i was reviewing some data from the 20th of november (big gap down spy day) and realized pulling history for my alerts on TI showed me stock alerts it had not shown me live - twice as many tickers alerts. a little more digging and I realized the timestamps on the alerts started right after the first minute after the open, but the prices being alerted had to have occurred much earlier, anywhere from 930:15 to 930:50 depending on the ticker. so the short version appears to be they couldn't process scans with the market volume early that day, and some alerts were not sent at all (and others late, which were not flagged as so and I did enter tickers into strategies based on late alerts). I do have independent time-stamp on this as well as I pulled up charts/strategies based on the alerts given. Once in a whlie a ticker pops up in history with TI which did not alert live, but this was unusual in the quantity and the lag involved. are high volume open days like this a problem with all scanners? I haven't used equityfeed in years and don't even recall if I can scan for what I want to do on it, but I am open to any suggestions which might offer a reliable alternative.
I don't have direct experience in external scanners, but methinks your answer is in the 5-word phrase, "big gap down spy day". To translate, that's like "Everybody-and-their-brother trying to get out of X, and INTO Y, to maximal degree, and *all* at the same time." Yow. Kind-of a "Let's break the internet" day.
%% Strange how sometimes just when you need good data-- some goof.LOL I may my keep a SPY or QQQ window open, reguardless of me trading/investing them. Paper charts are best, along side computer charts . Less errors on paper charts or can correct them before market opens.My old Esignal daily chart went goofy, when i was aiming @ taking a profit out; but monthly chart still worked + bid/ask still worked....... LOL+ true.
yes, obv. the heavy mkt volume at open is the stress point, though I would think these data folks would be doing the hardware side based on such stress models. Alternately, was data from the exchanges themselves slow?
No, market data from (most) exchanges come via multicast so you either get it in time or don't get it at all.
here is an example of the lag in TI I found - (stock is FIVE, price is last column) FIVE BELOW INC 09:31:00 AM 11/20/2018 FIVE 105.34 FIVE BELOW INC 09:31:00 AM 11/20/2018 FIVE 105.41 FIVE BELOW INC 09:31:00 AM 11/20/2018 FIVE 105.75 these prices almost had to be the 930:14-16 second range, but they were time-stamped and received by me at 931. a whole set of different stocks printed alerts in the next few seconds (including some that only show up on review of history, didn't show up live) which appear to work through the 930:xx minute.
I had a chat conversation with TI, and they indicated the issue was on their end, in terms of not being able to process the data in a timely fashion very close to the open. My own observations indicate this is most likely an issue on a day the market gaps (either way). Today (dec 6th) early alerts ran 10-15 seconds late the first minute, 5 seconds or so early in the 2nd minute, and were timely after that, for example. Alerts 30-40 seconds late is a long time, though (actually even 10 seconds can be a long time under certain circumstances). Are the other real-time scanner products similar in high-stress environments?
what platform and/or data feed could one start with to DIY? I am assuming would need to locate remotely with a group I could scale cpu power with. not sure but I suspect the population of stocks I would need to scan is at least a couple thousand early in the day.
Here is what I would build like for myself. The goal is to have full control end-to-end. 1. Colo at NY4 (need to be as close to data as possible) 2. Data feed (raw would be best, but at the least streaming ALL symbols like polygon.io) 3. GUI - MotiveWave or some basic Java Based windows displaying the results. It's too much time and money to invest for manual trading, imo. If you do algo trading, then you can add better/faster execution which may make it worthwhile.