How we beat the FTSE 350 four times over

The leading stocks in this selection made an average total return of 4.4 per cent - four times that of the FTSE 350 (the appropriate benchmark) over the same period. As the screen is re-run, however, it is not clear whether excess returns are due to systematic factors or, as the efficient market hypothesis (EMH) would have it, that the outperformance is entirely serendipitous.

The strong form of the EMH (there are thee forms of EMH) is based on the premise that share prices reflect all information, public and private, and therefore no investor can make abnormal returns, except by luck. A challenge to EMH is evidence for behavioural finance phenomena, such as the tendency among investors to rush into momentum stocks, in the realisation that they have previously underreacted to good news about companies (Jegadeesh and Titman, 1993). Another frequent observation is the outperformance of stocks that have rated highly on value factors, such as a low share-price-to-book ratio.

Value and momentum selection methodology

Quantitative research company First Trust Global Portfolios has based value and momentum strategies on the work of leading academics (such as Fama & French, and Carhart) and their system inspired Investors Chronicle's model.

In the screen, momentum stocks are rated on three-month, six-month and 12-month price returns. As a quality test, share-price-to-sales ratio and one-year revenue growth are also looked at. The average ranking across all five factors decides each company's composite momentum rank.

For the value tilts, companies are assessed on price-to-book (P/B) ratio, return on assets (ROA) and the ratio of share price to free cash flow (PFCF). The average score for the three factors is used to assign a composite value rank.

Finally, using the highest of their overall value or momentum scores, the companies are placed in one list. Where Investors Chronicle's methodology differs substantially from First Trust's is that rather than take the top quintile of all stocks in the ranking, we analyse just the top 100. This is still too many companies, so the top 20 (minus any companies where takeover documents are being awaited) make up the final shortlist.

With such a small sample there is the potential to become heavily concentrated in just one or two sectors. This makes it difficult to ascertain whether performance (positive or negative) is down to specific industry factors or the broad risk factors that the system aims to capture. With this in mind, the other four quintiles of the hundred stocks selected by the system are also being tracked.

Value beats momentum in the top quintile from November

The top-ranking portfolio for November ended up containing nine value and eight momentum stocks. Over the past five months, the value stocks have produced a stellar performance, up 10.75 per cent including dividends. Consistent with patterns observed in the past, the momentum stocks displayed a negative correlation and were down by 2.68 per cent, after net dividends. Overall, this meant a 4.43 per cent total return for the 17-stock portfolio.

November 2015 value and momentum composite, top quintile performance

Source: Bloomberg & Investors Chronicle

The system may have thrashed the FTSE 350 benchmark, but there are question marks as to whether it can continue to outperform. Firstly, in total, the 100 companies in the wider value and momentum screen achieved a total return of just 0.93 per cent as opposed to the FTSE 350, which was up 1.14 per cent. The EMH explanation would, of course, be that the top quintile outperformance was a fluke and the disappointing overall returns are evidence that it is impossible to consistently beat the market with a system.

As mentioned, the other concern is the potential for high sector concentration in the top-quintile portfolio. In November the list was reasonably diverse, considering the selection criteria. Re-running the screen in April 2016, however, there is an overwhelming bias towards housebuilders and companies with exposure to the UK's booming property market. Many of these companies are currently on Investors Chronicle's buy list, but the wisdom of holding all of them is extremely questionable, in case the bubble bursts.

April 2016 value and momentum composite top quintile

Source: Bloomberg & Investors Chronicle

With serious misgivings about trying to replicate the top quintile as a live portfolio, is there a practical use for the composite value and momentum screen? The high exposure to UK property may pay off over a short period and the portfolio could in future rebalance to become less concentrated. This would, however, be a haphazard way to manage risk and the system is probably not a suitable basis for investing.

Interestingly, the value picks of the 100 stocks screened in November have made an overall gain of 2.49 per cent, beating the FTSE 350 benchmark. Some commentators have opined that it is time to switch into value from growth/momentum strategies, and perhaps our screen can be used to observe trends.

If not a composite portfolio, maybe tactical tilts?

Market timing is difficult and it would be risky to try to beat the market by simply switching between all value and all momentum stocks. An alternative approach would be to add value or momentum factor 'tilts' to sample portfolios partly selected according to market capitalisation index weightings.

In index construction, tilting weights away from market capitalisation to favour companies with strong factor characteristics is a way of adding exposure to those factors, while controlling deviation from the benchmark. This is done by applying a multiplier (based on metrics similar to those we used for our value and momentum scores) to recalibrate the index.

These solutions at the index level are not really appropriate for us to copy, but it will be interesting to observe a crude take on the tilting methodology, which while unsophisticated is an easy experiment in factor-led strategies.

Before beginning, ground rules must be made on diversification. For these experimental 20-stock portfolios, we will have a maximum of two stocks in any one sector. This may have the effect of diluting some of the factor exposure, but it will reduce unsystematic risk. Furthermore, we will only hold one mega-cap stock (as defined below) per sector.

Value and momentum 'tilt' portfolios

Using the value and momentum rankings for FTSE All-Share companies, from our earlier analysis, we can create two new portfolios. The FTSE 100 weightings will be referenced to select large company holdings, but value and momentum portfolios will be selected from the screened FTSE All-Share universe.

Firstly, isolating the top 10 per cent of the FTSE 100 by market capitalisation (excluding Royal Dutch Shell 'A' shares), gives a list of 10 companies that together comprise approximately 41 per cent of the index. There are two pairs of companies in the same sector - Royal Dutch Shell (RDSB) and BP (BP.), and GlaxoSmithKline (GSK) and AstraZeneca (AZN). In these instances, for the value portfolio we will exclude the company with the lowest value scores and for the momentum portfolio exclude the company with the worst momentum score. It just happens that Shell and GSK beat BP and AstraZeneca, respectively, on both measures. Due to these diversification rules, each portfolio will have eight mega-cap holdings.

Next, another 12 holdings will be selected for each portfolio, from the rest of the FTSE All-Share universe; which has already been screened according to value and momentum factors. Institutional investors prefer to apply factor tilts to market capitalisation weights for liquidity reasons, but for our portfolios we will equal-weight the 12 factor holdings and the eight mega-caps.

'Live' testing

The value screen and the momentum screen are not based on back-testing. This is with good reason as, especially where value stocks are concerned, there is a danger of statistical look-ahead bias. This is where knowledge of complete data - for example, full-year reported sales - is assumed and used to work out screening metrics in back-tests. In reality, full-year results used to compute a ratio may not have been known at the reference date for the back-test, so simulations can effectively behave as though the stock selection had clairvoyant insight. This can seriously skew results and give a misleading picture. These test portfolios, therefore, will be tracked from this point and rebalanced every six months going forward.

At first sight, there is still a high concentration of exposure to the UK property market in the value portfolio, and another danger for both portfolios is that the mega-caps could underperform. Alongside these selections, it will also be interesting to keep an eye on our composite value and momentum system, to see whether it can overcome reservations and continue to outperform.

April 2016 mega-cap and value tilt split

Source: Bloomberg & Investors Chronicle

April 2016 mega-cap and momentum tilt split

Source: Bloomberg & Investors Chronicle