Systematic BTC Options Trading Strategies: Yield Generation

• Samneet Chepal

Note: This piece is only for educational purposes and does not constitute investment advice.

Given the ultra-low interest rates across the world, investors have been forced to search for new opportunities to earn a yield on their capital. In traditional markets, there are many strategies that involve the use of systematically selling options that provide relatively consistent returns if the risk is managed carefully. As such, it would be interesting to see if we can create a similar yield strategy for the budding crypto options market.

Below are a few guiding points to keep in mind when building this strategy:

• We’ll focus on BTC options given they have the most liquidity.

• To avoid the massive gamma risk of selling short-term options, we’ll focus on mid-dated ~25 delta options with an average maturity of 25 days.

• The ROI is in terms of BTC and all returns in this analysis are compounded.

• The options are sold at the mid-price (average of bid and ask).

• We collateralize each short option with 1 BTC to reduce the risk of liquidation.

An initial approach could be to systematically sell 25 delta options and roll them continuously as the option expires. If we believe BTC will rally upwards past our chosen strike or stay stagnant at current prices then we’d sell a put. Conversely, if we believe BTC will fall or stay near the same price we’d sell a call. Although one could argue that BTC’s asset returns have a positive drift, selling put options haphazardly exposes us to massive tail risk such as the price action in March 2020. We can do better by adding in a few constraints listed below:

• Include a 14/30 day EMA trend filter to determine the market regime. In bull markets (14 day EMA > 30 day EMA) we’d sell puts whereas in bear markets (14 day EMA < 30 day EMA) we’d sell calls. This trend overlay can serve as a rough quantitative metric to assess which side to be on.

• Exit the position if the option’s maturity is less than 2 days. This is done to avoid the short gamma risk near expiry.

• Exit the position if the option’s price is less than 0.0015 BTC. The reasoning here is that if we’ve been able to collect most of the option premium it’s probably not worth holding onto this low risk-reward position.

• If the option ends up deep in the money (> 80 delta), we’ll cut our losses and free up collateral for the next trade.

If any of these criteria are met, we will close the current position and move onto the next trade. With these constraints, we can see the initial backtested returns of this strategy are encouraging. Overall, the total return is around 105% with a Sharpe and Sortino of 2.46 and 3.24 respectively.

While this is a good starting point it’s important to think about underlying risks with this strategy. Sharpe and Sortino ratios can help us assess risk-adjusted returns, however, these metrics don’t fully account for the inherent tail risk associated with selling naked options. Even though we didn’t encounter a severe drawdown in the backtest, it doesn’t mean this strategy is immune to future black-swan events. One way to improve the risk-management of this strategy could be to refine the EMA timing signal so it can better assess changes in market regimes. Furthermore, we could add in additional criteria such as liquidating a position once stop-loss limits are breached.

Another approach we looked at was to explore selling spreads rather than individual options. The logic here is similar to the strategy above but in this case, we’re buying a low delta option to hedge the naked exposure. Therefore under this strategy, our individual trades will have bounded losses, however, on both an absolute and risk-adjusted basis this strategy underperformed the original approach. The total return for the spread strategy is around 63% with a Sharpe and Sortino of 1.89 and 2.43 respectively.

Overall, running this strategy will largely depend on an investor’s risk tolerance and return objectives. An individual with a lower risk preference may wish to utilize the spread strategy knowing they are protected from tail risk. Conversely, market participants with larger balance sheets and tolerance for risk may be comfortable with the baseline approach in an attempt to earn higher returns.

Aside: Call and Put Selling Strategy EMA Window Parameterization

A useful suggestion I received was to assess the stability of the EMA windows and see whether the final cumulative profit was sensitive to small changes in the EMA parameters. Currently, the fast and slow EMA windows are 14 and 30 days respectively. In order to assess whether our current results for the call and put selling strategy are reliable, we can re-run the model across a range of different window periods (10–20 days for the slow window and 20–40 days for the fast window). Below we can see some heat maps outlining the results for the Sharpe, Sortino, and overall cumulative returns.

It’s encouraging to see the Sharpe, Sortino, and cumulative returns are relatively stable across different window periods near the 14/30 day EMA region. To further improve this analysis, we could run this optimization over a larger range for both the slow and fast windows (although this would be very computationally expensive).