by Rohit Alluri
Reminiscent of the dizzy ICO boom of 2017, crypto is now witnessing a boom in derivatives exchanges that allow traders, arbitrageurs and risk managers to speculate on crypto prices by lending out capital to them. The inherent volatility of crypto prices gets magnified using leverage and often leads to unintended consequences, such as winners losing out on their gains and a cascade of liquidations. This is especially true for exchanges that do not have sound internal risk management practices. In an emerging, volatile area like crypto, derivatives risk management is also very much in its embryonic stage. Unlike a spot exchange, the technology underpinning the risk engine and the matching engine is also quite a non-trivial endeavour and even with established leaders, there are frequent instances of system overloading, outages and flash crashes leading to trader dissatisfaction. These are also the vectors that new entrants compete on — scalability, latency etc, assuming they manage to establish baseline liquidity and volume.
Insurance funds are mechanisms created by derivatives exchanges to manage risk effectively and to make sure that enough funds are available to pay out the gains to winners.
Let us illustrate the complexity involved in settling gains and losses for traders on derivatives exchanges using a simple example.
Consider two traders A (long) and B (short) who take opposite positions on BTC with 10x leverage. Assume that the price of bitcoin is $10000 and the initial margin deposited by both the users is 0.1 BTC. As the leverage is 10x, both the traders have an overall exposure of 1 BTC each. Trader A is long 1 BTC and Trader B is short 1 BTC.
Let’s say the price moves up 10% to $11000 per BTC. Trader A makes profit on his trade while Trader B loses money on his short trade.
Trader A made a profit of 0.1 BTC on his investment, while Trader B lost 0.1 BTC. As Trader B’s initial contribution of 0.1 BTC is wiped out, his order is now subject to liquidation.
The price at which a trade’s margin drops to zero is called the bankruptcy price. In this case, $11000 is the bankruptcy price for the Trader B.
Now the exchange takes control of Trader B’s position and to ensure that Trader A’s profits are whole, the exchange sells Trader B’s position at $11000 per BTC. Trader B’s $1000 margin will be moved Trader A’s account as profit.
But in real life, it is extremely difficult to make sure that the losing position gets liquidated at exactly $11000 because of extraneous factors.
Let’s take a scenario in which the losing trader’s position is liquidated at a price that is worse than the bankruptcy price, meaning above the bankruptcy price for short trades and below the bankruptcy price for long trades.
In the above example, let’s assume that the order gets liquidated at $11500, $500 above the bankruptcy price.
Trader A’s profit would be $1500 ($11500 — $10000) while Trader B’s theoretical loss would be -$1500 ($10000 — $11500). But in this case, Trader B deposited only $1000 as margin and his maximum liability is $1000, leaving with the winning trader short on his gains by $500 because the exchange was not able to liquidate the losing trader’s position at his bankruptcy price.
In order to prevent these kind of situations, exchanges prefer to liquidate the losing positions at a price different from the liquidation price to account for price slippage when selling large positions on the exchange.
Going back to the above example, if an exchange were to liquidate Trader B’s position at $10900, it gives them a buffer of $100 to make sure that the winner’s profits can be kept whole. If the losing trader’s position gets sold at say $10950, the winning trader’s accrued profit of $950 ($10950 — $10000) can be fully funded by the losing trader’s margin ($1000). However, once an order gets liquidated the losing trader loses his margin completely and anything that’s leftover after funding the winner’s gains is moved into a separate vehicle called the “Insurance Fund” to protect traders against calamitous events in the future. As long as the exchange can liquidate an order at a price better than the bankruptcy price, there will be a net positive inflow into the “Insurance Fund.”
Below graphic compares entry price, liquidation price and bankruptcy price for both long and short positions.
For small orders, the final execution price is very close to the liquidation price as they can be executed with relatively negligible price slippage. In the case of large orders, if the market is moving unfavourably, exchanges might not be able to liquidate losing positions at a price better than the bankruptcy rice all the time. This is when exchanges tap into their Insurance funds to make the winners’ ‘whole’.
Click here to read the postmortem of a $416 million liquidation on OkEx that triggered a cascade of liquidations, resulting in the exchange unable to offload losing traders’ positions at a price better than the liquidation price.
As stated in the above example, when the losing trader’s position is sold at $11500, $500 above the bankruptcy price, the winning trader is shortchanged by $500 and the exchange takes out $500 from the insurance fund to make his profits whole again.
The concept of Insurance Fund is not unique to crypto derivatives exchanges. Traditional exchanges such as CME and CBOE also have safeguards to handle situations where one or more clearinghouses default on their obligations. However, with traditional exchanges, it is the clearinghouses that are responsible for managing the trade positions of their clients and the significant reputational damage that can stem from a default shifts the burden of risk management onto the shoulders of the clearinghouses. In the event a clearinghouse defaults, an exchange can use its financial resources to ensure that any outstanding financial obligations are fulfilled.
As a majority of crypto derivatives exchanges deal with anonymous retail investors, the onus is completely on the exchange to manage risk effectively.
The figure below shows the funds available at CME’s disposal in the event of a clearing house default.
BitMEX Insurance Fund:
BitMEX is the leading crypto derivatives exchange in terms of trading volume, and also boasts the largest insurance fund among its peers. At last count, BitMEX’s insurance fund hoards about 31.3k of BTC, a figure that accounts for 0.2% of total BTC available in supply. Over the past one year, BitMEX’s insurance fund has grown in size by over 200% from 12k BTC to 31K BTC now.
Figure 2: BTC in BitMEX Insurance Fund
Deribit, the second largest crypto derivatives exchange, has only around 280 BTC in its insurance fund.
Fig 3: BTC in Deribit Insurance Fund
The below figure compares the size of insurance fund per dollar daily traded on an exchange. There is no ideal range for this metric yet, but a high multiple indicates that an exchange is too punitive on traders while liquidating their positions, while a low multiple implies that the exchange has little buffer to protect winners against catastrophic price shocks. Especially in the quasi-regulated/loosely regulated, opaque regulatory and corporate structures that many of these entities operate in, an insurance fund that is excess of 30,000 BTC ( as in the case of Bitmex) is bound to attract scrutiny.
What Caused BitMEX’s Insurance Fund To Appreciate So Fast?
In May 2018, BitMEX’s insurance fund had roughly 6000 BTC. Now that figure has skyrocketed by over 5x to 31300 BTC in fewer than 18 months. One of the reasons this might have happened could be due to the prevalence of extremely high leverage on trades on BitMEX. Earlier this year, BitMEX released some information on how levered the orders on BitMEX are.
While a majority of contracts are levered up by <10x, a significant number of contracts are deployed at 90–100x leverage, resulting in a bi-modal distribution of contracts outstanding.
Fig 5: Total Number of Contracts at each Effective Leverage
As we discussed earlier, exchanges are incentivized to liquidate orders at a price better than the liquidation price to protect themselves against price slippage. However, the price difference between liquidation price and the actual bankruptcy price varies with leverage, This means that for orders with low leverage, the liquidation price is much closer to the actual bankruptcy price and the spread is wider for orders with high leverage.
The below table illustrates the price spread between the actual bankruptcy price and bitmex’s liquidation price as a function of the initial price. As the leverage increases, the price spread between the actual bankruptcy price and the liquidation price widens. This results in traders having lesser buffer at higher leverage, leading to orders getting liquidated sooner than expected if the market moves against the position.
For example, at 100x leverage with $100 in initial margin, a price drop of 0.5% is enough to trigger liquidation as opposed to a theoretical 1% price drop that is required to wipe out the margin. At 10x leverage, the liquidation is triggered after an 8.7% price drop, which is much closer to the theoretical 9.1% required to wipe out the initial margin.
Assuming an initial margin of $100, the below figure plots the initial margin in $ terms that will be taken over by BitMEX at the time of liquidation. The maximum amount of capital that can be added to the insurance fund is highest at 100x leverage.
Fig 6: Maintenance Margin Requirement (Y-axis) Vs Leverage (X-axis)
From a trader’s perspective, then the amount of risk assumed at 10x leverage is exponentially higher than the amount of risk assumed at say 3x leverage. Paradoxically, the extra risk assumed per leverage starts going down once leverage crosses around 50x. In other words, once you are going high with leverage, you might as well go all in. This could explain the curious behavioral pattern exhibited in Fig 5, in the trade leverage information revealed.