Talk:2010 flash crash/Archive 1

Requested move

Mistake in duration: It says that it reached the bottom at 2:47. Is says "10 minutes later, at 3:07" which is actually 20 minutes later. Jcgul (talk) 19:07, 12 October 2010 (UTC)

May 6, 2010 market event → May 6, 2010 Flash Crash — I think we should change the name of this page to something more specific rather than the present name which is a bit vague. Yes, it was a market event but it was a crash, although temporary. I see it most commonly refereed to as the "Flash Crash". I think the name is fitting and descriptive of what happened. Although we won't know for awhile (maybe never) what happened, it was a crash in the markets. NortyNort (talk) 16:35, 9 May 2010 (UTC)

I agree that this is what the press is referring to it as. Probably a good move. Can't we just be bold and move it? --Rajah (talk) 18:57, 11 May 2010 (UTC)

I've moved it. I'm removing the movereq tag as well. --Rajah (talk) 18:59, 11 May 2010 (UTC)

Good call. "Flash crash" seems to be the label the news services are using, like Reuters. Edison (talk) 20:08, 11 May 2010 (UTC)

I think May 6th will be remembered as the Flash Crash just like October 19, 1987 was remembered as Black Monday.--NortyNort (talk) 20:19, 11 May 2010 (UTC)

time span of market move

Editors have added multiple times that the major indices moved over 9% in about 10 minutes - while the indices did move this amount over the entire trading day, the move in this ten minute window was not this large; furthermore, citations do not make any reference to a 10 minute window. I have taken out the "10 minutes" qualification - the only other way to make this factually correct is to change the percentage move to what actually occured in 10 minutes, which is between 6 and 7 percent. —Preceding unsigned comment added by 209.160.84.1 (talk) 15:50, 12 May 2010 (UTC)

Merge proposal

I suggest that we merge this with Wall Street panic of May 6, 2010. Both are about the same subject. Physalia physalis (talk) 23:41, 12 May 2010 (UTC)

Support. A third article was merged with this one today. I suggest keeping this one though, it has all the tags, etc. and is more thorough. --NortyNort (talk) 01:36, 13 May 2010 (UTC)

Just looked and found the Move has been completed. A redirect was put on the Wall Street panic of May 6, 2010 article by User:Pnm on 2010-05-19T04:53:01. Edit comment was "(←Redirected page to May 6, 2010 flash crash)"

Nanex analysis

No mention of the Nanex analysis? Written up in The Atlantic. TRS-80 (talk) 16:00, 5 August 2010 (UTC)

Theory with no basis?

In response to this:

"According to this theory, high frequency traders increased the noise in the system by stuffing non-executable orders (outside the highest bid and lowest offer) into the system to introduce delays that they could then exploit."

Unfortunately looking at how HFT works I see no way this would actually create a profit for any firm, and the article cited does not say how it would either. Financestudent (talk) 22:11, 26 August 2010 (UTC)

It is speculative. The cited article does say that Lo and Kearns both believe that these orders were 'latency testing' orders. Also, as originally written, our article is clear that this is just a theory. --RegentsPark (talk) 22:24, 26 August 2010 (UTC)

BTW, the addition that defends high frequency trading is not necessary. First, it is not relevant to the non-executable orders, and second, given all the hoopla over the nanex findings, I would not use the phrase 'directly disproved' because there are plenty of sources that are still putting forward this as a theory (in fact, it appears to be the theory du jour!). Perhaps, instead, we should try to make it clear that all the stated causes are theories and that no one really has an explanation for the flash crash (the NYT article is a good source for that statement) --RegentsPark (talk) 22:29, 26 August 2010 (UTC) I removed the CME quote since it is a generic statement about HFTs (and pre-dates both recent analysis of the data, as well as the interim report published by the SEC).--RegentsPark (talk) 23:05, 26 August 2010 (UTC)

If you are removing the quote would it be fair to place it else where or at the very least keep some reference to the source as a discrepancy to the nanex theory. It was actually not a generic statement and was the concluded results of their investigation into the crash and therefor quite relevant. Financestudent (talk) 00:16, 27 August 2010 (UTC)

According to the quote: However, there is no visible support of the notion that algorithmic trading models deployed in the context of stock index futures traded on CME Group exchanges caused the market fluctuations in question, and my reading is that its main focus is the futures market. I assume that there must have been speculation that HFTs trading in stock index futures (possibly because of the ETF connection - which needs to be added as a theory) were responsible for the crash and the CME came out with a statement that there was no evidence that that was the case. If there was speculation about trading in the S&P or other futures as a cause, then that should be added, along with the CME statement (in brief). Otherwise, I think the current formulation is ok. I removed the market manipulation bolded point (you were right about that - market manipulation as a theory is highly speculative) and rephrased it as HFT trading and with the manipulation theory properly contextualized. Do you know if HFTs trading in index futures was thought of as a theory? Point #4 talks about put options on the SP500 but not about futures.

One thing we need to do better is to explain that the causes are not understood. The lead in under 'theories' needs a rewrite for sure. --RegentsPark (talk) 02:28, 27 August 2010 (UTC)

It might have been till that report, there was large directional bets in futures which I think was noted but I would agree with you that it is misunderstood. Furthermore there really is no way a HFT system could profit from the market having lag or error considering the opposite is true and that would effectively prevent any of the HFT firms from making money or trading properly. If that is untrue feel free to show how a HFT firm would benefit from that. Financestudent (talk) 20:10, 27 August 2010 (UTC)

I tweaked the language a bit to make it more in line with the source (the Lo and Kearns explanation) so that it is clear that experts believe that manipulation of the sort described by Nanex is not practical. --RegentsPark (talk) 03:45, 28 August 2010 (UTC)

(od) BTW, I'm thinking that the article should be at Flash crash rather than at its current title. Flash crash redirects here and has become the preferred name of the event. Sounds reasonable? If there are no objections, I'm going to move it over the next couple of days. --RegentsPark (talk) 03:47, 28 August 2010 (UTC)

Discrepancy in "Technical Glitches"

Sorry, don't feel comfortable updating this myself (too new to wikipedia) but there is a discrepancy in the suggestion that movements in AAPL might have affected the CQS SIP. AAPL is a UQ SIP reported instrument. The two security information processors are completely different technologies run by two separate companies. If there is going to be a suggestion that AAPL movements contributed from a technical perspective it should be against suggested delays in the Tape C (UQ) SIP. - (vigilsrs) 10:31, 07 Sept 2010 (UTC)

I don't think that the AAPL and CQS glitches are meant to be related. The text talks about the CQS glitches and then says 'at the same time' about the AAPL and BID glitches, implying that these were independent glitches. Feel free to wade in and tweak the text if you think it is not clear!--RegentsPark (talk) 14:33, 7 September 2010 (UTC)

Are those trades valid?

I heard that some people said that those trades during that period was void. Is this true? With a reference please! Jackzhp (talk) 04:52, 8 November 2010 (UTC)

Not all trades. About 21,000 trades in 326 securities were cancelled. (See [1]) --RegentsPark (talk) 15:48, 8 November 2010 (UTC)

naming

putting Dow Jones in the page name doesn't make sense —Preceding unsigned comment added by 70.180.238.179 (talk) 03:53, 10 January 2011 (UTC)

Prehaps we should use the full name "Dow Jones Industrial Average" or the common abbr DJIA. --Quest for Truth (talk) 08:33, 10 January 2011 (UTC)

Yea, I don't think this was a good page move. Weren't all major indices effected by the crash? Also, based off of common naming, this crash was the flash crash of 2010. I don't think there was a need for the move. I proposed reverting it.--NortyNort (Holla) 13:02, 10 January 2011 (UTC)

Agree. The crash had nothing to do with Dow Jones, nor with the 30 components of the DJIA. It affected all US stocks on all exchanges. I renamed it back to a more sensible title, although if anyone can think of a better name I'd welcome that. Owen× ☎ 16:06, 10 January 2011 (UTC)

I have no problem. Looks good, thanks.--NortyNort (Holla) 21:13, 10 January 2011 (UTC)

Order flow toxicity

Order Flow toxicity looks like marketing speak that has been inserted into the Wikipedia document - Having it state that it is patent pending means nothing. Further because of the very nature of it being patented means that it does not have any authoritative source using it (Wsj for example) and hence is likely first hand data being plugged in. —Preceding unsigned comment added by 120.61.137.226 (talk) 20:17, 28 March 2011 (UTC)

99.236.29.247 (talk) 08:22, 4 April 2011 (UTC)Should the 3 sections of this article really be "Background", "Aftermath" and "Order Flow Toxicity". That's not a logical breakdown- the "order flow toxicity" is not an appropriate third section. Perhaps make a note about it in the aftermath section, but it certainly does not warrant a section along side the first two.

Someone added a sentence about order flow toxicity to the first paragraph, and I moved it from there to the bottom of the page. At best it's early research. I don't see how it's encyclopedic at this point and it seems that's the consensus here too. I will remove. MarketsGuy (talk) 21:03, 12 April 2011 (UTC)

As a matter of fact, order flow toxicity is the ONLY explanation accepted and published by a top, peer reviewed, scientific journal. Being the only scientific theory widely accepted so far, it should not be removed from the article. The official report has not been peer reviewed and accepted by a scientific journal, and you have kept it in the article.(PaulTheOctopus (talk) 23:46, 3 June 2011 (UTC))

So far as I know, "order flow toxicity" is not a widely understood term with a universal definition, and is used by the authors to describe their own theoretical construct. It may well be the only explanation accepted and published by a journal - but as you point out that statement needs to be qualified with "so far." There may well be a number of papers coming - should they all be discussed? Should they all be mentioned in the first paragraph? The official report is included here first because it is the only US government study of the event and has been presented by government regulators as a definitive analysis of the event. It was also widely cited and discussed as such by secondary sources. And if Bloomberg is correct that the "order flow toxicity" metric is the subject of a pending patent filed by the authors (see http://www.bloomberg.com/news/2010-10-30/-toxic-orders-can-predict-likelihood-of-stock-market-crashes-study-says.html), shouldn't that be mentioned? User:MarketsGuy|MarketsGuy]] (talk) 20:27, 7 June 2011 (UTC)

I moved the sentence "One hour before its collapse, the stock market registered the highest reading of order flow toxicity in recent history" PaulTheOctopus added to the first paragraph down to the section he created called "Academic Research," which he uses to showcase a recent journal article. The sentence - particularly if it's in the first paragraph - might lead a reader to conclude that the footnoted article is a widely accepted view of the Flash Crash, when for now the paper seems to be emerging research into the event. MarketsGuy (talk) 20:49, 7 June 2011 (UTC)

Which timezone?

Which timezone is referenced for given dates and hours? GMT? Local time? -- 78.53.67.247 (talk) 09:15, 13 June 2011 (UTC)

Too many words dedicated to the CFTC report

Should we rename the article "The CFTC report on the flash crash"? The CFTC did an Ok job, but it does not justify dedicating 3/4 of the article to one of several explanations, particularly from a source that has not been peer-reviewed.(PaulTheOctopus (talk) 03:24, 14 July 2011 (UTC))

As a member of the committee to which the SEC/CFTC report was addressed, one of the authors of the very paper PaulTheOctopus propounds here co-wrote a document, along with two Nobel laureates in economics and others, which said the SEC/CFTC report "provides an excellent picture into the new dynamics of the electronic markets..." See below. MarketsGuy (talk) 23:47, 16 July 2011 (UTC)

I'm reverting your rearrangement of this entry to restore the SEC/CFTC section. The SEC/CFTC report is not merely "one of several explanations," it is the definitive explanation of the event by the federal government, based on an empirical analysis of data available only to the government. As such, unless and until that data is released to others and analyzed, it stands as absolutely authoritative regardless of whether it's yet appeared in an academic journal.

I'm also reverting your deletion of the reference to the Kirilenko paper. The qualification that academic research must be limited to peer reviewed journals seems a convenient way to limit discussion to your preferred theoretical article. In fact, because he's Chief Economist of the CFTC, Kirilenko had access to far more detailed data than any other researcher to date, and his paper has been widely read and commented on by competent authorities. It is well worth noting that Kirilenko's paper is in fact cited in the very same article you're propounding, so to the question of whether Kirilenko has been reviewed, I'd say it certainly looks like it - see Easley, D., M. López de Prado, M. O'Hara: "The Microstructure of the ‘Flash Crash’." MarketsGuy (talk) 17:45, 14 July 2011 (UTC)

I do not believe or suport any explanation, just point out that most experts disagree with the CFTC story. Two reasons why this reference must be removed: 1) If the article is based on "secret" data, its results cannot be reproduced. The scientific method demands transparency, and the paper you insist to push forward cannot be validated. What if all those calculations were simply made up to fit a petty bureaucrat's agenda? 2) Government studies obey different standards than academic papers. They are addressed to a different public and they are not at all authoritative. Read the CME's response to the CFTC study ... Finally, you are giving way too much credit to the federal government or its employees. In matters of quant finance, they are understandably behind. Remember Madoff? I ask you to refrain from adding more comments on the CFTC. This is an article about the flash crash, not about the CFTC understanding (or lack of it) on this topic. (PaulTheOctopus (talk) 01:37, 15 July 2011 (UTC))

I'm afraid I'm going to revert. Here's why. First, your statement that "most experts disagree with the CFTC story" is unsupported. Next, taking your statement that "the scientific method demands transparency" at face value, we would probably exclude the paper on the VPIN metric because it is apparently soon to be patented and is therefore proprietary - and hence not "transparent" itself.

Your statement that "government studies obey different standards than academic papers" is also unsupported. What's your evidence?

Let's use an analogy on the main point. Suppose we were talking about a plane crash. The crash is investigated by the NTSB. The NTSB has black box data and all kinds of forensic capabilities on the plane itself. It then writes a report citing all of that evidence. Wouldn't that report be authoritative unless and until there was broad consensus around an alternative? Of course it would. The NTSB's report would surely be more broadly authoritative than, say, a professor of aviation engineering's theories about what went wrong.

Finally, though you may disagree, the SEC and the CFTC are the regulators for the equities and futures markets, and by virtue of that role their report has primacy in the discussion, at least for the time being. That's why the paper you introduce cites all of their work, and that's why when the SEC/CFTC report was released it made headlines around the world. MarketsGuy (talk) 18:26, 15 July 2011 (UTC)

You cannot be serious ... An NTSB investigation does not constitute Academic Research. Look, I'm only asking you not to hijack the article by dedicating 2/3 of its content to a single source. A Government report based in "secret" information is not authoritative, it is authoritatian. It is not based on peer-reviewed consensus, but imposed at gunpoint. How definitive was the SEC investigation that cleared Madoff a couple of years before he blew 50 billion dollars from private investors? Ask the thousands of victims about your beloved SEC "opinions". As you know, if anyone at the SEC had an undergrad knowledge of math and statistics, they would have realized that Madoff was running a scam. So please, spare us the sermon that Government studies are the ultimate source of scientific knowledge. Wikipedia is a free encyclopedia, not an instrument of Government propaganda. Cut your tirade by half, and let other points of view be known.(PaulTheOctopus (talk) 19:49, 15 July 2011 (UTC))

A couple of points. For the first time in this exchange, you didn't delete Kirilenko's paper from the "Academic Research" section you created - congratulations on letting other points of view be known there. Next, "Early Theories" as a section is an artifact from before the release of the SEC/CFTC report, where a number of speculations about the cause of the Flash Crash were detailed. Putting it after the details of the SEC/CFTC report violates chronology but certainly sequences legitimacy.

On the topic of legitimacy, this isn't the place to discuss your view of the government or of its employees, but it's undeniable that the SEC/CFTC report has been widely noted and discussed - making headlines around the world - and remains the definitive version of events. We should also remember that the subtitle of the SEC/CFTC report is "Report of the Staffs of the CFTC and SEC to the Joint Advisory Committee on Emerging Regulatory Issues." And who was on this Joint Advisory Committee? Among others, none other than a co-author of the very paper you propound here, Professor Maureen O'Hara. If the report is as flawed as you imply, why didn't O'Hara do something about it? In fact, O'Hara (along with two Nobel laureates in economics, among others) said the following about the SEC/CFTC report you attack:

The Committee believes that the September 30, 2010 Report of the CFTC and SEC Staffs to our Committee provides an excellent picture into the new dynamics of the electronic markets that now characterize trading in equity and related exchange traded derivatives...The Staff Report describes in considerable detail the impact of one institutional algorithmic order on the CME’s E-mini market and direct reaction to the E-mini price movements on the markets for large numbers of equities securities and ETFs.

http://www.sec.gov/spotlight/sec-cftcjointcommittee/021811-report.pdf

So once again, I'm restoring the section on the SEC/CFTC report above "Early Theories." The section has the virtue of telling readers what actually happened by summarizing the only detailed account of the event yet published. And I remind you again that not only does the paper you propound cite it, but a co-author of that paper was on the very Committee for which the report was written, and that the co-author, Maureen O'Hara, as well as a couple of Nobel laureates in economics, among others, applauded the report.

Finally, the section on the SEC/CFTC report is about 950 words of the article's current 4500 word length. That's a quite bit shy of the "3/4 of the article" or the "2/3 of its content" you have complained about. MarketsGuy (talk) 23:33, 16 July 2011 (UTC)

I have repeatedly said that I personally like the SEC/CFTC report. But personal opinions should not matter when contributing to Wikipedia. I do not favor any particular theory or try to diminish/trash others. We have to let all reasonable views be known, and allow the reader reach its own conclusions. As a matter of principle, I oppose your argument that a Government report is by definition the ultimate scientific authority. An Administrator will pick this up and settle the controversy.(PaulTheOctopus (talk) 14:37, 17 July 2011 (UTC))

I never made any argument at all that the SEC/CFTC report was the "ultimate scientific authority." I said it was "the definitive explanation of the event by the federal government" and that "the SEC and the CFTC are the regulators for the equities and futures markets, and by virtue of that role their report has primacy in the discussion, at least for the time being." I pointed out it is the most exhaustive description of the Flash Crash to date and that when the report was released it made headlines around the world. To my knowledge, there is yet no published refutation of it. Finally - and very importantly, given your complaints about government competence and peer review - the report was written for, and overseen, endorsed and cited by your own expert. It was also overseen and endorsed by two Nobel laureates in economics and a handful of senior industry notables, as well as, of course, by the SEC and CFTC themselves. These facts may make you angry, but they are all facts, amply footnoted, which I believe to date also establish both the legitimacy and primacy of the report in this discussion. MarketsGuy (talk) 16:47, 17 July 2011 (UTC)

I will not engage in an edit war. An Administrator will take care of this. Read the message posted on your talk page.(PaulTheOctopus (talk) 17:00, 17 July 2011 (UTC)) You are quite right on this point and my apologies. MarketsGuy (talk) 17:09, 17 July 2011 (UTC)

Restoring deleted sections

I have restored the latest version of August 4, 2011. User 76.104.152.229 (IP address from Washington DC) deleted entire sections of the article, namely the criticism of the SEC-CFTC investigation, without justification or comment in the Discussion page.(Berklabsci (talk) 15:41, 7 August 2011 (UTC))

Did VPIN reach its historical high one hour before the crash?

Berklabsci added the following paragraph to section Academic research:

A few months after its publication, the Journal of Financial Markets was forced to retract Andersen and Bondarenko's study, due to evident errors in their calculations.[48] A rewritten version of Andersen and Bondarenko's paper was later published, which was also proven to be erroneous by two independent Government studies. These two studies have pointed out various errors that explain Andersen and Bondarenko's failure to compute VPIN.[49][50]

The paragraph refers to AB (14a): Andersen, T. and O. Bondarenko, VPIN and the Flash Crash. Journal of Financial Markets, forthcoming. http://ssrn.com/abstract=1881731.

For now, I removed this paragraph, which consists of incorrect, unsupported, very biased, and (arguably) slanderous claims:

1. The link to the retraction notice, does not say that AB (14a) was retracted due to evident errors in their calculations. This claim is false and is contradicted by available sources. However, if Berklabsci insists on it, then, by Wikipedia policies, he must provide a reliable source that directly supports the material.

2. The next claim that AB (14a) was "also proven to be erroneous by two independent Government studies. These two studies have pointed out various errors that explain Andersen and Bondarenko's failure to compute VPIN.[49][50]" is also false.

Neither of Berkeley Lab studies (http://ssrn.com/abstract=2318259 and http://ssrn.com/abstract=2062450) asserts that AB (14a) failed to compute VPIN. In fact, this Berklabsci's claim is completely absurd, since neither study even cites AB (14a)! The first one does not mention any paper of AB. Again, if there exists a Berkeley Lab study which confirms that VPIN did reach its historical high before the flash crash, then, by all means, Berklabsci should add this reference to the article. Otherwise, it's misleading to just say that somebody disagrees with somebody on some (unrelated and marginal) issue.

3. As a side note, it is highly disputable whether Berkeley Lab studies can be called independent. Lopez de Prado, one of the VPIN creators (ELO), works at the Lab, he has been a co-author of Lab scientists, and he has served as an adviser/consultant for the Lab studies. Moreover, ELO, who have strong financial interests in the adoption of the VPIN metric, have served as editors of WBGLR’s paper. There is nothing wrong in two author teams (WBGLR and ELO) sharing information and objectives, cooperating, and promoting each others work. However, they are no longer independent teams. In the article, the Lab's studies on VPIN should not be called independent.

In conclusion, I suggest Berklabsci rewrite his material taking into account the Wikipedia's core principles: neutral point of view (WP:NPOV), no original research (WP:NOR), and verifiability (WP:VERIFY). NMLDP (talk) 02:03, 30 September 2013 (UTC)

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

I want to point out that Berklabsci deleted my explanations on Talk page, which is against Wikipedia rules. In particular, "The basic rule ... is that you should not edit or delete the comments of other editors without their permission." (WP:TPO). See: https://en.wikipedia.org/w/index.php?title=Talk%3A2010_Flash_Crash&diff=575091491&oldid=575081618 NMLDP (talk) 04:46, 30 September 2013 (UTC)

Points 1. and 2. above seem to call for a response from Berklabsci, including some firm evidence. Cyclone35 (talk) 23:42, 2 October 2013 (UTC)

Apologies for the unintended deletion while editing Talk comments. (Berklabsci (talk) 16:08, 6 October 2013 (UTC))

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Berklabsci, the apology is accepted. For the record, though, there wasn't any editing by you of Talk comments. Just deletion of one section, nothing else was added or changed.

Overall, your last rewrite is an improvement. It now states:

An independent study conducted by a Federal Agency has found errors in Andersen and Bondarenko's study. In particular, scientists at the Lawrence Berkeley National Laboratory have shown that Andersen and Bondarenko did not compute correctly VPIN's false positive rates. [48]

I am happy to see that you no longer insist on inserting the false and inflammatory statement covered in point 1 above. For point 2, I am glad that you removed reference to one Berkeley Lab study (which does not cite AB at all) and another false claim. Nevertheless, I respectively ask you to re-write your passage again and to be very clear which specific claim from which of the AB three papers is disputed by the Berkeley Lab report. Importantly, please make sure that the disputed claim is relevant to the content of the WP article.

The issue at hand is whether VPIN reached its historical high one hour before the flash crash. If there exists a Berkeley Lab study which confirms that fact, please provide a direct quote, as required by Wikipedia. Computation of false positive rates in unrelated exercise has nothing to do with the content of the article. In fact, the issue does not arise in any ELO papers, is completely marginal, and seems to stem from a basic confusion of WBGLR about AB procedure (http://ssrn.com/abstract=2331106). Berklabsci, your current statement remains misleading and might appear related to a personal feud with AB. But we are clearly making progress. NMLDP (talk) 01:12, 7 October 2013 (UTC)