3

In a recent question (Were the shareholders of ZTNO (Zoom Technologies Inc.) just hustled?), I asked about an equity (stock) that appeared to suddenly go to zero. Other community members helpfully responded in the comments that it was likely a glitch.

In our ensuing discussion, community member @Fattie stated that these types of glitches are actually quite common. They did not know why, and felt it would make a great question.

This question asks just that: Why are there so many glitches in stock market data?

Flux
  • 17,043
  • 8
  • 68
  • 124
End Anti-Semitic Hate
  • 1,062
  • 10
  • 27

2 Answers2

2

Why are there so many glitches in stock market data?

The obvious answer is human error, aka GIGO (Garbage In, Garbage Out). Errors are made with stock market data entries.

Web site providers of free data subscribe to services that collect data. There's no real incentive (free?) for the provider to fix errors. The classic example of this is Yahoo Finance though they do not have a monopoly here. You can find a number of "bad data" posts on various Stack BBs and even more if you search the net.

Reliable data comes from two sources:

  • Data vendors who check for errors and clean up their database ($ubscription fee)

  • The stock exchanges who do the same. While not infallible, Time & Sales is the closest thing to perfect. It will prove the bid and ask every time it changes as well as shares and size for every trade.

Bob Baerker
  • 76,304
  • 15
  • 99
  • 175
  • Thanks Bob. Isn't much of this automated by computer, thus not requiring human intervention? – End Anti-Semitic Hate Oct 23 '20 at 16:40
  • 1
    One would assume that much of this is automated by computer, thus not requiring human intervention but how much is 100% minus "much of this"? If the answer is greater than zero then that's where the problem lies. I'm a retail guy so I have no clue what that number is. What I can tell you is that there have been many questions posted here about bad data and since I have access to Time & Sales, I have checked, often reporting that the bad data is the fault of the web site provider. – Bob Baerker Oct 23 '20 at 16:51
  • @BobBaerker How do you get access to Time & Sales data? Is it through your broker (e.g. IB)? Do you have Time & Sales data even for OTC stocks such as ZTNO that OP is referring to? – Flux Oct 23 '20 at 16:54
  • 1
    Yes, I get Time & Sales data from IB. OTC BB stock data is a fee based subscription which I do not subscribe to since I rarely trade BB stocks. And when I do, it's at another broker. I would assume that Time & Sales data would be included but I don't know if that to be a fact. – Bob Baerker Oct 23 '20 at 16:59
  • 1
    @RockPaperLz-MaskitorCasket There's probably a difference between historic data (time & sales) which can be "patched up" later if affected by a glitch and (near) current prices which can't (such as Yahoo show). My guess from your first question is you checked ZTNO's price on a few (free) sites when you saw the zero. I suspect that many sites showing (near) current prices will get their feed from one or two data sources: if one of those had a glitch of some kind (network interruption etc.) then all sites sourcing that data will be affected. – TripeHound Oct 23 '20 at 17:06
  • @TripeHound Your guess is correct. I checked several free sites, such as Google Finance, which were all wrong. I can easily understand why a network interruption would cause current prices to not get reflected, but why would such an occurrence show a price of zero? – End Anti-Semitic Hate Oct 23 '20 at 17:11
  • @RockPaperLz-MaskitorCasket If you've not been able to get an updated price, you've got three main choices: (a) show 0.00, (b) show the last price you did get; (c) show an "unavailable" message. Each has pros and cons: (c) is "honest" but might make a site look bad, and isn't helpful if someone isn't interested in the up-to-the second price. (b) would help with that, but carries the danger of someone who IS interested not realising the displayed price is out of date. (a) has the danger of scaring people into thinking their stock has crashed! – TripeHound Oct 23 '20 at 17:17
  • 1
    @TripeHound Great analysis. I guess it depends on how their system works. Usually, when there is no new data, most systems show the last known data. Option (c) seems best, as it is honest, with (b) the next best, but it seems sites like Google use (a) which is the worst and, as you mentioned, has only cons and no pros. – End Anti-Semitic Hate Oct 23 '20 at 17:22
  • @BobBaerker I've often wondered. there are *two different concepts*. (A) the actual underlying Time+Sales data has mistakes, and (B) the various websites F'up when they summarize/display the Time+Sales data. It seems to me that "B" is completely commonplace. But. Does "A" ever happen?? Or does Time+Sales data tend to be rock solid / definitive? – Fattie Oct 23 '20 at 17:23
  • @RockPaperLz-MaskitorCasket I don't think you can analyze it that closely. it's more like this: (i) programmers are utterly useless (ii) programmers at free websites are more so useless (iii) very few programmers can properly deal with data sources that change over time, have a confusing / archaic mixture of triggers, polling, wildly varying databases/apis/etc (iv) regarding displays of market data, "everyone knows" that they are very often F'd up, so, on top of i, ii, iii nobody cares that much anyway! – Fattie Oct 23 '20 at 17:26
  • @Fattie I dare you to post that on StackOverflow! :) :) :) – End Anti-Semitic Hate Oct 23 '20 at 17:27
  • 1
    @Fattie - Time & Sales isn't infallible. People make errors. The big difference is that the stock exchanges strive to provide correct data as do fee based data vendors some of whose products are very expensive (tick data?). Web sites like Yahoo, Google, MarketWatch, etc. pay a vendor and that's why an error is redundant across the board. There's little, perhaps no incentive for that vendor to correct mistakes. – Bob Baerker Oct 23 '20 at 17:35
  • Data collection has been going on for over a century (ticker, newspapers). Computer data, maybe 40-50 years. I'm not sure that programmers are major culprits in this bad data issue. Perhaps they are occasionally at fault when systems are changed (reprogramming). I've been paying for data for almost 35 years. In the 80's, it was by the quote, LOL - cheaper at night. No data correction then. I've subscribed to EOD data from Thomson Reuters for maybe 20 years and they update 2 days of data every night (today's numbers and an overwrite of yesterday, in case there were errors yesterday). – Bob Baerker Oct 23 '20 at 17:44
1

You’re mistaking common and frequent. Stock pricing glitches are common, but not frequent. That’s because of the vast volume of data — prices every few seconds for thousands of equities means that there are millions of data points every day. If one out of every ten million of those data points is wrong, then there will be several errors per week, which makes them common. But one error out of every ten million is a very low frequency of errors.

Mike Scott
  • 23,450
  • 2
  • 65
  • 77
  • If investors think they lost all their money even just once a month, I'm guessing they would consider it common and frequent! – End Anti-Semitic Hate Oct 24 '20 at 22:57
  • Your point makes sense, but I think it's misleading at best to say that this means such errors are "not frequent". You're using "frequent" only in the narrow sense of "how many erroneous events out of the total number of events". But in common usage "frequent" also refers to "how many erroneous events per unit of time". In this common sense, several errors per week is pretty frequent. I'm not sure there's much relevance to your notion of frequency; the question is really "why is the number of errors so high on a visible-to-the-naked-eye level". – BrenBarn Oct 24 '20 at 23:20