How to exploit Momentum Factors in the Bitcoin Futures Market

18 May 2021
By Navdeep Sahote, Product Manager

The surging demand for Bitcoin has fuelled price increases for months, hitting a record high of $63,000 in April, only to fall off a cliff a few weeks later. We have analysed how a Bitcoin Futures momentum factor strategy performs compared to buy and hold.

Momentum is a core element of factor-based investment strategies and relies on the continuation of past patterns of return. This pattern is demonstrated in the digital asset space more than anywhere else and investor herding is one of the cornerstones of momentum trading.

Investors like winners and Bitcoin has entered parabolic phases resulting in high returns with greater risks. Trading this momentum is therefore a valuable but challenging proposition. To overcome some of these challenges, we used SIGTech’s quant trading platform to help us identify an optimal strategy. Although the last few months have seen strong growth in alternative digital assets, such as Ethereum, picking the right asset for a momentum factor strategy is crucial. Bitcoin Futures are an obvious choice as they are highly liquid, volatile and show the strongest price-swing momentum in the digital asset industry.

1. The Bitcoin Futures universe

We defined buy and hold as the baseline strategy, then analysed how momentum factors performed against it. We chose a rolling futures strategy where exposure to the front-month contract is constantly maintained as the baseline. It had an annualised return of 40.1% and a Sharpe ratio of 0.81. The return series, given a starting portfolio value of $1000, is shown below.

Rolling BTC Front Contract Strategy

After defining the baseline, we analysed multiple momentum factor strategies. For this task, we used the simplest measure of momentum, which is total return over a fixed lookback period. This period is denoted in days. A buy or sell order would then be placed given the sign of the returns over this period. Since lookback periods are completely arbitrary, we backtested a range of time periods to find an average annualised return and sharpe.

2. Single lookback period

Sharpe versus Momentum Window Length

Backtesting a range of lookback periods, ranging from 1 - 200 days, reveals an average annualised return of 22.5% with a Sharpe ratio of 0.62. On average, a random lookback period does not result in better performance than our baseline. However, taking a 3 month lookback period achieves Sharpe ratios around 1.25, with correspondingly higher annualised returns

It was clear that short-term lookback periods performed the best. Based on this information we decided to add a layer of complexity to our backtests with a focus on shorter periods.


Receive our latest blogs, trading strategy ideas and insights

3. Dual lookback period

Sharpe versus Momentum

We also simulated strategies that used two different lookback periods. The signal depended on whether the short span was above or below the long span. As shown above, a simulation of different combinations of long and short spans took place. These simulations are computationally-intensive, but with the help of SIGTech’s parallel compute function the process was approximately eight times faster than our standard compute instance. The standard instance took over 30 minutes to complete all simulations, whereas parallel compute only needed four minutes.

Here, the average annualised return was 49.5% with a Sharpe ratio of 0.91, which now outperforms the baseline rolling futures strategy. From this, it was clear that the lighter area indicating a higher sharpe was our area of interest. A strategy that ran with a short span of 7 days and a long span of 40 days returned a Sharpe ratio of 1.30 and annualised return of 95.1%. However, it’s maximum drawdown was 53.1%. We attempted to decrease this factor by testing a long-only momentum strategy with the same parameters.

4. Long-only strategy and comparisons

Comparison of logarithmic returns

A long-only strategy, which used a negative signal as a risk-measure rather than a short entry, returned an annualised return of 114%, Sharpe of 1.65 and max drawdown of 33.7%. Interestingly, as of 1st May 2021, our long-only strategy closed its long position (opened around mid-October 2020), while the long-short strategy had flipped short. By 18th May the short strategy had returned 19.54% from its short position around $56,500.

Momentum factor strategies can be incredibly effective in the Bitcoin Futures market. However, as shown above, random measures of momentum are unlikely to result in performance better than the returns from simply holding Bitcoin. Knowing which measures of momentum to use, as well as their corresponding parameters is imperative. There are many opportunities in the ever-growing Bitcoin Futures space for sophisticated investors. Having the right technology to identify patterns fast and accurately will separate the good from the bad.

CME Bitcoin Futures Data is available to all SIGTech users. Get in touch to find out how SIGTech could help identify signals for your next strategy.