Andy
Posts:4
 |
| 12/14/2007 11:47 AM |
 Quote
 Reply
 Alert
|
| I've implemented both CQS and NYSE Quotes feed processing and now I'm looking at differences in the two feeds. I'd expect that every quote that appears on NYSE Quotes also appears on CQS (from participant N) and vice versa. Is that the case? I seem to be seeing slightly more quotes on CQS than on NYSE, even though I see no packet loss on either feed. |
|
|
|
|
kgilchrist@nyse.com
Posts:0
|
| 12/14/2007 3:46 PM |
Quote
Reply
Alert
|
Andy,
Can you give us some high-level numbers so we can check against our end of day files?
Please specify the date(s) that you analyzed.
Thanks,
Kevin |
|
|
|
|
Andy
Posts:4
|
| 12/17/2007 9:42 AM |
Quote
Reply
Alert
|
| I'm seeing maybe a little less than 2% more quotes from CQS than from NYSE Quotes. So far I've only looked at quotes for IBM over short time windows (several minutes). Note that I'm excluding CQS "status" messages (halts, resumes. etc) since they don't include quotes. |
|
|
|
|
Ariel
Posts:163
|
| 12/18/2007 3:46 PM |
Quote
Reply
Alert
|
| Andy, what you are seeing is a definite possibility. The NYSE Best Quotes feed is a latency centric feed, as a result if the system detects a newer quote (quote 2)for a symbol is immediately behind the latest quote (quote 1) received than the system will consider quote 1 stale and will send out quote 2. This is way you may see slightly less quotes on the NYSE Best Quotes feed |
|
|
|
|
Andy
Posts:4
|
|
Andy
Posts:4
|
| 12/21/2007 1:28 PM |
Quote
Reply
Alert
|
| Actually, one more question: if a newer quote can "overwrite" a stale quote, that implies that the quotes are queued before they're sent out. How long are quotes queued for? Is that constant, or does it vary with load? |
|
|
|
|
Ariel
Posts:163
|
| 12/21/2007 2:57 PM |
Quote
Reply
Alert
|
| The quotes are not necessarily queued. We don't a specified time at all before we send them out. If the system sees a newer quote in the buffer before it fills up the packet to 1400bytes than we discard the oldest quote. This scenario is rare and happens about 1-2% of the time. |
|
|
|
|