In traditional financial products, ETFs and Indices solve the problem of generalized investing rather than very specific investing. They track wide swathes of sectors or assets in order to give a utilitarian choice to an investor for a generalized investment vehicle. As such, this is a giant business due to this use case. Indices generally use an arbitrarily designed weighting to come up with a price level that serves as a reference for the cohort of individual investments within that Index. This is not something that Bitcoin needs. Initially, there was a need for a generalized price of a single asset such as BTC due to the wide variation in this asset’s pricing across multiple venues. OTC pricing, opaque fees and misreporting contributed to all this.This has ceased to be the case with most money-transmitter or service businesses like venues reporting accurate pricing along with size/depth.
Clearly using reference pricing as a function of some exchanges, such as equal weighting after removing n-outliers exposes the reference price to one or more exchanges suffering issues. Those could be purely operational or accidental and could also be nefarious or genuine attacks. This has been seen more recently with some high-profile exchanges that caused a domino-like effect on positions on certain large derivative exchanges and auto- liquidated users en-masse. By using a CBBO filtered for actual trading sizes, any exchanges that are affected by the above problems do not even show up in the consolidated quote. This increases reliability for the users of the exchange as they are protected from importation of reference pricing threat vectors.
On July 14, 2019, a sharp decline in Bitstamp’s ETH-USD was observed that resulted in a $60 spread in dollar terms (equivalent to a 25 percent spread) between Bitstamp and other major exchanges. This event has many similarities to a similar market movement on Bitstamp’s BTC-USD market on May 17, 2019. Although one possible explanation is trader error, a less charitable interpretation of this event is that this is another example of deliberate selling designed to trigger forced liquidations on long futures positions, margin calls on long margin positions, and stop losses on spot markets.
Bitstamp is one of three constituents in BitMEX’s Ethereum price index which is the price index that BitMEX’s Ethereum perpetual futures contract is priced on. Due to the presence of up to 100x leverage on BitMEX (and other exchanges that offer cryptocurrency futures products) and the outsized impact that futures products have on spot markets, the incentive to engineer price movements is clear. During a three minute window between 09:51 and 09:54, a total of 10,186 Ethereum was market sold, clearing the order book, and causing a large spread between Bitstamp’s market and other major exchanges. This quantity of Ethereum only amounts to approximately $2.7 million in U.S. dollar terms using prices immediately prior to the event.
Since Bitstamp’s ETH-USD is one of three constituents in BitMEX’s price index, the sudden drop in price caused roughly $17 million in notional value to be forcibly liquidated on BitMEX’s ETH-USD perpetual futures contract. More troubling is that this engineered price movement seemed to be a catalyst of a broader market sell-off which affected other assets including Bitcoin. Major BTC-USD markets simultaneously sold off during this time including forced liquidations of roughly $60 million in notional value in BitMEX’s BTC-USD perpetual futures contract. The overall market would continue to sell-off through the remainder of the day with Bitcoin momentarily reaching below $10,000.
To further understand the malicious effects of an ill-designed reference price index or mark price index on traders, we performed a theoretical comparative analysis for liquidations on Qume and BitMEX on Sep 24th when the price of BTC fell by >$1000 in a matter of minutes.
Firstly, we plotted the reference prices for Qume and BitMEX for a period of 80 minutes during with the price of BTC experienced a meteoric crash. As seen from the chart below, Qume’s reference price is smoother and is less subject to price fluctuations as it is outlier agnostic but takes the liquidity weighted price meaning where you can buy and sell at least 5 BTC at any instance, whereas BitMEX’s reference price was prone to price fluctuations elsewhere as it is calculated based on the price feed from just 2 exchanges (Bitstamp and CoinbasePro).
The price spread between Qume’s reference price and BitMEX’s mark price showed meaningful divergence in a 10-minute window during which the spread widened to over $500 because the price of BTC had dropped to $7900 momentarily on Bitstamp and not in line with prices on other exchanges, causing the BitMEX reference price to drop to as low as $8000 in that period. This anomalous behaviour of steep rise/decline in price could work against traders by liquidating their positions when they should not have been.
The below figure shows the price divergence on Bitstamp during the flash crash event.
We created theoretical trade positions at different entry prices and leverages to measure the number of liquidations that would be triggered on BitMEX and Qume after the flash crash on Sep 24th. We varied the entry price from 9200 to 10000 in increments of $50 and the leverage from 6.0x to 9.0x in increments of 0.5x. We calculated the number of liquidations at each leverage on both exchanges to highlight the importance of having a robust reference price that is not skewed by the price feed from a single exchange.
At each point of leverage, the number of liquidations on BitMEX were higher than on Qume, highlighting the robustness of Qume’s reference pricing and therefore better protection for traders in extremely volatile price scenarios.
Liquidations On BitMEX (Leverage vs Entry Price)
Liquidations on Qume (Leverage vs Entry Price):