What Caused this Drop (and What Didn’t)


by Marat Garafutdinov, an Analyst with HASH CIB
m.garafutdinov@hashcib.com

We have recently published a report on the past two weeks+ and upcoming few months in the cryptoasset markets. Unfortunately, it is unavailable to the general public. If you are a financial/digital asset markets professional, you can fill out the form on our website with actual (and verifiable) information to request access. It goes against the ethos of the public blockchain space, but we want to stay universally compliant and set a certain bar for ourselves and, hopefully, others. Here we share some of our high-level market data analysis. Any substantiated critique or comment is very welcomed. Thank you.


14–25 November 2018 saw a massive sell-off in cryptoasset markets, drawing fears of significant prolongation of the bear season in digital assets with last of the recent market entrants possibly capitulating. The sharp move was somewhat expected, as the market volatility had been trailing at historical lows for almost all of the most liquid cryptoassets. The usually volatile market’s kinetic energy had been suppressed for two months at this point — it was destined to shoot off. The direction of the outburst was unclear.

After the sudden price fall on the 14th of November the hindsight consensus on what caused it in the following days revolved around multiple themes, as the market was looking for a narrative. From what we’ve learned from analysing the crypto markets is that there is always some under-the-rug activity that goes hand in hand or even precedes the price action. In the realm of emotions, which still vastly applies to digital asset markets, trends are regularly set up, rather than ridden. While the fundamentals are often times detached from sentiment. Even top cryptoassets aren’t immune to this. Let us explain it on this recent case.

The initial markets sell-off was presumably caused by two large ETH short positions on BitMEX crypto derivative exchange early on Wednesday, 14 November, 2018. At 9:48 (UTC) we can see a significant price decline in ethereum, probably started on BitMEX[1]. The ETH/USD pairs on Bitfinex and Binance also started moving the same exact instance. It is hard to pinpoint the very exchange and the very millisecond the sell occurred as the most liquid crypto exchanges are well inhabited by algorithmic traders. However, judging by the market data we collected, the largest downward swing seems to have started on BitMEX — which could also be attributed to largely leveraged positions allowed on this exchange. While Bitfinex and Binance spot exchanges could also have caused the first dump in ETH/USD markets, it is the BitMEX price action that took it further, so we stick to our assumption.

BitMEX market data, view on Plot.ly

The prices of other market leaders following the sell dropped as well, but the amplitude of the movement in ETH was higher. The tendency of the previous several months to oversell this currency remained, without markets taking significant notice. One hour before the real November 14 crash (which started at 16:15 UTC) at around 15:14 UTC the last red flag could be noticed (see below) - once again ETH prices were the first to go down. BTC was also affected, but 10 minutes later it recovered the losses, while ethereum remained under pressure. We would like to note that the market data we collect from various exchanges shows bettering efficiency of crypto markets, with Bitfinex and Binance (the other two ETH/USD markets we looked into closely here - for a number of reasons beyond the scope of this article), trading in a very close tandem in the most liquid pairs.

BitMEX market data, view on Plot.ly

Taking into account the fact that cryptoassets are highly correlated, such divergence couldn’t be avoided by algorithmic traders. We are rather safe to assume that the spread deviation between ETH and other top traded assets was a signal for some of them to enter the buy ETH/sell BTC position. As a result, one hour later (16:16 UTC) we witnessed the beginning of a new global downtrend.

BitMEX market data, view on Plot.ly

In order to quantify the move and make sure it was an intentional play we calculated the volume of BitMEX’s ETH/USD market during the alleged short position takings and compared them to random 5-minute intervals within an hour range (+1h/-1h) of the same trading hours timeframe — sampled through days preceding and following the dump. The table below shows a significant uptick in volume during the short sells we identified. Even compared to the bigger of the trading days that followed. Again, this could very well be attributed to initial volatility wave being caught by BitMEX traders, but the November 14 uptick still stands out, in our view.

BitMEX market data

As we haven’t seen indicators of massive spot market dumps preceding the move (usually following large OTC deals), we assume a deliberate attempt to front-run a sell-off in ETH on BitMEX. Looking closer into one of the possible precursors — the ICO-capitulation story, we’ve checked the data we’ve been collecting on ICO treasury accounts since July 2018. Generally speaking, ETH moved from such accounts up to November 14, 2018, amounts to just 0,56% of current total ETH supply, while the price has fallen 60% during the same period. There was no uptick in ETH moving from any of such accounts preceding the sell-off. The outflow of ICO funds we tracked during November 14–30, 2018, was 286 036 ETH (with 82k ETH moved just on the 28th of November). In $ terms this amounts to $36m.[2]. Given how liquid the ETH/USD markets were during these days with volumes of $2b, $1.8b, $1.8b, $2.7b and $2.3b on November 16- 19 and 27[3], respectively, we believe that it’s a relatively moderate factor to affect ETH price at this point.

Starting from the 14th we saw some activity, but only after the sell-off had begun. 22% of the moved ETH (up to 30 November, 2018, 12:00 UTC) ended up on the exchange accounts, 23% was sent to the Wrapped Ether smart contract, and 55% ended up in the unknown address. While we cannot be sure that most of the unknown destination transactions are just on-going costs (presumably, paid in ETH and never directly hitting the market in full), we are confident that 23% or $9m worth of ETH[4] sent to the Wrapped Ether smart contract never hit the markets.

Wrapped Ether (wETH) is widely used to then post collaterals for the Dai stable coin system on Ethereum. Dai is a dollar-pegged stable coin issued by a trustless distributed system (dubbed Maker) that allows ETH holders to borrow in $ terms without ever touching the centralized crypto-to-fiat exchange or other, also centrally issued, dollar-pegged stablecoins. Due to ETH being volatile, Maker issues over-collateralized dollar-pegged Dai tokens that can be traded for other cryptoassets. In both instances ETH was moved to top up the Dai collateral smart contract as the ethereum price started approaching the margin call price level, we assume. Also, the Wrapped Ether (wETH) tokens can be used on Radar Relay decentralized exchange – where a predominant, but yet ignorable volume is attributed to the wETH/Dai pair.

So, in reality just $8m[4] worth of ICO treasuries’ ETH directly hit exchanges during November 14–30, 2018. Given the data we have, we are confident to say that ICO projects reacted to the market conditions, rather than dictated them.The treasuries currently still hold a minimum of 3.4% of the total ethereum supply. Our findings have been confirmed by separate researchers. And contradicted by some — most probably, due to reporters equalling all of the ETH moved to ETH directly sold.

There were also two Ethereum blockchain-related news around the dates of the initial market sell-off. One of them was Nvidia admitting on Thu, 15 November 2018 a decline in their Q3 earnings due to reduced GPU-heavy crypto (i.e. mostly ETH) mining. Nvidia stock (NVDA) plummeted as the news was broken. So it was the extended ETH price drop causing NVDA to finally follow rather than the other way around. The other news, though, was more important. As the US Securities and Exchange Commission announced having settled enforcement orders with two 2017 Ethereum-based ICO issuers and followed it in a few hours with a general statement on the token issuance and trading on Fri, 16 November 2018, ethereum price came under additional downward pressure compared to BTC.

BitMEX Market data

Speculating on whether someone really front-ran the SEC announcements is beyond our reach. The ETH short seller could have made their profit on sheer expectations of market breakdown given the tight range it kept trading prior to the fall. Yet, our findings bring us to another conclusion: it’s the price action in ethereum crypto and narrative around Ethereum blockchain that really push the markets (even if the actual data doesn’t match the market sentiment) — with the momentum only then carried by the BTC price moves. Similar to last year, when Enterprise Ethereum Alliance announcement sparked the whole bull run of 2017 with ICOs soon taking it further up and only then BTC driving it home.

source: Coin Metrics

As soon as the market sell-off began, BTC as expected, became the main focus of the analysts and commentators. Despite much of the attention around Bitcoin being sucked into the turmoil surrounding the smaller Bitcoin offshoot — Bitcoin Cash (BCH), which split even further into two more chains — the actual underpinnings of the drama surrounding that split were minor and negligible.

To quantify the extent of possible BTC sell-off by the BCH miners competing for which fork has the most hashrate (mining power) support we calculated the delta, resulting from the increase of combined hashrate on both of BCH forks. We assumed all of it to be financed solely by selling BTC rather than switching over from mining BTC (which is more likely in majority of cases). We have calculated the cost of: a) buying all of the equipment necessary for the bump up; b) rent-only cost and c) a 50/50 combination of both. Thus, from the figures obtained (see below) we concluded that BCH mining war has little to do with the actual BTC price, even if we assume that all of it was financed via BTC selling. Given the recent multibillion-dollar BTC daily volumes, we see it as hardly possible. Also, based on our calculations, aggregate BCH hashrate increase could only amount to 7% of total BTC hashrate reduction over the week of 14–21 November, 2018.

Outside of the Bitcoin Cash-related fears of BTC losing both in miner-backing and price, multiple commentators have found that Bitcoin hashrate has nevertheless significantly reduced, possibly even preceding the price fall. Below we have calculated current bitcoin mining profitability in China (as the sort-of-benchmark region for BTC mining) with one major caveat: our calculations were based solely on Bitmain’s model S9 hardware specs — as the most commonly employed equipment in China. Given how competitive the mining market is, we assume roughly similar and characteristics (possibly, customized) mining hardware currently used by miners in terms of its hashrate and electricity consumption characteristics.

source: HASH CIB, Statista

The two charts here show major discrepancy in Bitcoin mining profitability for an average Chinese miner (above) and the bigger players (below). What it tells us is that some decent percentage (bigger players, Bitmain’s publicly available data being the reference), if not the majority, of Chinese hashing power could possibly have just now come close to mining at a loss. Also, if our calculations are correct, it means that the smaller players most probably haven’t been able to accumulate much of BTC on top of what’s been sold to cover their costs — probably having little reserves left to dump in covering losses.

source: HASH CIB, Quartz

We believe that most of the miners have already been drained by the continuous price decline of 2018, so possibly not having that much BTC to sell in the first place. Given the recent difficulty adjustment on November 16–17 and the one coming in a couple of days (presumably, less significant)— with massive hashrate outflow due to the price decline, we expect the mining profitability to improve for the remaining miners. On top of that you’ve got a long awaited newer version of Bitmain’s miner hardware coming out in a month with shipments currently estimated for 20–31 December 2018. So again, just as with ICO treasuries in ETH, miner industry in BTC follows price. We don’t see it’s shrinking to be a major factor in past two week’s BTC market performance.

If you would like to know the assumptions we used in our calculations please contact our analyst directly via email (presented above). Inquiries by unidentified readers will be ignored.

[1] BitMEX offers synthetic asset trading with USD-denominated pairs, all collateralized with actual BTC. Its ETH/USD perpetual swap contracts (launched in August 2018) is its second most liquid market after the BTC/USD perpetual swap, we assume. However, BitMEX never officially reveals its historical market data.

[2] Aggregate of all transactions, each time priced at average ETH price for the day funds were moved, per CoinMarketCap. We have neglected transfers of less than 1200 ETH (amounting to 24 435 ETH or $3m through 14–30 November).

[3] Priced at average ETH price for the day funds were moved, per CoinMarketCap.

[4] We measured the markets according to an average 2-day lag (with November 30, 2018, missing).