Exploring the tangle (part 2)
Deposits @Binance
In my previous article I demonstrated how we can recognize the exchange withdrawals in the Tangle. Binance in particular displays a recognizable pattern, which operates more or less as follows: a new address is filled up to a nice round number, usually 5 Ti. Then the withdrawals are distributed in bundles of 5 transactions at a time, about 50 times a day (where the change goes to a new address each time). As soon as the value of the hot wallet drops below 1 Ti, a new wallet is filled with 5 Ti and the whole story starts again from scratch.
For this (my second) article I have explored the deposits that ultimately are the source for the above mentioned withdrawals. In the past 4 weeks there have been 7 series of withdrawals which showed the fingerprint of Binance. I say fingerprint, because it cannot be explicitly deduced from the Tangle who is responsible for a certain transaction. But all kinds of characteristics of the bundles do reveal the character of Binance. It is however possible that another exchange shows similar behaviour, or that Binance uses other wallets with a different modus operandi. With that in mind, I have compiled the following.
By analysing the transactions preceding these 7 series of withdrawals I got a pretty good idea of how Binance handles its deposits. And there are a number of interesting aspects to that. Below I will first describe what the working method is, and then substantiate it with figures and references to specific transactions.
In principle a deposit with Binance remains at the address where the user deposits it, until Binance actually needs it to initiate withdrawals. As soon as Binance fills a new wallet (to be able to pay out withdrawals with it), the largest deposits are merged up to a total of 5 Ti (or actually a bit more, where the change is put aside for a while). Sometimes a few transactions of about 1 Ti together are sufficient (e.g. this one), but usually many more deposits are required (like here). It doesn’t matter how long ago these iotas were deposited, they are aggregated from large to smaller until the desired total is reached.
This has an interesting effect: before the start of a new series, the ‘whale’ transactions are merged, and these are always recent transactions (because in the previous series, about five days earlier, the same thing happened to what were then the largest transactions). This is followed by older but still large transactions. As a result, the threshold between new and old deposits is slowly getting lower and lower. I will illustrate this with a tangible example. On August 14 a new wallet is filled; the largest deposit is 800 Gi, which took place exactly 5 days earlier. The smallest deposit is 24 Gi, which was sent to Binance on 29 November 2018.
Four days later, on August 18, a new wallet is filled. First with 19 large deposits, varying between 26 and 1241 Gi, all deposited in the preceding days, and then (still in the same bundle, but with higher indices and timestamps) with 52 deposits between 23.5 and 24 Gi, almost all deposited several weeks or months earlier.
The large deposits are therefore always paid out first, while the small deposits remain invisible to the outsider. This makes it extremely difficult or even impossible to calculate the total amount of deposits. However, with some assumptions we can say something about the relation between deposits and withdrawals. To do this we first look at the distribution of the withdrawals.
The figure above shows the distribution of the iota withdrawals at Binance over the period August 5 to August 29. Because of the skewed distribution I use a logarithmic scale, which means that 6 is a Megaiota, 9 a Gigaiota, 12 a Teraiota. The smallest withdrawals are 0.5 Miota, which surprises me because I thought 10 Miota was the minimum for a withdrawal. The most common amount is 1 Gi. Apart from the peaks at round amounts (100 Mi, 10 Gi etc.) the distribution looks quite similar to a normal distribution.
If we now make the same plot for the deposits, a completely different image emerges. Of course it’s not right to compare the two plots (withdrawals and deposits) with each other, because they relate to a different timeframe. The withdrawals in the first figure are all made within a period of about 4 weeks. The deposits in the second plot relate partly to the same period (especially the large ones), but also partly come from a (much) more distant past, as explained earlier. It would be better to compare equal periods with each other. In doing so, we must also take into account the fact that we haven’t seen the small deposits of Binance yet.
In the two plots above we compare all deposits and all withdrawals exceeding 24 Gi in the period between August 8, 14:20:00 and August 27, 13:20:00 GMT. During this period the iota price decreased very slightly from $0.272 to $0.263. In this time and value range we have a full picture of the deposits and withdrawals:
* 85 deposits with a total value of 20.1 Ti
* 128 withdrawals with a total value of 13.9 Ti.
Unfortunately we only see the top of the deposits, so we can’t make any firm statements about the balance between all deposits and all withdrawals. Perhaps this will improve over time as the threshold drops further.
However, we can state that there are significantly more large withdrawals than deposits. In the period mentioned above, more large entities probably collected iota from Binance than put iota on the exchange. However, the total value that these parties have collectively absorbed is smaller than what has been deposited in this segment, and that indicates that a redistribution is taking place from large parties to smaller parties. Of course, a deposit does not automatically mean a sale, nor does a withdrawal always amount to a recent investment. I would also like to reiterate that this analysis is surrounded by all kinds of uncertainties and assumptions, but I think there is sufficient reason to believe that the above is true. However, there is no certainty, so use this information with caution.
Fun fact: prior to this research I thought that large exchanges like Binance made the number of addresses with value on the Tangle be smaller than the number of persons/entities that own iota’s, because the deposits are bundled. On the contrary: small deposits remain on the Tangle for a long time, and large deposits are redistributed to all withdrawals. In this way, Binance ensures that the number of addresses on the Tangle under Binance’s management is greater than the number of investors who have deposited iota on Binance. This is important to know when interpreting the number of addresses, as can be seen on https://thetangle.org/statistics/tokens-distribution!
Next time I want to look at other exchanges, at what happens in the chain between a withdrawal and the subsequent deposit, and at all kinds of other transactions that we can distinguish. And if there’s time, I will continue to monitor the deposits and withdrawals as shown in this article. If you have any questions or suggestions, please leave a message below!
Disclaimer: For this analysis I have assumed that Binance works systematically and in the same way. For the time being, everything indicates that this is indeed the case. However, it is possible that in certain situations this method is deviated from. This can lead to errors in the above analysis.
