How AQRU Trend works
AQRU Trend is an algorithmically-managed index, designed for people who haven’t got the time to watch the market all day, and are looking for a simple solution aiming to outperform Bitcoin’s returns but with lower volatility. The AQRU Trend algorithm waits for markets to establish a clear multi-month uptrend, and then invests in any of ten cryptocurrencies. It automatically rebalances into USD-pegged crypto assets when uptrends break down, and takes care of all allocations when invested.
Unlike index trackers that are always fully invested, AQRU Trend is a trend-following system that dynamically manages your exposure to the crypto market, depending on medium-term price movements, making it ideal for those who are looking for a hands-off solution.
But how does it actually work?
Scroll down to read our Chief Investment Officer Philipp Kallerhoff’s technical paper on the methodology and execution of AQRU Trend.
And here’s our CEO Phil Blows explaining it all.
AQRU Trend’s methodology and execution
By AQRU Chief Investment Officer Philipp Kallerhoff.
The AQRU Trend strategy is constructed as a time series momentum strategy, following Hurst et. al. 2013. To determine a positive trend in each asset, the strategy simply considers whether the asset’s excess return is positive: a positive past return is considered an “up trend” and leads to a long position. On the flip side, the strategy does not take any short positions, but instead moves into a cash position.
To generate the signal, we use 1-month, 2-month and 3-month look-backs. The 1-month strategy goes long if the preceding 1-month excess return was positive. The 2-month and the 3-month strategies are constructed analogously. Combining the three strategies leads to three possible signals for each asset:
- If all three strategies are long, the signal is 100% long.
- If two out of three strategies are long, the signal is 30% long (2 out of 3 long signals minus 1 out of 3 short signals).
- In all other cases the signal is 0% and the strategy goes into a cash position as described above.
The size of each position is chosen to maximally diversify the annualised volatility across the traded assets, following the methodology of Moskowitz, Ooi, and Pedersen (2012). Specifically, the number of dollars bought/sold of an asset on a given day is the inverted volatility (to promote lower volatilities) of an asset divided by the sum of inverted volatilities (i.e. each asset has a volatility which gets inverted and the sum of these volatilities). This approach leads to 100% exposure if all assets have a 100% long signal and 0% exposure to the market if the signal is 0% for all assets. All Exposure below 100% is moved into a cash position (see above).
The ex-ante annualised volatility for each instrument s is estimated as an exponentially weighted average of past squared returns
where the scalar 365 scales the variance to be annual and r is the return for time t. The parameter δ is left with a standard value of 0.94, which gives about 50% of the weight to the last 10 days of returns (see ‘Exploring the Exponentially Weighted Moving Average’).
The AQRU Trend algorithm is planned to run every Tuesday and outputs the size of each position, then a set of trades are executed to adjust each position accordingly. If the size of a position is above a specific threshold then the trade is split into chunks in order to minimise market impact. The algorithm and trade executions are administered by Accru Finance Ltd.