Nasdaq “test” data gets published, causing confusion over tech stocks
- Thread starter JournalBot
- Start date
Can we get an update if anyone tries to file a lawsuit against NASDAQ for emotional trama. My guess is they'll be at least 25 notable suits filed in the next two weeks.
Upvote
13
(15
/
-2)
I wouldn't bet on it. Most of the market today is not people, but competing machine-learning algorithms. Unless the human programmers go in and purge this data, it could easily influence all sorts of internal models in unpredictable ways.
Upvote
48
(48
/
0)
Did a small web app for Nasdaq, not super surprised. Making minor front end css only changes required an approval process 7 people and 2 countries long, often passing technical description through decidedly non-technical people with the direction that ultimately came back often only casually resembling the initial request.
Edit: I realize that they're blaming the vendors, but to me this sounds like a communication problem; it just seems unlikely that if the test was clearly communicated that so many different outlets would have erroneously published test data.
Edit: I realize that they're blaming the vendors, but to me this sounds like a communication problem; it just seems unlikely that if the test was clearly communicated that so many different outlets would have erroneously published test data.
Upvote
18
(25
/
-7)
The UTP QDF provides only BBO (best bid/offer). I highly doubt anyone is using this for trading or modeling. That's done using other feeds that can be used to compile books of greater depth.
It looks bad, but was of little danger.
Upvote
13
(13
/
0)
NASDAQ would be the wrong target. The correct target would be those that erroneously published known incorrect information.
NASDAQ does this kind of testing on a regular basis. Anyone who works with them should be well aware of this, and known better than to publish such information.
Upvote
23
(23
/
0)
Now all we need are some enterprising lawyers willing to file for emotional distress on behalf of an algorithm. If there's a buck to be made....
Upvote
8
(8
/
0)
I saw something similar last week, where for maybe an hour or two after hours Yahoo Finance was showing a close value of $0 for everything, which led to various apps using that data (including iPhone's built-in stock app) showing a 100% drop in price.
Upvote
3
(3
/
0)
The data will stay because the social media and news media reactions are valid outputs of that data input, and that is relevant data, even if it isn't valid pricing data.
Anyway, I doubt that the top algorithms consider the market unpredictable - in the short horizon - to any great degree any longer. Which is not to say that the market won't crash, only to say that they'll profit from it when it does, shortly. The unpredictability of even individual investors has probably been taken away by facebook, fake news and twitter.
Upvote
-3
(1
/
-4)
A LOT of the algos are using Twitter and other social media as an input so the fact that the pricing got out into those outlets could cause a bit of a problem.
Upvote
2
(3
/
-1)
I realize this was a planned and announced test, but I still can't help but thinking of this:
Upvote
43
(44
/
-1)
These data feeds are notoriously unreliable. And then there are all the sorts of bugs and fuckups in the front-ends.
While I haven't ever been directly affected, the brokers I use usually have a "sorry we messed up the quotes for $AAPLQ today" apology somewhere.
The only real way to get the real price action is to tap directly into the exchange during market hours. Then you only have to worry about fat fingers and flash crashes. Fake L2 data feeds should be unrealistic enough most of the time, too.
Other than that, you can check the company website and news sites.
As usual, think before buying rope.
While I haven't ever been directly affected, the brokers I use usually have a "sorry we messed up the quotes for $AAPLQ today" apology somewhere.
The only real way to get the real price action is to tap directly into the exchange during market hours. Then you only have to worry about fat fingers and flash crashes. Fake L2 data feeds should be unrealistic enough most of the time, too.
Other than that, you can check the company website and news sites.
As usual, think before buying rope.
Upvote
2
(2
/
0)
Um you realize the environment at Nasdaq in regards to change control and the like has nothing to do with NASDAQ right? All financial institutions have that requirements, and IT regulations for them have gotten insane. I work in IT at a medium bank and I remember when we went from the small classification to medium, got about 10x the work thanks to new regs for size. NASDAQ is an even larger organization with even more onerous regs.
Thanks to Enron, Williams and the like for all the trouble.
Upvote
3
(3
/
0)
The more I learn about the markets, the more I'm convinced that the guys running it are inept morons.
Upvote
-6
(0
/
-6)
As TFA said, and as has been pointed out a few times in the comments, already, this wasn't Nasdaq's fault.
Bloomberg and whatever other data vendors released it to the public are the ones to blame.
The change control processes around everything in the financial industry are nuts. I know. I work for one of the big 3 multinational data vendors. This stuff is pretty well-communicated and there's plenty of advance notice given. The fact this happened at all can probably be traced back to one or two people at each organization that let it leak who were supposed to have thrown some sort of switch before the test, but didn't because they ducked out of the office early for the holiday and were on the road or something.
And I'm sure they got paged and had a stern postmortem with their superiors and half their department, once it was over.
Of all the things that could go wrong in all these systems, this was probably one of the "best" things that could have gone wrong while still being client-facing. Much less destructive to real dollars than an outright outage or bad data during market hours.
And this is also one of the reasons why most investment bankers don't just rely on a single platform to do their jobs. Many IB firms have a mix of Bloomberg, FactSet, and/or Thompson-Reuters workstations for the various strengths each platform has.
Good to know we're all inept morons, though. Nice to have a reality check from time to time...
Upvote
6
(7
/
-1)
I always thought this is wrong, very wrong. There should be people in the other end, not machines.
Upvote
1
(1
/
0)
To be fair. If a failure is going to happen, it's going to be there.I realize this was a planned and announced test, but I still can't help but thinking of this:
Upvote
1
(1
/
0)
Upvote
1
(1
/
0)
Also, totally. I used to be a data consumer, using direct feeds, at one of the big international banks, doing software in equities.
News like this usually go out months in advance. They typically, but not always, pick random Saturdays to do industry-wide tests though, and not one of the lightest-staffed days in the year where when something like this happens people won't be around to catch. Also the advantage of doing this on a Saturday (instead of hours after) is that we'd have more time to triple-check our work to make sure things don't break.
I won't lay all the blame at Nasdaq, but their choice of timing was incredibly poor. I'm surprised they didn't push back harder.
Upvote
1
(1
/
0)
- Status