How ClearMacro Transforms Raw Data Into Alpha

by | Jan 31, 2020

At A Glance

ClearMacro was approached by a hedge fund client to build an investment signal using equity risk premium (ERP) data.

We tested the signal and found that it had value; we ran a simulation where we went long the top five rated equity countries, and short the bottom five (equally weighted and using equity ETFs). We re-ran the simulation every month and changed the top five and bottom five holdings in accordance with the simulation output. The performance was acceptable, but not great (IRR of 3.99%, and drawdown of 13.06%). However, we found that combining the new ERP signal with two existing ClearMacro signals vastly improved the performance (IRR of 7.47%, and drawdown of 10.27%).

The takeaway for investors is that transforming raw data into investment signals has value in and of itself – it can help investors allocate long/short to different equity markets. But the value can be improved substantially when combined with other signals, increasing the alpha you can capture.

 

The Backstory

At ClearMacro we transform data into alpha, extracting the return insight available in structured and unstructured information.

One of our hedge fund clients recently asked us to create a bespoke investment signal using the ASR Composite Equity Risk Premium data. They wanted to see if transforming the raw data into a ClearMacro signal could generate alpha in global equities.

 

What is the ASR Composite Equity Risk Premium?

The data is a composite of nine academic models representing different models that estimate future long-term equity returns relative to long-term bond returns.

 

How ClearMacro Transforms Raw Data Into Investment Signals

We transform raw data into a standardized signal score, that moves between 1 and 10.

A signal build always starts with a plausible hypothesis – in this case, that high long-term return estimates are also good for return outcomes over shorter term horizons. In the case of the Equity Risk Premium signal, a score of 10 represents an equity market that has the highest expected risk premium compared to its history, while a score of 1 represents an equity market that has the lowest expected risk premium compared to its history.
Figure 1 below shows the history of this signal, represented by the blue line. The purple line is the performance of the SPDR ETF. The right-hand axis shows our signal score.

Figure 1 – Transformed ERP signal plotted against S&P 500 data

Why Does This Matter to Investors?

Data is a valuable, but risky, asset. Sourcing the right datasets, and transforming them into signals, is both expensive and time consuming. By taking care of the heavy lifting, ClearMacro saves the client the effort of doing it themselves, which in turn saves time and costs.

 

How Useful is This Signal?

To test this, we ran a simulation where we allocated across global equity markets based on signal ranking – we went long the top five rated equity countries, and short the bottom five (equally weighted, and using equity ETFs). We assumed a notional portfolio of $1m, and then allocated $200k to each of the top five countries, and -$200k to each of the bottom five countries. The transaction cost was assumed to be 0.75%.

We re-ran the simulation every month and changed the top five and bottom five holdings in accordance with the simulation output. We also ran a similar simulation for all our other value signals (both short and long-term value), as a comparison.

The results of the test are provided in Figure 2 below. The performance of different “value” signals were ranked high to low, based on six criteria (IRR, Sharpe, 5-year Sharpe, Sortino ratio, Drawdown and Equity correlation). We have highlighted the combined rating in red. The user can change the weighting of each of these criteria, e.g. if he wanted to only focus on IRR, he could put 100% in the IRR column.

The Composite Equity Risk Premium signal is the top ranked signal, with an IRR of 3.99%, and drawdown of 13.06%.

We also ran a simulation using the raw composite equity risk premium data, but it did not make the cut, and is therefore not included in the list below; i.e. transforming the raw data into a signal is critical to extract maximum value from the data.

Figure 2 – ClearMacro’s existing value signals, and the new Composite ERP signal, rated against each other

Why Does This Matter to Investors?

To extract the maximum value from raw data, it needs to be transformed into a signal. And once a signal is created, it needs to be tested. By running a simulation we determine how good the signals is at allocating to equities, and subsequently provide our clients with a list of equities they should go long, and short, on a monthly basis, to generate alpha.

 

What if we Combine the Signal With Other ClearMacro Signals?

Aggregating signals generally provides better performance than if we use a single signal. We therefore tested if the new Composite ERP signal has any value as part of a multi-factor (signal) model.

We found that the signal does indeed have value, and it made the cut to be one of only three signals in our selected / best-in-class multi-factor tactical allocation model for developed market equity countries (Figure 3). The other two signals are the Strategic Leading Activity signal and the Alexandria Equity News Sentiment signal. The performance when combining the signals was an IRR of 7.47%, and a drawdown of 10.27%.

Figure 3 – When combining the new ERP signal with two existing signals, the performance of the model improves

Figure 4 below shows the cumulative performance of the model (purple) vs MSCI World total return in USD (blue).

Figure 4 – Performance of the three-factor model, versus the MSCI World Total Return

Why Does This Matter to Investors?

A signal on its own is powerful, but when combined with other signals it’s even more so. We provide our clients with a library of signals, and we test them to see which combination of signals offer the best alpha for their chosen asset class.