Aims, tasks and assistance to the project Active position: What to do? Chronology of fighting the fraud of Binance
Soon: Guangying Chen (Heina Chen) About, questions-answers, plans, events (2021-2023)

Sylvian Ribes: How top exchanges inflate volume: Binance, HitBTC, Okex, Huobi

We recommend you to read:
- Binance's Scary Truth: Trading volume falsification as exemplified by Eth2 stacking by 60 times the real volume

The article is from 2018, but its subject vividly demonstrates the "business of casino exchanges" such as Binance in 2021. In current realities, it is quite possible to declare that 90% of trades (and 99% in alts) are fake artificial volumes, you can add "FTX Kitchen" - similar casino to the list of exchanges. The situation is different for BTC market, but there are also enough serious percentage of fake trading activity - that's hidden reality, which they don't want to show to the public.

Everyone remembers when Okex froze cryptocurrency withdrawal due to owners' arrest, and how it affected crypto market? Absolutely nothing, BTC rate didn't fall/increase - no reaction at all. The question is why? Because, Okex - falsifies its volumes up to 95%, just like other casino exchanges Binance, HitBTC, Okex, Huobi and FTX. The true transaction volume is on the CME and others like it.

But, the situation in 2018 and now (2021) is no different - only the amount of falsified volumes is no longer $3 billion, but probably more than $30 billion, and the volume of fake trades - as were in the area of 95%, and remains so.

Let us bring to your attention a translation of an article by trader and investor Sylvain Ribes:

In this post I will demonstrate why I think that more than $3 billion of all cryptocurrency trading volume is fabricated and how OKex, the №1 exchange by volume, has become the main offender, because up to 93% of its trading volume doesn't really exist. I will try to prove it by analyzing open data.

When I started collecting data for this post, I had no idea I would be writing about fictitious volumes. At first, I planned to collect data on the liquidity of cryptocurrency assets, which could complement trading volumes. I thought I would have an interesting indicator to analyze the value of the asset.

I decided to collect quotes stacks from major exchanges for this purpose and check how much the price of a particular cryptocurrency collapses if it is sold for $50,000. In this article I use this figure as a "slippage" (slippage) (the definition of the term is in the appendix). I then adjust the slippage figure by selling more or less on each exchange, depending on its volume, and varying the sales volume based on the market capitalization of the cryptocurrency.

I would expect slippage to be a decreasing function of volume, with the fact that, depending on the particular currency, there may be some differences. At the end of the day, if you have huge trading volume on a particular pair, there should be a lot of competition between market makers to fulfill the bids of active buyers and sellers. This competition should flatten the quote stacks and reduce spreads.

Am I right?

It turned out that this was an obvious trend, yes. However, where I expected small differences between specific cryptocurrencies, I found huge discrepancies between exchanges. Not at all the kind that you can just manually dismiss ("sort of, it's just their users behaving differently"), but the kind that can only be explained by the fact that some numbers are inflated to 95%.

The leader of the group is OKex, №1 in both CoinMarketCap and LiveCoinWatch rankings, with a total trading volume of $1.7 billion.

OKex is a ghost town

Explanation: Slippage = f(Volume), OKex, Kraken, Bitfinex, GDAX

The graph represents the average slippage volume and trading volume for all pairs of accounted cryptocurrencies with daily trading volumes over $100,000 on four major crypto exchanges over 24 hours: OKex, Kraken, Bitfinex and GDAX.

For example, you may notice that the blue dot at the bottom of the chart represents a pair on the GDAX exchange with trading volume of about $200 million, with a slippage volume of less than 0.1% The chart is simply amazing. It shows, although all of the first three exchanges behave similarly, how all of the pairs on the OKex exchange marked in red exhibit much higher slippage volumes relative to their total trading volumes. As I explained earlier, this can only mean one thing - most of the volumes on OKex are completely fabricated.

In addition, I decided to remove all slippage data greater than 4% from the array for readability. If these data are entered, the OKex graph will be even more absurd and a logarithmic scale will be required:

Explanation: Slippage = f(Volume), OKex, Kraken, Bitfinex, GDAX? -logarithmic scale

If you decide to sell only $50,000 worth of assets, the slippage will be more than 10%, although such pairs show trading volumes up to $5 million. During the analysis (06/03/18) the following pairs were considered - NEO/BTC, IOTA/USD, QTUM/USD. Highly or low-liquid assets.

Although these figures by themselves certainly prove that the vast majority of trading volumes on OKex are rigged, I have not personally seen how they do it. So, I went to their platform and looked at the trading history of some pairs. They do falsify volume, but in a ridiculously obvious and artificial way:


Explanation: Volume = $1 billion. * sin (period).

Compare this perfectly neat graph as a continuous sine wave to what happens on a real exchange:


Rapid rise, sharp fall, avalanche effect on high volatility. Not at all like an oscilloscope chart like in school.

"But it's about day and night cycles in China!" I don't think the above chart speaks highly of OKex engineers, who really thought about how to make fictitious trading volumes less noticeable than just constant transaction flows, but all they got was a flawless sine wave.

How bad is it?

Although it is obvious that most OKex volumes are falsified, how to calculate whether 90%, 95% or 99%? I suggest using the following method:
  • Make a list of reliable exchanges, which stably operate without such violations;
  • Make a regression on their data sets in such a way as to be able to predict the trading volume based on the observed slippage;
  • Compare the reported OKex volumes with the volumes calculated according to our model.
For my calculations I used data from the following exchanges: Bitfinex, GDAX, Poloniex, Bistamp, Gemini, and Kraken. Taking into account the volatility of data sets with low trading volumes, I also decided to change the indicator used: instead of analysis of sale for $50,000, I modeled with $20,000.

Here's what the data from the reliable exchanges look like, this time showing volume as a function of slippage:


Explanation: Volume = 4.4/Slippage? -?5.5

Note, due to very volatile data any model becomes absurd when slippage exceeds 0.7%. The model proposed above works best for slippage below 0.7%. After this threshold, the only reasonable assumption is that the expected volume is less than $1 million.

As you can see, if you enter OKex data into the array outlined above, the picture comes out all wrong:


Out of 28 pairs of selected cryptocurrencies with trading volumes over $100,000, only 11 showed slippage of less than 0.7%, namely:


Explanation: OKex data, estimated % of falsified trading volumes

The overall ratio of falsified volumes for these pairs is, according to the model, a shocking 93.6%. If we add remote pairs (slippage > 0.7%), the situation does not change much in either direction. Perhaps the regression I used does not work very clearly with large volumes, as there are no reliable data sets available. In that case, the only honest solution would be to remove BTC/USD. The figure is still high to the point of absurdity: about 92.9% of all OKex trading volumes are most likely falsified.

Huobi exchange, not many differences

Just like OKex, Huobi was closed as a result of strict legislative measures taken in China, but then reopened under the license. Using the methodology described above, we got the following results:


Explanation: Huobi data, estimated % of falsified trading volumes
81.8% of volumes are falsified, not as bad as the nearest competitor, but still a lot. If you look at the trading history on Huobi, it is immediately confirmed that while they look much more organic than on OKex, the constant, not particularly conspicuous, fake trades still provide significant support to the overall volumes: 
Explanation: Real volumes *don't* stick to any constant baselines
The Chinese armada of swindlers 
You may or may not have noticed, but just recently CoinMarketCap added a group of Chinese exchanges, all of them showing quite high trading volumes, but for some reason no one has ever heard anything about them. Apparently, most of them use the same user interface and trading engine.
Although the list could go on, first of all I mean the following exchanges: Lbank, Exx, RightBTC, CoinEgg, Zb, BitZ, Bibox, CoinEx, BTC-Alpha.
These platforms are so unceremoniously falsifying their trading volumes that it's not even worth analyzing them using a model, and you can see for yourself... It's just a shame that CoinMarketCap and LiveCoinWatch put these crooks in the same row as law-abiding exchanges, which sometimes have a bad time.
HitBTC and Binance Casino 
For a number of reasons, I was suspicious of the two altcoin trading leaders, HitBTC and Binance. Here's what their results are compared to a few "decent" exchanges:
Explanation: HitBTC and Binance compared to model exchanges
It's easy to see that in certain volumes, both exchanges, especially Binance marked in orange, are significantly less liquid and therefore suspect.

Doing the same analysis as before with OKex and Huobi, we get the following results. First HitBTC:
Explanation: HitBTC data and calculated divergence volumes
The above figures do not seem too significant, although they do prove that HitBTC is somewhat less liquid than the sample exchanges. This small difference between reported and recalculated trading volumes could appear for a number of reasons, including a simple deviation from the general norm.
Binance's results, however, are more interesting: 
Explanation: Binance data and calculated discrepancy volumes
The difference with our mathematical prediction of 70% is worrisome. But let's not forget that the input data for our model is the slippage on a particular pair, which is not necessarily caused by the whole trading volume.
Of course, I know from my own experience that Binance restricts trading using robots rather strictly. For some time I even argued with them, trying to convince them how silly these restrictions are, because they only hinder the growth and liquidity of the exchange.
It is possible that because of these restrictions, many people who use market-making strategies on several exchanges do not use them on Binance, because they will be permanently blocked without even knowing what limits cannot be exceeded.
Without a few professional market makers, it is easy to see how the stacks of quotes will become thinner and my model will no longer work. Nevertheless, it may come in handy for keeping a close eye on Binance's reported trading volumes in the future, although analysis of trading volume history does not reveal any obvious suspicious activity.
Although I have virtually no doubts about my statements, the numbers should not be taken lightly. And here's why:
  • As I noted, with respect to Binance, restrictions on the use of trading robots play an important role. It is easy to use better robots that improve liquidity on specific exchanges.
  • Commissions can also have an impact. The larger they are, the less market makers will exceed their quotes and reduce spreads.
  • I have only collected an average sample of 24 hour data and have not controlled for outliers. I am not a scientist, nor am I a fan of statistics, nevertheless everything looks like the results obtained are quite reliable. I am willing to listen to other arguments.
  • Icebergs and hidden orders. Some of the exchanges we have reviewed may offer their users the ability to hide their limit orders. At the same time, considering that Bitfinex also offers such options and at the same time works almost the same way as the other "model exchanges", I believe that the volume of liquidity in question will not change much if we put aside the influence of icebergs and hidden orders.
  • Different groups of users may behave differently on different exchanges, although, based on my personal experience of algorithmic trading, I can say that the influence of such differences on the overall situation is at least not systemic.
So, should we bother? 
Some might argue that "given the fact that the market is unregulated, such actions are not even illegal, so why shouldn't the exchanges do what they want?" and they would be wrong. Precisely because the market is unregulated, responsible behavior benefits the players in that market themselves.
By spreading the word and boycotting exchanges that use unfair practices is the least we can do. Some may say that "they are not hurting anyone," and they would also be wrong. First of all, by inflating their trading volumes, they get an opportunity to defraud gullible common investors. What's more, if you're not even a gullible ordinary investor or a venture capitalist, these actions can and do affect the valuation of some cryptocurrencies. In particular, up to 75% of Bitcoin Cash and Litecoin trading volumes are fixed on one or another fake exchange. Only OKex steadily trades more than 30% of each of the above-mentioned currencies.
Explanation: Markets for Bitcoin Cash on the day of its latest plunge
Showing mostly artificial volumes, these currencies seem more interesting to traders because they attract much more attention than they actually deserve (volumes are a good cure for volatility). Also, if there are consistently rigged trading volumes at the top of the spread, people tend to become somewhat overvalued by such cryptocurrency.
Dishonest exchanges are also more likely to attract more customers than honest ones, with more realistic volumes and liquidity, while allowing users to trade faster and/or lose less due to slippage.
Finally, falsification of trading volumes, even if not strictly speaking an illegal operation, may well be a precursor to further, more serious violations, so all users should exercise extreme caution when deciding to trade on such exchanges.
By my calculations, more than $3 billion in trading volume is unrealistic. Probably more. For some reason, this practice, if not encouraged, is at least ignored by popular data aggregators and most of their users, when all one has to do is look at the numbers to see if something is wrong with them.
Right now, crypto-assets are under serious bear market pressure after the bull rush of 2017. I am convinced that growth will not continue until there is a sufficiently healthy trading environment. The capitalization of the entire ecosystem and awareness of the cryptocurrency market has grown so much that such blatant manipulation simply cannot be allowed.
We all keep saying, "Cryptocurrencies don't need regulation!" Now is the time to prove it. Otherwise, the current state of the cryptocurrency market seems to confirm that the free market does not work.
P.S. Addition 
My definition of slippage is somewhat incorrect. I define slippage as the percentage change between the observed average price between quotes and the lowest price at which I would agree to sell the asset.
For those who want to examine the data I collected on their own, here is a link to the raw data, sheet 1 I used the CCXT Python library at to collect the data.
Finally, this post does not list all the exchanges that I think are falsifying their trading volumes. At a glance, if you just look at the charts of trading history Tidex, Liqui and Wex are up to it, although I haven't counted how much their volumes are falsified (although I haven't counted them, they are probably heavily falsified). There are probably many more of these exchanges. If you analyze the futures contracts on OKex, preferably with a different methodology, there are bound to be surprises, because OKex claims huge volumes of futures. шаблоны для dle 11.2

Search by keyword: