Hacken Token
$ -- --.--

Unusual A/B Testing: Which Trading Volume Manipulation to Choose?

Nowadays, we often hear the crypto industry referred to as the “Wild Wild West” by those in the traditional financial and business sectors. This is in part, due to the immaturity of the blockchain field, whose infancy results in a lack of regulations, business dealing standards and requirements for those companies operating in the market.

As a result, it’s possible for “smart people” to earn “easy money” by manipulating less educated and easily led participants in the market. This behavior negatively affects the prosperity of the blockchain field as a whole since it causes serious market players, including investment banks and institutional investors, to avoid crypto and crypto related businesses altogether, considering them to be a ‘scam’ or a bubble just waiting to burst.

Logic suggests that all well-established crypto businesses, including the biggest exchanges and token issuers, would likely be the most interested parties in the continued development of the crypto field. It is imperative that they take all possible measures to convince the rest world that blockchain and cryptocurrencies are undervalued and sustainable.

But unfortunately, this is not the case…

There are entities that prefer to make fast money whatever it takes. One of them is the biggest Korean crypto exchange – Bithumb. The exchange applies various trade volume manipulation techniques and then uses a multi-factored approach to conceal these manipulations.

Bithumb, established in 2013, is one of the oldest South Korean crypto exchanges. It provides crypto to fiat trading for 66 coins against Korean won (KRW). Currently, it is the second largest exchange globally with $1.4bln 24-hours reported volume according to CoinMarketCap (CMC).

Let’s get down to business.

Read also our investigation on two other Korean exchanges manipulating the trade volume

Bithumb Success Overview

In September 2018, Bithumb was at the bottom of the CMC Top-10 with about $350mln of daily volume. However, during the second week of October its 24h volume increased to over $1bln and continued to rise right up to November 11th, peaking at about $4.4bln. Such a significant grow eventually led Bithumb to overcome Binance and Bitmex in order to top CMC rank by reported daily volume.

Fig 1 shows Bithumb sitting on top of the CMC rank with $4.1bln 24h trade volume on November 9th, which is more than 10 times greater than its average daily volume during the summer. Due to these figures, we decided to investigate the trading activity on Bithumb.

Trading Activity Analysis

We examined the exchange’s 10 most active coins: BTC, LTC, ETC, XMR, ZEC, DASH, BTG, OMG, QTUM, and WTC. In particular, we reviewed their charts over different time frames retrieved via API and analyzed the historical trade data from June 1st to November 22nd. We also calculated the correlation between trade volume and price volatility and reviewed transactions during particular periods more closely.

To begin with, we calculated the Price-Volume Correlation (PVC) by the 10 most active coins using the methodology from our recent research and we received the following results (see fig 2).
Fig 2 shows the PVC falling since July. In September the metric entered the red territory (values below 0.5) that we consider influenced by non-market mechanisms. The obtained results suggest a high chance of foul play taking place on Bithumb since the lower the PVC value the more likely trade manipulations have taken place.

So, let’s dig deeper into the charts and trade data.

BTC market analysis

While looking on the BTC daily chart it’s hard not to notice a significant rise in volume that started on August 25th and then abruptly dropped on November 12th (see fig 3).
Fig 3 shows the BTC daily trade volume increase tenfold in September from 3.2k BTC on average from the beginning of the summer. Such volume performance looks very strange considering the absence of fundamental factors and the fact that we are not looking at a newbie trans-fee mining exchange but at the largest South Korean crypto platform by trade volume.

We must also note that our analysis suggests that the trade volume has been intentionally inflated on Bithumb starting from the end of August and continued even today. We can divide this 3-months long “volume pump” into 3 periods with different characteristics of trading activity specific to each of them:

  • First period – from August 25th to October 7th
  • Second period – from October 8th to November 11th
  • Now ongoing third period started on November 12th

The first period of Bithumb’s volume pump is characterized by a peculiar trade volume performance featuring an activity spike during the first few minutes of an hour (11 a.m. local time each day) delivering 90-95% (!!!) of daily total volume. Such synthetic trading activity has formed a peculiar “comb-like” pattern on the hourly chart (see fig 4).

The second period is characterized by irregular trade volumes not aligning with price moves. There are multiple volume spikes in low volatility periods and sharp price swings on much lower volumes (see fig 5).
The pump in the “second period” lasted for over a month intensifying closer to its end as daily trade volume gradually rose from 22k BTC (October 8th) to 87k BTC (November 11th) with a peak of 106k BTC on November 6th. Then the flood has stopped and daily trade volume dropped by more than 56 times to 1,538 BTC on November 12th.
BTC trading activity returned to its natural level on November 12th, since even a dramatic fall in price and increased volatility couldn’t drive volumes to previously pumped levels (see fig 6).
Onthe third period of the pump starting on November 12th, BTC trade volume shows a fairly natural performance: increasing when volatility rises and spiking while the price goes through lows or moves sharply (fig 7).

While looking deeper into the trade data, we calculated daily Average Transaction Size (ATS) first hand.
On fig 8 one can see BTC ATS from the beginning of the summer till November 22nd and its notable rise which started on August 25th. The daily measure increased by 8.7 times from 0.21 BTC on average in the period preceding the pump to 1.83 BTC on average in the pump’s “first period”.

BTC ATS in the “second period” experienced further grow beginning from October 15th with a peak of 5.88 BTC on November 11th. Think it over: the average size of 14,823 transactions carried out in one day on Bithumb is 5.88 BTC (about $37,600). And that was happening not during Bitcoin dramatic fall in price, but a few days before that when it traded in a tighter range around $6,400. Eventually, ATS dropped to 0.17 from the beginning of the “third period”.

Let’s take a closer look into BTC trades on particular days. To begin with, this is how the trading looked like on August 8th, a day from the period preceding the pump (see fig 9).
The scatterplot of August 8th BTC transactions (fig 9) displays apparently natural trading activity with a small number of large trades performed throughout the day.
The histogram of August 8th BTC transaction size distribution (fig 10) shows that 10,557 trades (66.85% of the total) have the size less than 0.1 BTC, and 229 large trades over 2 BTC form only 0.015% of the day’s total number. The ATS of the day is 0.234 BTC.

Next, let’s look into BTC transactions on September 9th from the “first period” of the pump.
The scatterplot of September 9th, 2018, BTC transactions (fig 11) pictures an extraordinary pattern, a period of intensive trading activity with a great number of large trades and varying trade sizes. The intensive period lasted for only 5 minutes starting at 11 a.m. (see fig 12).
During that incredible outbreak, almost 7,500 trades (39% of the day’s total) were carried out with total volume of 34,600 BTC that accounted for over 94% (!!!) of the whole day’s volume. ATS of trades was 4.62 BTC, while ATS over the rest of the day – only 0.15 BTC.

From this, a very interesting questions arise: what caused this incredibly intensive trading activity which only lasted for 5 minutes of the day?
Besides, the histogram of BTC transactions distribution during the 5-minute “splash” (fig 13) shows only 910 transactions (12.1% of total) are smaller than 0.1 BTC but the number of large trades (4,544) greater than 2 BTC account for over 60.5% of the total number.

Let’s look at the trading activity in the “second period”.
The scatterplot of transactions made on November 11th BTC (fig 14) shows another interesting result. There are multiple series of equal or nearly equal trades (withing orange rectangles) and a greater overall number of large trades than in the previous period.
The distribution histogram shows that 4,686 trades (31.6% of the total) are less than 0.1 BTC and a number of large trades are greater than 2 BTC — 6,547 trades — that accounts for over 44% of the total transactions. This is a very high percentage, especially compared to 0.015% on August 8th, the day of the pre-pump period.

Let’s look at the trades of one particular hour from 3 a.m. to 4 a.m. local time, containing one of the series marked on fig 14.
On fig 16, one can see a large transaction series of around 10 BTC accounting for a vast part of total trades number. While looking deeper, we counted the number of transactions of the same size and their shares in the total.
The pie-chart on fig 17 displays the most frequent transaction sizes and their shares in the total trades number in the observed hour. As we see more than 75% of 1,330 transactions are made by only five closely commensurate and fairly large trade sizes.  

Altcoin markets review

After a more detailed analysis of BTC trading activity on Bithumb, we will review other coins, not in as much depth but to show the tendencies, which are pretty similar to Bitcoin’s.

Thus, in addition to BTC, there is an evident “comb-like” pattern on the “first period’s” hourly charts of LTC, ETC, XMR, ZEC (see fig 18) and some other coins including OMG and BTG.
The gradually rising daily volumes abruptly dropped to pre-pump levels on November 12th (except QTUM volume that dropped on November 10th) and can be distinguished on the daily charts of LTC, ETC, QTUM, WTC (see fig 19) and some other coins.
The ATS of the four coins experienced a significant increase during the “first” and “second period” of the pump as well (see fig 20). 
WTC stands out from all the coins we observed, as it was only listed on the exchange on the last day of August and had the shortest pump period which started on October 28th and lasted till November 11th. For or that reason its pump was one of the most intensive. The inflated daily volume of Waltonchain jumped by 350 times from 348k WTC (on average prior to the pump) to 122.5mln WTC (on average during the pump) only to then drop by by 1,450 times in one day from 206.7mln WTC to 141.8k WTC on November 12th.

Fig 21 featuring the WTC hourly chart shows 4 to 12 hours long intensive trading activity periods on each day with insignificant volume through the rest of time.

Moreover, the coins turnover during the pump on Bithumb exceeded its capitalization by several times. For instance, on November 8th it was 4 times larger (see fig 22).

Interestingly, in contrast to the other pumped coins, XMR and ZEC volume didn’t drop to their pre-pump levels on November 12th. While XMR volume maintained the average daily values of the “second period” after sliding down from the peak, ZEC volume slid from the peak as well but dropped to August levels on November 19th, spiking the next day to half of the average “second-period”value and dropped again on November 21st. ATS performance of both coins was in line with their trade volume trends (see fig 23).
And finally, the ongoing “third period” of the pump beginning on November 12th is characterized by an abrupt drop in the trade volume in previously pumped coins down to the pre-pump levels except aforementioned XMR and ZEC. Along with that, a new wave of pump started in DASH, OMG, and BTG (see figs 24-26).

Subscribe to our research

Enter your email address to subscribe to Hacken Reseach and receive notifications of new posts by email.

    hackenproof logo

    The world trusted Bug Bounty Platform. Run custom-tailored Bug Bounty Programs to secure your business and assets.