Our last post on using factors as filters described the factors we expose to users and described some things to be careful of when applying the filters. The method of sorting is a more intuitive concept and is overall, less impactful to a strategy than the filters chosen.

A sort is only applied after the filtering has occurred. This is the way it has to happen for good reason

• Filtering based on factors shrinks the universe of stocks down to only those stocks that pass each and every filter. This means that on a given day your investment strategy will have a list of the stocks that have passed all the filter criteria
• The sorting is then employed to order these stocks so you get for example the top 15 stocks
• Any sorting can therefore only be employed after this filtering process - as otherwise if 100 stocks passed all the filters, unless we employ some preference or sort, there would be no way to know how to select your top 15 stocks. This is why one sort, at least, is needed and enforced.

When using more than one sort, we do not incorporate any ordering. Each factor is used to generate its own individual score for each stock and we calculate the final sort-score for a particular stock by summing up all of the individual rankings for that stock - this is why order is not important. The final top 15 stocks are then chosen by looking at the total sort-score of all the stocks against each other. We can show a small example of this below, where Nike, for instance, has the highest momentum and liquidity. If for this case, we only wanted to return 2 stocks the sort would return NKE and AMZN as they have the two best final sort scores.

Some of the most popular factors to sort on, and they were the ones we used in our Xplore value strategy example, are

• momentum
• volatility
• liquidity

These are very popular sort factors since they represent some of the most natural and important concepts in terms of investing. Sorting your final list would be a prudent thing to do, in order to ensure for example that the least volatile stocks or the most liquid stocks are included in your strategy.

Other users, more familiar with fundamental company financial data, may instead choose to order by specific factors. For example, you could sort by price-to-earnings, operating cash flow ratio or by gross-margin. With our Xplore tool, you can start to build your knowledge and experience and start to experiment.

We are going to take our first value strategy example we had designed and the only thing we are going to change is the sorting method applied to it. This will hopefully illustrate the sensitivity of an investment strategy design to its underlying choices. We can see that the strategy itself produced some decent returns and a good Sharpe ratio and has performed quite well since covid hit.

Our first value strategy backtest output showing the total return, sharpe and portfolio graph.

Part of the reason we chose to sort by momentum, volatility and liquidity for example was that our factor selection was used to identify the value stocks and then we wanted our sort to incorporate some other information - such as liquidity and volatility.

Now let's change the sort and remove those three and replace them with one sort. We choose a classic value style metric such as price-to-earnings and choose to sort using LOWEST order, thus giving us what appear to be the stocks that exhibit the most value. The evidence is clear, with this choice of sort we lose an estimated 5% annually and our Sharpe ratio is halved from what it was before.

We can see the clear impact of the choice of a single price-to-earnings ranking, instead of momentum, volatility and liquidity on the backtest of our strategy.

Our Xplore Stocks tool is designed to be very dynamic and robust enough to investigate all sorts of investment strategy design decisions. Please let us know any feedback and enjoy building your own strategies!