There’s been this meme circulating that 70% of all trading volume on the exchanges is HFT, and that the average holding period for stocks is 11 seconds. Punch into Google “Average Stock Holding Period: 11 Seconds” and you get 850,000 results.

The problem is, none of these “data points” are backed up with real data. I set about tracking down where this meme came from.

I had first read this number in an interview with Peter Cohan in Marketplace. I tagged Peter, and he gave me Bloomberg as the source for the 70%. It turns out that the Bloomberg article in question quotes Raymond James analyst Patrick O’Shaughnessy. He appears to made a guess that “High-frequency trading may accounts for 70 percent of share volume in the U.S.”  But that “may” mean it is a only a guess, not backed up with data. Somehow, that 70% estimate somehow morphed into a fact. I’ll reach out to O’Shaughnessy to see what he says.

As to the 11 seconds holding period, here is the NYT:

“The founder of Tradebot, in Kansas City, Mo., told students in 2008 that his firm typically held stocks for 11 seconds. Tradebot, one of the biggest high-frequency traders around, had not had a losing day in four years, he said.”

This 11 seconds data point is 1) a rough number, 2) without much supporting data; 3) spitballed to a group of students, 4) for one small HFT firm.

Other people have looked at holding periods and came up with far longer holds. David Hunkar recently wrote:

“Based on the NYSE index data, the mean duration of holding period by US investors was around 7 years  in 1940. This stayed the same for the next 35 years.  The average holding period had fallen to under 2 years by the time of the 1987 crash. By the turn of the century it had fallen to below one year. It was around 7 months by 2007.

Similar pattern exists in the UK also as shown in the chart above. There the average duration has fallen from around 5 years in the mid-1960s to less than 7.5 months in 2007.

Over the past 15 years even in international equity markets, holding periods have fallen. The Chinese market was red hot until few months ago. However the duration for the Shanghai stock market index is close to just 6 months.This shows that Chinese investors do not have a long term horizon.”

The best data I’ve tracked down confirms that range of holding times.

>

Conclusion: The 11 second number is wrong. There is no data supporting that. Estimates of HFT of 70% of trade volume are just that — estimates — and we have no evidenciary proof or data that HFT trade volumes are 70% . . .

>

Update: O’Shaughnessy writes back “That’s an estimate I put together by speaking with various industry

He adds: “HFTs are the new market makers, which puts them on essentially one side of all exchange trading. Exchange trading is around 70% of all trading activity, so that gets them to 35% share. Then you have a lot of exchange trades where the hft algos trade against each other, plus hfts do a lot of dark pool trading too. 70% may be a bit high but I would wager it is easily 50%.

>

Sources:
Flash Trade Halt Backed for Nasdaq, Bats as SEC Votes
Whitney Kisling and Jesse Westbrook
Bloomberg, Sept. 18 2010
http://noir.bloomberg.com/apps/news?pid=newsarchive&sid=aCHizjnQq73E

Speedy New Traders Make Waves Far From Wall St.
JULIE CRESWELL
NYT, May 16, 2010  
http://www.nytimes.com/2010/05/17/business/17trade.html

Category: Markets, Trading

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

37 Responses to “No, Average Stock Holding Period Is Not 11 Seconds . . .”

  1. but, the Consensus says: “The Science is Settled.”

    worry not BR, just a flavor of a new/old Fad..

    http://search.yippy.com/search?input-form=clusty-simple&v%3Asources=webplus&v%3Aproject=clusty&query=Lysenkoism

  2. garo says:

    This shouldn’t be that hard to calculate. If we only want the average, look at the total float across all stocks traded on an exchange and divide by the average daily volume. That should give you the average hold time. I suspect that the median would be a more informative number.

  3. takloo says:

    BR… thanks for the myth busting!

  4. baychev says:

    well, TrimTabs estimates HFT to be comprising 80% of market activity. and the founder of TradeBot is not the founder of a small HFT shop… but the founder of BATS.

    so, 11, 15 or 20 seconds the problem is that stocks are no longer an investment tool but a traded commodity where price and not fundamentals matter. you may regard them any other way at your own peril.

    ~~~

    BR: BATS is huge — but Trimtabs has stunk up the joint lately

  5. globaleyes says:

    I saw it with my own two (global) eyes:

    He said “long term means after lunch” .

    He bought it and before I finished my spumoni, he sold that sucker.

  6. Guillermo says:

    It is important to look at the bigger picture. Many HFT programs are in the business of providing momentary liquidity to pick up rebates. Those kinds of operations are, by definition, short operations. If we don’t measure and count the holding times of market markers, we shouldn’t count those of HFT liquidity-providing programs. I am not arguing their liquidity is of good quality, that’s a totally separate argument, I am just arguing that if certain firms are acting as liquidity providers, we shouldn’t count their holding periods along those of investors.

    I’d also like to see a full distribution of this, something tells me the distribution of holding periods has a huge hump in the left and a big fat long tail on the right. If we look at the big picture, it makes sense. If some stat arb program trades 100,000 shares in one day and 10,000 investors hold their 1000 CLX shares for 2 years you’ll end up with 51,200,000 shares with a holding period of < 1 day and 10,000,000 shares with a holding period of over 2 years… see what i mean?

    the entire way of measuring the statistic is flawed

  7. MBD1120 says:

    Loosely related topic: when did it become the standard to judge how relevant something is by Google hits? BR, you seem to do this quite a bit, but are not the only one.

    Clicking on your link,you see that by the 4th page, the hits are already irrelvant to the search item. So, maybe the 11 seconds meme gets 30-35 hits when typed into Google.

    Sorry for the nitpicking, but I see this quite a bit and it is misleading.

    Great writing in general, keep it up.

  8. [...] Debunking the 11 second holding period meme.  (Big Picture) [...]

  9. NotQuiteSo says:

    Quite a bit of hard data has come out this year.

    This paper (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1641387) shows HFT involved in approximately 74% of all trades for the stocks in the author’s sample.

    This paper (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1686004) shows that inventory half-life of HFT firms in the e-mini S&P500s is less than two minutes.

  10. NotQuiteSo says:

    In the SEC/CFTC report on the flash crash, *before* the crash itself, HFT firms were involved in approximately 46% of all equity trades in a large sample of trading activity on May 6.

    http://www.sec.gov/news/studies/2010/marketevents-report.pdf

    (See footnote on page 45.)

    An important caveat which you can take as you like, the authors note that “However, we believe that the data provides a useful means to evaluate the extent to which large HFTs participated, and withdrew from participation, in the combined Nasdaq and OTC markets on May 6. This data should not be used, however, to estimate total HFT participation across all markets.”

  11. NotQuiteSo says:

    Sorry for the multiple posts, but Tradebot is not “one small HFT firm.” Tradebot was one of the first and most influential of the equities HFT firms, and it continues as one of the most significant HFT firms in the equities markets. Its founder, Dave Cummings, also founded the BATS Exchange, which is the fourth largest stock exchange in the US (out of 13). See http://www.batstrading.com/market_summary/.

  12. louiswi says:

    Conclusion: The 11 second number is wrong. There is no data supporting that.

    Your conclusion seems wrong. Just because there is no data, doesn’t mean the number is wrong. Further investigation is required. I know from one personal conversation, an HFT trader views it as “going out and doing a little mowing”. Does it add value anywhere. No! Is it good for the market in general? No! We used to tax the shit out of short term trading. Where is that now?

  13. constantnormal says:

    Isn’t there some “hard data” regarding the fraction of program trades that comprise all trading? I thought I had seen some tables from the exchanges on this matter some months ago … I know that all program trades are (probably) not HFT, but the point is that people/individuals play a pretty minor role in equities trading these days …

    As to the duration of the trades, I think the jury is still out on whether longer duration trades helps/hurts the market as a whole (it is possible, I think, that one might improve the responsiveness while at the same time making the markets less stable rather than more stable — there is likely a sweet spot of liquidity and stability somewhere that would be an attractive place to try and get the markets to operate at …).

  14. constantnormal says:

    Does anybody here have any positions that are a year old?

  15. VennData says:

    That “11 seconds” thing is Muslim, isn’t it?

  16. BigD173 says:

    As is not uncommon, the terms are not sufficiently defined.

    When you say “average holding period,” are you speaking of the period in which ALL shares are held, orthe period in which the shares that actually change hands on a single day are held on that day?

    Obviously, most shares are not traded on any one day. For example, AT&T (T) has approximately 5.91 BILLION shares outstanding, yet the trading in this issue averages only about 26 MILLION per day. So obviously the average holding period of all shares is MUCH higher than 11 seconds. Neither sophisticated math nor additional data is required to see that.

    But on a given day, it could well be that most of the 26 million shares (average) of T traded back and forth are being turned over quite rapidly. The data on this is likely to be very hard to get at, and whether 11 seconds is anywhere close to being the average holding period of the shares traded that day is very difficult to say. I don’t know where you would find the data to support or refute such an estimate.

    But the central point is, let’s be clearer about what we are trying to estimate.

  17. Lyle says:

    Actually HFT is the logical conclusion of the day trading. If you can make money trading on a minute to hour basis, now that we have gone to $.01 moves on stocks, then the faster the better. Plus computers make better short term traders than humans, not having to take breaks. The next thing is to get to millisecond holdings, and the transition to a short term casino is complete. Then only long term 1 year plus investing makes sense, as the noise from HFT tends to damp out with time. (Sort of like a plucked string, it quiets fairly rapidly). In that sense then the HFT just makes the moves that much faster, proving that humans trying to make money by trading is a futile exercise (resistance to the HFT trading is futile you will be assimilated to modify a phrase)
    And to answer constant normal I hold some stocks that have been held for 50+ years.

  18. VennData says:

    No, the dollar is not backed by nothing.

    It is backed by bonds held at the Fed. …dollar for dollar in fact.

  19. tradeking13 says:

    I first heard the 70% figure in this CNBC interview with Joe Saluzzi of Themis Trading.

    http://www.youtube.com/watch?v=_A28Zy9vR_A

    [He states the figure at 1:10]

  20. curbyourrisk says:

    Well…to say it is flat wrong…is wrong on your part. As you said there is no evidence either way. So, maybe it is, maybe it is not. Point of it is….the HFT trader is the market. And by saying the average trade is held 11 seconds…..makes that point perfectly clear. NO ONE IS INVESTING ANYMORE..

  21. kola.white says:

    I know guys who execute trades in less than 3 seconds. 11 seconds seems like an eternity…

  22. ToNYC says:

    In a bucket shop, Time only means that you lose over it.
    BTW, in what bar or club out on the East End do they use the word, “meme”?
    In other words, things happen, but not often enough for the reasons you think.
    11 seconds is way enough Time for a pick-pocket; and you still lose your lunch, no matter.

  23. TakBak04 says:

    @NotQuiteSo Says:
    October 28th, 2010 at 2:06 pm

    Sorry for the multiple posts, but Tradebot is not “one small HFT firm.” Tradebot was one of the first and most influential of the equities HFT firms, and it continues as one of the most significant HFT firms in the equities markets. Its founder, Dave Cummings, also founded the BATS Exchange, which is the fourth largest stock exchange in the US (out of 13). See http://www.batstrading.com/market_summary/.

    ————

    Well there we have it…the NEW ERA on Wall Street…and they might complain..”we didn’t see it coming” but it’s there and obviously many are aware of it… just not the MSFM (Mainstream Financial Media) who live top push us the latest ETF Conglomerate of other ETF’s and the Cramer’s and the rest of the Pump My Stock and Dump My Stock (but…I will give you a few “Winners” …just to keep you “teased back in the market,” and the Day Traders…who sadly don’t realize their time in the Sun is now going to “SunDowners.”

    Very Sad.. It’s beyond most of us except the most SAVVY …and who are they? They aren’t going to tell us..because they are raking in the $$$$$$$$$$$$’s BIG TIME. Why give away the secrets! The Markets belong to the “fleet of foot” not the “Old and Chartist, Timers and Cemtiroes Theorists of DOW and S&P)”…….It’s all changed..

  24. TakBak04 says:

    Excuse Typos….in haste (never answer in haste..you look like a fool) Frankly my Typo’s were so bad…I’ll leave it to look like a fool …shame…

    but I will correct:

    They aren’t going to tell us..because they are raking in the $$$$$$$$$$$$’s BIG TIME. Why give away the secrets! The Markets belong to the “fleet of foot” not the “Old and Chartist, Timers and Centuries of Theorists of DOW and S&P)”…….It’s all changed..

  25. dedalus says:

    Barry,

    You call TradeBot “one small HFT firm”??!!!

    Omigod. . . . Please stop. You’re a smart guy embarrassing yourself by that offhand remark; its akin to calling Pimco (or Calpers or OhioPers) “one small pension fund.”

    I know you’re on the East Coast but presumably you’ve heard of Getco (Global Electronic Trading Company), the firm that (along with TradeBot, Citadel & other HFTs) put the NYSE floor specialists out of business?

    FYI : In 2008 Citadel’s HFT group made $1 billion:

    “Malyshev led Citadel’s high-frequency trading group, with co-defendant Jace Kohlmeier as his top deputy. Together the men ran a department that made more than $1 billion in 2008, using computer programs to make rapid-fire trades across multiple trading venues.”

    Tradebot, Getco & Citadel by themselves probably trade one-third of all US stocks. (And an eleven-second holding period sounds a little too long to me.) Throw in another half-dozen or so HFTs and together as a group the HFTs probably represent 2/3 of all stock trading — more when volatility is low.

    http://articles.chicagotribune.com/2010-10-19/business/ct-biz-1019-chicago-law-20101019_1_jace-kohlmeier-citadel-files-computer-files

  26. fundaman says:

    Divide the total market capitalization by the average daily volume (in $). That will give the average holding time (in days).

    For delivery based transactions it’s typically one year. But for non-delivery transactions it should be much lower.

  27. [...] – Debating the average stock-holding period. [...]

  28. stat_arb says:

    So glad you’re bring *sources* and *facts* into the discussion.

  29. ToNYC says:

    Stop the bus! (…got to let my brother Jack, off.)
    Since it’s a few days before a major election, LET’S ALL TALK about suspicious packages that I get in my mail box and remember how lucky we are to be under security!..never try to change and screw up that delicate surgical extraction of your freedom, along with your 11 second real investments.

  30. stat_arb says:

    Another thing is, if you counted Dealers’ holdings on the actual trading floor — then the “average time a stock is held” won’t have changed as much.

  31. stat_arb says:

    @constantnormal Almost all of my positions are at least a year old.

  32. MikeW says:

    True?

    Since when have people based what they believe on whether something is true or not?

    People mostly believe what appeals to them, and the 11-second meme sounds good in the wake of all the mischief on Wall Street.

    It’s like the meme that J. Edgar Hoover was a semi-public cross-dresser, a preposterous claim based on the word of a woman who served time at Riker’s for an unrelated purgery and had personal reasons for slandering Hoover. The image struck a nerve among the public nonetheless, so it stuck.

  33. ToNYC says:

    Have you seen the look on JEH’s fat face at the pool with his constant companion Clyde Tolson?
    It pained him to use skinny sticks in his pina colada!

  34. ToNYC says:

    ouch!! Mike W Says:
    A first-time-in-this-here-blog double meme in consecutive sentences. Triple meme and you can get busy operatic practice. Fi-garo, Fi-garo, Fi-ga-ro.

  35. [...] is conducted by high-frequency traders who hold a stock for an average of 11 seconds.” is almost certainly incorrect: Conclusion: The 11 second number is wrong. There is no data supporting that. Estimates of HFT of [...]

  36. whatabute says:

    “I know you’re on the East Coast but presumably you’ve heard of Getco (Global Electronic Trading Company), the firm that (along with TradeBot, Citadel & other HFTs) put the NYSE floor specialists out of business? ”

    Not true. Technology put the “specialists” out of business. There are still specialists. They are called DMMs. They use technology instead of open outcry. They get a parity bucket instead of a look at an order before executing it or sending it on.

    The most important thing to realize here is that the market is very complicated. Most of you, although you want to, don’t know really anything about what you are talking about. Average hold time has decreased through the years as the market has become more efficient. Think about it. The faster liquidity is turned over, the more efficient the transfer of risk – in case you don’t know what this means, what im saying is, the quicker turnover leads to less margin for the trader in question, which means more money in the brokers pocket which theoretically should get passed to the investor. Unfortunately, because like i said before, the market is very complex and you dont understand it. Your broker does, but you dont. Right after you bought 100 shares in your ameritrade account for $10, ameritrade sold your order to Getcos dark pool for $0.20. Ameritrades $10 fee in the past was for the work/cost it took to execute your order. Now they just use it to advertise more and brainwash you more. The average hold time of Getco or Tradebot is likely sub ten seconds in Citibank. That number slowly grows the less and less liquid the stock is you are talking about. Make sense?