So, Have We Got Tethered?

Oleksandr Ivanov (aka cryptofan) is a crypto analyst, data scientist, and PhD candidate at University of Groningen. In his short research he applies the critical analysis of recent accusations that Tether supply has been manipulated in order to use pump-and-dump scheme on the price of Bitcoin.

Over the 2017 year the price of bitcoin skyrocketed.

So did the amount (equals to the market cap) of tethers (USDT).

Some people made a causation link between these two. Briefly speaking, here is the plot:

  1. Bitfinex (a cryptocurrency exchange and the owner of Tether) prints millions of tethers out of thin air.
  2. Uses tethers to pump the price of bitcoin with margin long trading.
  3. Dumps bitcoins for US dollars.

For more detailed explanations see here and here for example.

At some point the discussion whether or not Tether influences the price of bitcoin became one of the hottest topics in crypto community. Some people, like the person behind the twitter account Bitfinex’ed, actually made a name by speculating on this question.

So have we all got tethered? I approached this question from a statistical point of view. If the alleged Tether strategy is true, we should see a positive correlation between the change of the amount of tethers and the change of bitcoin price at some timeframes.

First, from the Coinmarketcap website I took the historical data of bitcoin price and the amount of tethers in 2017. Then I calculated the daily percentage change of bitcoin price:

and the daily change of Tether‘s market cap (the amount of tethers):

Clearly, Tether had some rock days, when the amount increased over 40% in one day.
Next, I plot the bitcoin daily percentage change in price versus the amount of tethers daily percentage change. Most of the points center around zero. Some points on the far right corner of the plot correspond to large percentage changes in the amount of tethers. These are the outliers. Interestingly, they mostly lie below zero in the bitcoin price percentage change axis.

Pearson correlation coefficient for bitcoin daily price percentage change and the amount of tethers daily percentage change is -0.17168 (p-value is 0.001, which is statistically significant). So there is a negative correlation between the change of the amount of tethers and the change of bitcoin price. Visual inspection of the plot about indicates that tether outlier points may have a strong influence on the correlation. Removing tether outliers (daily change of market cap > 20%) leads to the correlation coefficient of -0.08335 (p-value is 0.1159, which is not significant).

A valid argument is, what if tether is not used to pump bitcoin price immediately but after some time? We can statistically verify that by calculating the percentage change of bitcoin price for some number of days with a sliding window. This is similar to the moving average, but instead of the average, the percentage change is calculated between the first and the last day in the moving window. By varying window size in days I can calculate if the change of the amount of tethers correlates with bitcoin price within some number of days (I call it a lag). Here is the result (without removing tether outliers):

The correlation in all cases is still close to zero. In conclusion, this simple statistical test does not support the hypothesis that the creation of new tethers pumps the bitcoin price.

How come that the Tether controversy became such a hot topic in the bitcoin community? Let me speculate myself here. I personally attribute this to the confirmation bias. As soon as new tethers are created and the price of bitcoin within a short period of time goes up, people that believe in tether conspiracy theory present such an event in support of their theory. Whenever there are new tethers created and bitcoin price does not move or goes down nobody bothers to talk about this. In the same way, we know about almost every single airplane crash event and about very small percentage of car accidents events. It is much more safe to fly than to drive a car but people are generally more afraid of flying than driving in a car.

My analysis does not generally disprove that there are no manipulations with Tether. However, in that case, the plot may not be so simple as I outlined at the beginning of the article. Transparent public audit of Tether will shed the light on this issue.

  • Charlie Offenbacher

    Does the second half of your analysis include when the mass tether printings occurred or were they removed as outliers again? They are overwhelmingly the most important part of the data, and also most likely to not be spent in a same-day pump, since the numbers are astronomical.

  • Apples and oranges. The allegation is that bitfinex bought Bitcoin with tethers they created out of thin air on other exchanges to start rallies and create buy walls to keep the price moving upwards, then cashed them out during pull-backs and kept the cash. Only a very naive person would deduce that they have 2.2 billion USD’s in the bank that was deposited from customers. They might have 2.2 billion, but the accounting will show it was created by creating fake bitcoin and cashing them out. If they did that, they committed fraud. The US gov will audit their books and throw them in prison if that was the case, and crytpo will be cripped for a year or two until the average person regains faith in the institution. This week, in Congress, we will probably hear about tether and Bitfinex and what the US gov plans on doing to rectify that situation (indictments or trial)

    • cubanlove

      not possible as the US has no jurisdiction in Taiwan

      • Read the case of Liberty Reserves, and read about the powers of the Patriot Act. Any financial company in the world that claims to store or receive the US dollar gives the US gov jurisdiction to the USA. So Tether and Bitfinex might be incorporated in the Virgin Islands and have offices all over the world, it doesn’t matter. The long arm of the US gov can reach out and touch someone anywhere at any time if they claim to hold their sacred US dollars.

        There are literally 100’s of cases of foreign nationals and companies being put in US prison for scamming investors who paid with US dollars. US dollar is king, so it is used all over the world, and that gives US gov jurisdiction. Other currencies have the same rights. If a US investor scams someone with the Euro the EU would have jurisdiction and the US would arrest the crooks and send them for trail in Europe.

        • Let me put another way. Tether and Bitfinex owners claim to hold $1 USD for every tether. If they created tethers illegally out of thin air to create a 1.4 billion US dollars (the number I’ve read is most likely) and then used that money to buy Bitcoin, the owners will go to prison for fraud. US gov sees this in the same light as counterfeiting and the laws against that are very severe.

          • Леонид Морозовский

            I wonder what kind of fraud is that?
            For example, some company issues shares, sells them for bitcoins, then sells bitcoins for USD on account. Why is it a fraud? As long as assets balanced with liabilities it seems ok. So if real dollars match USDT tokens on the account everything seems to be ok.

      • sealclubber

        SWIFT

  • Is Bitfinex an anti-Wall St anti bank type company or Wall Street greed on steroids? I say the latter. Any investigation into the owner will reveal that Phillip Potter, one of the owners of Bitfinex, was a Wall Street reject that exemplified greed: http://www.nytimes.com/1997/10/19/business/faces-of-the-new-york-economy-big-earner-big-spender-little-burden.html

  • vdrhtc

    Excuse me, but what about this? Can you reproduce these results?

    http://www.tetherreport.com/

  • Donald Ralph

    I am not sure if i should be saying this here but i guess i want to share anyway. I started investing my bitcoin with this amazing platform http://www.simplecoinage.org, i invested 0.1btc and got 0.3btc back in 24 hours. They have really helped me a lot through this dip and sudden unpredictable time of bitcoin.

  • Author should have used weekly or monthly data points. Its not like tether printing and bitcoin buying happens on exactly the same day.

  • Ilya Bibikin

    The course is lowered by the stock exchanges and will be lowered. The exchanges benefit from high and frequent volotility, more transactions – more fees, more funds to lower and sow panic, and this is like an avalanche. The only solution is not to sell, but this is a market, you can not not sell, someone can do it. Who doubts and will talk about bans and government regulation – look here https://bitinfocharts.com/ru/bitcoin/address/3D2oetdNuZUqQHPJmcMDDHYoqkyNVsFk9r. For the year, almost 10% of the emissions of bitcoin, and this is only bitfinex. If you add all the exchanges, it turns out that the exchanges control the rate. Fiat – banks, crypt – stock exchanges.

  • One Of These Wolves

    I feel like comparing the daily change in tether market cap to the daily change in bitcoin price completely is not the right approach here.
    It completely ignores how the market was moving at that moment.

    For example, if the bitcoin price is moving down at -0.2%, of course one tether printing/buying run wouldn’t suddenly turn that into +0.2%. But it might cushion the drop.

    So what you should do is check how much the price changed after a tether printing run in relation to the price movement right before that.

    Btw, I’m also wondering about how you derived your data.
    Shouldn’t there be some days with a lot more than just 0.2% change?

  • G$

    Cryptofan –
    Strictly speaking your analysis is invalid. The Pearson correlation test you use is valid under two assumptions:
    1. the two variables have a linear relationship
    2. the two variables are normally distributed

    Regarding assumption 1, your graph gives a decent linear relationship between the % change in BTC and % change in Tether market cap. However, you fail to test the assumption that these two variables are normally distributed. So I did this for you.

    We can first use the Shapiro-Wilk normality test. The results are:
    W = 0.96893, p-value = 0.004646
    We see that the two p-values are NOT greater than the predetermined significant level of 0.05, implying that we reject the hypothesis that the distribution of the % change of BTC is normal.

    We can test for normality in another way, using qqplots (see below). It would be lovely if the observations fall on the straight line or within the gray area of tolerance. But clearly the upper and lower quartile of data do NOT map. This is important because these represent the days when the price of BTC changed significantly, either up or down.

    With this we can conclude that the % change of BTC is NOT normally distributed and thus you have no justification in using the Pearson correlation test to make your claim. You could however look into using a Kendall and Spearman test which does not require the variables to be normally distributed. I’m working on that analysis.

    https://uploads.disquscdn.com/images/82ff32859b31fe1ba280a8bb1e975a0dcc2a963fc5f388793b63e2aace56e92d.jpg

    • LinusVanPelt

      Thank you for your contribution, can you please share your results?
      The results from this analysis look opposite from http://www.tetherreport.com/, many are interested to know why.