Bitcoin Futures: Promises and Pitfalls
Two popular uses of futures contracts are as a hedging instrument or as proxy for a long exposure. In this memo we examine how well bitcoin futures have served in these two capacities since their launch in December, 2017.
The Cash and Carry Model
Futures prices are set equal to the price of the underlying asset on the futures expiration date, therefore, the futures contract must track the underlying asset’s price closely during its lifetime. This relationship is described by the Cash and Carry model. A simple version of the model looks like:
F = Spot price + cost of carry = S (1 + CT)
F = Futures price
S = Price of the underlying asset, also called the “Spot Price”
C = The cost of holding the underlying asset in % of S, also called the “Implied Finance Rate”
T = Time from trade date to the futures expiration date
Essentially, this model says that buying a futures contract has the same economic consequences as buying the underlying asset and holding it until the futures expiration date. This is because on expiration date, the price of the futures is contractually required to be the same as the underlying asset.
Note that this relationship can be violated due to forces such as supply and demand. However, because there is a predictable future outcome (on expiration date), it is then possible to use arbitrage techniques to enforce the relationship and push futures and spot prices back into alignment. That is, if the market mechanisms are in place to allow efficient construction of arbitrage trades.
The market must allow the arbitrage in both directions. In the case of bitcoin, the two arbitrage trades are:
- Long bitcoin and short bitcoin futures — This arbitrage trade can be constructed fairly easily at this stage in the bitcoin market’s development. Buying bitcoins can be done easily, storing the bitcoins may be expensive but custodial services are readily available, and, borrowing fiat money using bitcoins as collateral is also possible (but expensive). These costs are captured by the variable C (Carry) in the model. Lastly, shorting futures is easy as long as there is sufficient liquidity.
- Long bitcoin futures and short bitcoin — In this direction, the arbitrage trade runs into trouble. Going long futures is no problem given adequate liquidity on the exchanges. However, at this point in time, shorting bitcoins cannot be easily accomplished. There are only a few trading venues where shorting is possible, but it’s prohibitively expensive. Furthermore, simultaneous execution of a short sale with the purchase of futures (in order to lock in the arbitrage profit) is also very difficult.
With the above discussion, we should not expect the relationship between futures and bitcoin to be well behaved. Indeed, the delta line in the graph below (February 2018 CBOE bitcoin futures contract (XBT Feb18) vs. Gemini 4pm bitcoin auction prices (used by CBOE to settle the futures on expiration date)) shows significant discrepancies from the Cash and Carry relationship.
The Cash and Carry model dictates that bitcoin futures prices should be above bitcoin prices until expiration date, and, the futures premium should gradually diminish over time in an orderly manner. The graph shows an erratic behavior (the Delta line) with numerous days where futures fell below bitcoin spot.
Bitcoin Futures as a Hedge Instrument
The goal of hedging is to remove the fluctuations (a.k.a. volatility) in the Profit and Loss (P&L) statement. A good hedging program will “lock in” the P&L at a constant level and reduce volatility to near zero. This is a similar but different goal from the concept of diversification. Diversification reduces P&L fluctuations without sacrificing long term expected return (i.e., diversification does not lock down the P&L).
A typical scenario for utilizing futures to hedge is when there is a long-term position in an asset (e.g., bitcoin), and the futures is used to neutralize the P&L swings temporarily. For example, if someone purchased bitcoins as a long-term investment and wanted to avoid the large paper losses due to recurring selloffs, then shorting futures would be a way to accomplish that.
Note that the hedge trade is very similar to the arbitrage trade. There is a long side and a short side in both trades. The difference lies in the fact that when unwinding a hedge, the underlying long-term position remains in place. For an arbitrage, both sides of the trade are entered into and existed from simultaneously.
To be an effective hedge instrument, the futures needs to move in lock step with the underlying asset. In the previous section, we discussed the lack of arbitrage enforcement to align bitcoin futures to bitcoin spot prices. Therefore, it is expected that hedge effectiveness will be compromised.
Two statistical metrics are used to assess hedge effectiveness: R-Squared and regression coefficients (i.e., regression slope). These two metrics are computed for:
- Daily absolute changes — This approach assumes the hedge is constructed with equal quantity on both sides. For example, to hedge 10 bitcoins, go short 10 XBT futures (each XBT contract is scaled to 1 bitcoin).
- Daily % changes — This approach assumes there is an equal dollar amount on both sides. For example, to hedge $100k of bitcoins, sell short $100k worth of futures (this is difficult to do since futures do not trade in fractional contracts).
Daily changes are used to perform the study since in most cases P&Ls are computed on a daily basis.
Using the Gemini 4pm bitcoin prices and the XBT Feb18 futures closing prices, the following R-squared values and regression coefficients were computed:
- Daily absolute changes — 97% R-squared, 89% coefficient.
- Daily % changes — 98% correlation, 98% coefficient.
In both cases, the high R-squared values indicate that the regression coefficients are reliable (i.e., high levels of confidence) descriptions of the linear relationship between bitcoin and bitcoin futures daily price changes. However, the 89% coefficient for absolute changes indicate a suboptimal hedge effectiveness since the magnitude of change is not 1 for 1. The coefficient for % changes promises a better hedge result at 98%. Therefore, as a hedge instrument, bitcoin futures work best in an equal dollar hedging strategy.
The high correlation values are a pleasant surprise. It indicates that while futures price levels do not follow the Cash and Carry model closely, the daily changes are largely in sync with bitcoin spot price swings.
Bitcoin Futures as a Long Proxy
Professional investment managers often use futures to quickly gain exposure to an investment such as the SP500. SP500 futures are liquid and much easier to trade as compared to 500 individual stocks. And, futures do not incur a management fee like ETFs and mutual funds. The idea is to use futures to produce returns that are very similar to the actual investment. Note that under this scenario, the investment manager is not utilizing the leverage offered by futures. In other words, if the manager has a $1mm to invest, she will purchase only $1mm worth of futures and invest most (some cash will be used as futures margin) of the $1mm cash in low risk interest bearing securities like T-bills.
While futures serve as a cost-efficient way to replicate the returns of the underlying asset, it is not cost free. Recall the Cash and Carry formula:
F = Spot price + cost of carry = S (1 + CT)
The cost of carry (the carry) component can be thought of as the cost of using futures as a proxy to the underlying. Over time, the carry will diminish (as T is reduced), which acts to pull down futures prices. For example, if the spot price (S) stays constant, then the futures price will gradually fall towards spot until the two prices are the same at expiration date.
We can compute the value of C given each day’s closing futures price and Gemini’s 4pm spot price. From the futures launch date of 12/11/2017 to 1/31/2018, here are the statistics using the XBT Feb18 contract: (annualized carry)
The above statistics indicate that futures prices were very high relative to spot during the first days after launch. The 12% average carry during December was likely arbitraged and forced to come down. When futures are over-priced, the arbitrage is to sell futures and buy bitcoin, this trade can be executed efficiently. When futures are too cheap (e.g., when carry is negative), then the arbitrage is to buy futures and short bitcoin. As previously mentioned, this trade is not easily executed, so there were numerous days in January where carry was negative, see graph below.
The limited history of bitcoin futures does not allow firm conclusions to be drawn from the above analyses. The 5% average carry cost is simply a number to keep in mind and monitor going forward.
Conclusion
Bitcoin futures has been a welcomed addition to the digital asset market infrastructure. Two of the many uses of futures were examined in this memo: Hedging and long proxy. In the hedging use case, there is evidence that bitcoin futures work well in an equal dollar hedging strategy. For the long proxy use case, there is preliminary data pointing to a high cost of carry at 5% annualized. Additionally, if the proxy is maintained for a long time, then the cost of rolling futures contracts would be an added concern. These futures related costs should be compared with the costs of financing and maintaining actual bitcoin holdings, or, with the cost of investing in a fund that provides bitcoin exposure.
Michael Yeh is a co-founder of IronChain Capital, a digital asset funds management company, with responsibility for day to day management of fund operations. Michael has over 20 years of broad based experiences in financial services with companies such as Goldman Sachs, Barclays Global Investors, and Blackrock.
