A user of Reddit used Google Consumer Survey to survey the U.S. population about their GameStop ownership. He used a set of 3 surveys, with a total of 2200 respondents and with the question "Do you own shares in the company GameStop ($GME)?". Link to the post

The survey reported that 5.59% of the respondents had shares, with an estimated average share count of 34 (conservative). After processing this data, he reached the conclusion that at least 163.66M GME shares are owned by the US adult population. (Best estimate would be 382M shares owned by the US population, and a total of 450M+ shares worldwide). The reddit post has a detailed explanation of how this data is processed and interpreted, and to the best of my knowledge it is correct - please let me know if you think otherwise.

As per Yahoo finance, GME has 74.38M Shares Outstanding, so the number of shares own by shareholders would be 3x - 6x of the shares outstanding, if that makes sense.


  • Can a stock have more shared than reported? How can this happen?
  • What does this mean for the stock and its shareholders?

Link to surveys Survey 1 Survey 2 Survey 3


  1. These surveys were done during the June-July 2021 period, so well after the Jan-2021 squeeze.
  2. 2200 samples are more than enough to accurately represent the US population. 1000 Samples would enough to have a 3% margin error. You can find more info in Google, e.g. here or here.
  • 5
    You can’t possibly think a survey of 2,200 Reddit users accurately represents the approximately 252 million adults in the US. I’d say the discrepancy of 3x-6x GameStops audited financial statements indicates that there’s something wrong with the survey not that there’s some missing information to learn.
    – quid
    Jul 21 at 3:45
  • @quid 1) The survey is not of Reddit users, but the US population. 2) yes, a sample of 2200 can accurately represent the US adult population (2200 is actually overkill) . There is plenty of info in google, e.g. scientificamerican.com/article/howcan-a-poll-of-only-100
    – Daniel
    Jul 21 at 8:13
  • 3
    I strongly suspect that there is something that's skewing the results of this survey, even if right now I can't put my finger on exactly what it is. The results just seem too high to be plausible. (It is a pity they did not take the sensible step of asking a control question - do you own stock in some dull industrial company - and seeing if that also gave an unexpected result)
    – Andrew
    Jul 21 at 11:35
  • 1
    For example, while about half of American households have stock-market investments, only about 14% (in 2016, latest data) have individual stock holdings rather than eg mutual funds. Even assuming that's jumped since 2016, which is plausible, it's still implying that a very large proportion of individual shareholders were invested in one particular company...
    – Andrew
    Jul 21 at 11:51
  • 3
    I’m voting to close this question because not vaguely in the bounds of conventional numeracy.
    – Fattie
    Jul 21 at 20:18

Rehypothecation could conceivably result in a daisy chain of investors 'long' and 'short' on the same GME shares, as in the below scenario:

Suppose an investor bought some shares of a corporation and the broker-dealer loaned the investor’s shares to a hedge fund for short selling; and this hedge fund sold the shares to a third investor; this investor’s broker-dealer loaned the shares to a second hedge fund, and so on, so forth. And if this rehypothecation happens sufficiently many times, the number of short-sold shares of this corporation exceeds the number of its available shares or, in Wall Street parlance, short interest in the shares of this corporation as a percentage of float (SI % of Float) exceeds 100%.

If you are 'long', your shareholdings (the shares themselves, not any particular dollar value of the shares) are protected by SIPC. Any bad effects that rehypothecation might have on your broker will not result in your shares disappearing.


I don't trust the results.

By reading the links about the survey and the analysis: Using Randomized, Representative Surveying Data to Model $GME Ownership Among the U.S. Adult Population

It should be highlighted that this is not scientific research, and we’re not necessarily seeking a high level of precision in the data. A margin of error of 4-6% is certainly acceptable given the “tip of the iceberg” nature of the research, and the aims of the original thesis.

Yes you can use a survey of only a few thousand people to represent the US. But the question is does that collection of people who responded accurately represent the US population. It isn't enough to check age, race and where they live. It also depends on the makeup of people who would have access to the survey.

I also had a few problems with the questions as asked.

They asked if you owned shares. I have no idea if I own shares. I invest in several mutual funds. They may have owned shares. So should I have answered Yes or No? If I answered yes I would have has to guess how many shares I owned, which would have involved knowing what percent of the fund was tied up in Game Stop.

Also if a person shorted the shares, did that mean they should have answered yes or no? The people shorting the stock didn't own the shares they borrowed the shares. If a person lending shares also answered the survey were they supposed to answer yes of No. Easy to double count here. Looking at the investors via a fund, I have no idea if my fund shorted the shares, or was a lender.

This was said about the original survey:

That said, this research includes the participation of 300 individuals. Assuming a confidence level of 95% (meaning 95 of 100 survey respondents will provide a truthful and accurate response), this research has a margin of error of 5.66%.

A 95% percent confidence level doesn't mean that. It means that you have calculated that the data you have collected gives you 95% confidence that you have proven the hypothesis.

  • "meaning 95 of 100 survey respondents will provide a truthful and accurate response"... that is an awfully big assumption from a community that is highly biased. Without any other supporting evidence, I might accept if someone said that 5% of reddit users surveyed were truthful. Jul 21 at 16:03
  • @mhoran_psprep thanks for your feedback. What else would you control in the survey, apart from age, race and where they live? And why do you say it's not enough? As far as I know it's quite standard. (and yes, you can add more parameters and make it more accurate, but this 4%-6% should be enough )
    – Daniel
    Jul 21 at 22:48
  • The point about what should answer people that hold funds is a very good point. I would also not know what to answer myself, although I think at the end I would have said "no", so I don't have to count. I wonder how much this affects the survey.
    – Daniel
    Jul 21 at 22:52
  • Regarding the shorted shares point - that doesn't make sense to me. If you shorted a stock, you obvioulsy don't own any share of it, do you? Quite the opposite.
    – Daniel
    Jul 21 at 22:54
  • The survey with 300 samples is the smalles of all, and with the bigger margin error. Google gives this one a RMSE Score of 6.2%. The bigger one with N=2200 has a score of 3.9%. Even with a 6%, or even 10% error, that does not change the conlusions much
    – Daniel
    Jul 21 at 22:58

When shares are shorted, it creates synthetic shares. For example, if 100 shares are loaned for shorting, there are two owners of 100 shares for a total of +200 shares and one holder of a -100 share short position. Only one owner possesses the actual physical shares.

During the short squeeze, the short interest for GME was reported to be about 140%. So using your Yahoo number of 74.38M shares outstanding, 140% short would be about 104M synthetic shares. Add the physical share number and it's a total of about 178M shares. 382M to 450M shares seems a bit much.

Here's an article about this.

  • 1
    The survey was done in June 2021 - does that affect your answer? It seems unlikely to me that the short interest would still be so high several months after the initial squeeze. Jul 20 at 23:42
  • 2
    I doubt that the survey is accurate. If done in June, probably even less so since as you stated, short interest is likely to be much lower. Truth be told, I'm just speculating since I have no idea what the actual short interest peak was. Jul 21 at 0:10
  • @BobBaerker. If you think the survey is not accurate, can you explain why? I agree that the results look extraordinary, but to my knowledge the surgery is perfectly fine.
    – Daniel
    Jul 21 at 9:55
  • 1
    A survey is an estimate. It is not a precision analysis. Are the results accurate? I don't know but given factual data from Yahoo Finance (short interest numbers) from February, I think not. Jul 21 at 13:20
  • Of course a survey is an estimate, and of course is not a precision analysis. The question, or the post, never implies that, and it actually talks about error margin. The question you need is: is the margin error acceptable? We are talking about huge discrepancies here, and even a big error margin (which I still can't think would be the case) wouldn't change this big discrepancy.
    – Daniel
    Jul 21 at 23:04

Not the answer you're looking for? Browse other questions tagged or ask your own question.