Portfolio Active Risk – Create (or Upload) and Assess

The app takes away much of the complexity of building a new portfolio or improving the return/risk profile of an existing one. When creating a new portfolio, the app organises the stock selection with 12 shortlists and provides a target number of holdings for each in a way that is consistent with your choice of benchmark. Once the selection (or the upload) is complete, the app calculates the various sources of active risk, flags the more provocative ones, and suggests options to mitigate their impact.

Portfolio Assessment

What is active risk?

Active investing seeks to generate high active return (i.e. portfolio return above index return) with low active risk (i.e. keep the volatility of portfolio’s active returns in check).

Active risk therefore reflects the propensity of a portfolio to generate returns that will deviate from that of the chosen index of reference (ie. benchmark). It is different from volatility which reflects the absolute risk of the portfolio, irrespective of a benchmark.

A higher volatility will increase the risk of large absolute drawdowns (i.e. many consecutive days of large negative returns), whereas a higher active risk will increase the risk of large relative drawdowns (i.e. many consecutive days of large negative active returns).

The active risk of a portfolio will tend to increase as:

  • its number of holdings decrease
  • its number of smaller stocks relative to larger ones increase
  • its differences (e.g. in country/industry allocations) with the benchmark increase
  • its benchmark is more concentrated and volatile

Number of portfolio holdings

You may upload portfolios with up to 120 positions but when creating a new portfolio, the app will restrict the number of holdings to between 20 and 40. The lower bound corresponds to the minimum recommended to capture most of the advantages from diversification, while the upper bound corresponds to the maximum beyond which the portfolio rapidly loses its potential for outperformance.

Portfolio Upload

The app provides a simple .csv file template that can be used to upload an existing portfolio, which only requires two sets of data:

  • A column named Ticker to identify the stock
  • A column named Qty to report the corresponding number of shares.

You can of course directly upload any .csv file that contains these two fields (for instance a .csv file downloaded from your broking account), just make sure that the tickers follow the right “ExchangeTicker ExchangeCode” syntax.

FEVR LNFevertree’s exchange ticker followed by the ticker for the London Stock Exchange
AAPL UWApple’s ticker followed by Nasdaq’s GS exchange code
9983 JTFast Retailing’s ticker followed by Tokyo Stock Exchange’s ticker

Choice of benchmark

The broader the universe selected, the broader the investment opportunity set. That’s why outside of Japan (given its size and unique characteristics), butterwire only proposes regional and global benchmarks.

Portfolio weights

For new portfolios, the calculated number of shares to buy for each holding corresponds to an equal weight portfolio. Once the composition of the portfolio is confirmed, investors may revise the number of shares by using the Edit function, bearing in mind that it will affect the return/risk outlook of the portfolio.

Holdings Selection

Portfolio builder for smart alpha selection

This functionality organises the stock selection based on the benchmark chosen to:

  • minimise the chances of risk concentration in specific industries and countries, while nudging you to select stocks with above average base (and recession) scores
  • diversify your exposure to value and momentum scores (unlike with fitness, similar scores are more liable to perform in sync)
  • achieve your (ballpark) controversy target

The builder creates 12 shortlists with typically 1 to 3 stocks to select (+/- 1). Each shortlist can be extended by activating the “Show More Stocks” option, in which case the initial list of candidates will be complemented by lower score options as well as “high octane” candidates (represented with a brown background).

Unlisted stocks can be manually added at any time.

Portfolio risk assessment (and mitigation)

Once the selection is complete, the app moves to its assessment screen (pictured above) which visualises the various risk exposures embedded in the portfolio. Outside of the charts reporting scores (see Style bets), the unit used in the charts is “number of holdings equivalent” (this is to make it easier to visualise how many more/less holdings are exposed to a given risk relative to the benchmark). The colour coding is equally intuitive: green means within target range, red means above target range, blue means below target range (except for the style bets bar chart which uses amber to reflect a score that is too close to that of the benchmark). Nothing prevents you from going above/below any of the target ranges, but the app gives you the option to try and mitigate any material active exposure:

Portfolio return/risk outlook

The app’s default setting is to target an expected return of around 2.5% above the benchmark (active return) with a standard deviation of around 5% (i.e. active risk/tracking error of around 5%) and a market exposure close to 100% (ie. portfolio “market beta” between 0.85 and 1.15). It seeks to achieve this by maximising the contribution from company-specific risks while keeping other risk factors in check (ie. low active positive/negative exposures with regards to specific countries, industries, macro variables, etc). The app will therefore alert investors when portfolio return, or risk parameters fall outside of default target ranges:

  • Too low an expected active return suggests trying to raise the overall base score of the portfolio (click on the base score bar in the “style bets” section for assistance in doing so)
  • A “mitigate” option will appear if the market beta of the portfolio is deemed too low or too high (meaning the portfolio is liable to underperform, respectively in upmarket or downmarket conditions). As always, this is not a prescription as there may be perfectly valid reasons to do so. For instance, an investor may run a high beta portfolio but also keep a high cash balance; as the calculation excludes any cash holding (whose beta is 0), a portfolio that shows a beta of 1.2 but also holds 20% cash would have a market beta of 80% x 1.2 = 1.0
  • An estimated active risk that looks too low suggests a portfolio that is too much like the index (ie. closet-indexing). Too high a value suggests that portfolio returns are liable to deviating substantially from that of the index; if coupled with high volatility, this points to an overly aggressive portfolio and the potential for large swings in absolute and relative returns. Active risk can be raised or reined in by affecting the exposures taken, be it specific country / industry / macro / size exposures or by changing the portfolio controversy score (use the mitigate function for suggestions on how to do so).

Mitigation example

Risk mitigation

Click on bar (or mitigate button when applicable) corresponding to the unwanted exposure and a window will appear with the following selections:

  • Top two holdings most contributing to the selected risk exposure – click on the bin icon to remove any of them from portfolio construction
  • Top two holdings which cumulate both a high exposure to the selected risk and a high contribution to other sources of active risk -- click on the bin icon to remove any of them from portfolio construction
  • List of stock ideas whose risk profile would have an offsetting effect on the risk exposure of the portfolio – click on the shuffle button to see more offset candidates, and click on the check icon to add any of the suggested stocks

Style bets

This part of the assessment is the most important as it shows the portfolio’s base score relative the selected index, and it is this score that drives the estimation of portfolio “alpha” (active return forecast). The app’s target setting is to achieve a base score of at least 1.5 over that of the benchmark. If the value represented for base score is below 1.5 (amber-coloured bar or blue bar if below 0), clicking on the bar will open the mitigate window with suggestions to replace the lowest score selection with higher score ones. The same applies to Fitness, Value and Momentum where the higher the score the better, with one caveat for Value and Momentum. Whereas stocks with high Fitness scores do tend to perform for reasons that are largely company-specific (ie. little added risk from increasing the portfolio fitness score), it is not necessarily so for Value and Momentum which are “style factors” that are more liable to generating correlated returns. The app therefore computes a risk indicator of insufficient diversity of value and momentum scores (see value and momentum dispersion), with an option to mitigate the risk when the dispersion is deemed too low. Likewise, for the portfolio controversy score, where too high a value may result in the kind of volatility and exposure to large drawdowns (sequences of very negative daily returns) that may prove too risky and taxing for most investors.

Size bets

Everything else equals, one should always favour the biggest stock by market capitalisation over the smaller one: it’s easier and cheaper to trade, plus it’s more economical from an active risk budget standpoint (as larger stocks carry a bigger index weight). The problem is that they are a lot less large stocks to choose from than they are smaller ones (for instance in North America, 20 “mega-cap” stocks represent 30% of the total market capitalisation of the largest 1,000 listed stocks), and so the pool of interesting candidates will naturally be over-populated with smaller stocks. As a result, an active portfolio will tend to be “under-weight” the larger stocks and “over-weight” the smaller ones. While there is no conclusive evidence that over long (20 year) periods the difference in returns between portfolios tilted toward smaller versus larger firms is significant, the volatility of this difference in returns is however significant and so size is a relevant consideration for shorter-term (1-5 year) horizons. The app’s default setting is to keep the underweight to mega and large cap stocks to over -5 holdings equivalent and will suggest mitigating the exposure whenever it falls below this number.

Country and Industry bets

These charts help visualise the extent that a country or industry exposure is broadly in line with the chosen benchmark (green bar), significantly positive (red bar) or negative (blue bar). The app doesn’t take side as to whether to go “overweight” or “underweight” a given country or industry and absent a personal preference or conviction, so should you. Any deviation away from the benchmark has a cost in the form of higher active risk, so one needs to believe in the likelihood of getting paid for taking the risk, for instance if deciding to follow some of the country exposure recommendations presented in the “Markets” section of the app.

Global macro bets

This section provides an indication of how (in)sensitive the portfolio can be expected to be relative to the index, when a specific macro variable moves up by 10%. While a reported impact of +/- 1% suggests a negligible exposure of the portfolio relative to the index, high numbers reflect a clear conviction on the direction of the macro variable, whether intended or not. For instance, a portfolio showing a +5% relative sensitivity to a 10% rise in oil price (i.e. to long-dated Brent futures), would have significantly underperform the index back in Sep-14 when oil price expectations collapsed. The mitigate function is available to deal with any such exposure whenever needed.

Pair-wise correlations (double-down bets)

The app tests the correlation of each holding with all others and reports those that look suspiciously elevated. This is a particularly important consideration with concentrated portfolios where each holding is expected to contribute to performance in its own way, so that if one fails it does so in relative isolation from the rest. A high pair-wise correlation suggests one may be taking twice the same bet, potentially exposing oneself to twice the disappointment.