch with a 10 foot pole.
I'd say they have made a business decision to send the entire thing and let the user filter as required. Any large shop can very easily and very quickly customize and distribute the data as it is being received. If they have a requirement for truly-real-time data they probably have exchange co-located machines to receive data and execute orders that run independently of a NxCore source. The shop would also have the entire set of all data for developmental work. Why should NxCore do it? It would greatly increase their support costs It could also get into a cost-per-quote per symbol discussion with potential customers and again increase their support costs. I think they are happy with the customer base they serve. Jack
Then I would think that, as a "Software Engineer", you'd be able to write a filter for whatever data you want. That's trivial compared with the other tasks involved in integrating a new data API, in my experience. But why even bother with NxCore (or any full-feed product, for that matter) when much cheaper alternatives such as tal, esignal, and many others offer (in fact, mandate) the vendor-side filtering you describe? https://toolkit.taltrade.com/