A new stock screen to exploit a market quirk

To state the bleeding obvious, investors like it when a stock hits a new high. But what if price highs are linked to something less obvious, namely the efficiency with which a company’s earnings are interpreted?

That’s a key insight from a recent paper by Tobias Kalsbach and Steffen Windmüller of the Technical University of Munich (TUM), which regular readers may remember from a column I wrote in the magazine earlier this month (‘A simple way to add more 'alpha' to your portfolio’).

In the study, the pair sought to explain why it can sometimes take markets a long time to digest news with positive implications for earnings, and to find when this post-event price ‘drift’ might be at its most powerful (which may be gradual). To do this, the academics looked at a phenomenon known as the anchoring effect, which relates to the way an individual’s choices or views are influenced by a reference point or 'anchor' that may be irrelevant.

As a market proxy for a psychological anchor, Kalsbach and Windmüller chose 52-week price highs. Across 23 developed markets between 2004 and 2021, the study found that when stocks traded near their yearly highs – an irrelevant data point – investors were more reluctant to buy in on the back of positive news.

This adds psychological nuance to the literature on momentum investing. The inference of the study isn’t that investors should simply buy stocks on the back of positive earnings surprises. Instead, the findings suggest that companies that have posted positive earnings surprises and whose stocks trade near their 52-week highs are most likely to lead investors to underreact to good news.

By extension, investors who buy these mispriced situations should be able to benefit from the slow digestion of the good news, as the market forgets about the psychologically irresistible anchor of the 52-week high watermark. Indeed, the paper found that by factoring this into a trading strategy, investors might add 6.8 percentage points of alpha per year.

Mirroring the anchor effect

For the past couple of weeks, I’ve been fiddling around with several spreadsheets to try and approximate the TUM team’s methodology. The compromise I’ve reached, which I’m calling the Double Top screen, is naturally a lot cruder than the academics’ back testing. But it’s also designed to exploit this potential source of outperformance in real time, and in a user-friendly format.

The screen sets just two tests, which are as follows:

• A share price within 5 per cent of the 52-week high.
• An upgrade in the next 12 months’ earnings per share forecast of at least 3 per cent, over the last month.

For any reader who regularly trawls through the Ideas Farm section of Investors' Chronicle, the rules should be familiar. We already provide lists of UK stocks which have just come close to their one-year highs, and those which have seen the largest one-month bumps in broker earnings forecasts.

While the price test finds stocks whose investors may be 'anchored', there are a couple of downsides with the earnings test.

The first is that broker upgrades are not company news. In their study, Kalsbach and Windmüller use a “high frequency methodology…to identify scheduled and unscheduled firm-specific news” and draw their inferences about delayed reaction. Our screens lack such sophistication. And while our data provider, FactSet, does provide some information on company guidance, there isn’t enough consistency or completeness in the figures for them to be useful.

This is important, for two reasons. First, because the TUM paper’s findings showed that unscheduled news requires more time to be digested. Second, the study also found that analysts’ upgrades were less likely if positive news affected a stock trading near its 52-week high.

In other words, neither trading updates nor brokerages are the perfect proxy source for slowly digested news. Despite this, I am going to stick with forecast upgrades. While it’s likely analysts are affected by the same price anchors as investors, they are also often forced to re-evaluate their recommendations and forecasts when a target price looms into view, or news occurs.

Indeed, the paper found that analysts are influenced by “firm-specific news as they change their recommendations after the arrival of news”. They may not always get the balance right, but analysts’ jobs and incentives require them to recognise good news when they see it.

Other considerations

When building a screen from scratch, there is always the tendency to engage in what statisticians and machine learning experts call ‘overfitting’, or adding too many parameters to perfect an output. It’s always worth remembering that quantitative methodologies will never reveal a scientific rule about investing or markets.

As such, I’m going to overlook the temptation to add more filters to the screen. I could throw in a few tests for quality, exclude stocks trading at a wild premium to the market range, or cull investment trusts. But while the screen’s methodology is blunt, its differentiation lies in its focus on price. Adding fundamental criteria risks polluting the strength of the idea it seeks to capitalise on. 

One aspect of the screen I am not yet sure of is its frequency. Since I retired the quarterly No-Thought portfolio in July, there has been a natural gap in the annual schedule for the kind of momentum-focused strategy that the Double Top screen embodies.

However, there are two reasons why this might not work. For one, while both the No-Thought portfolio and our Momentum Classics screens have demonstrated the power of momentum strategies over shorter time horizons, running the Double Top every three months would make momentum the theme of around a fifth of our weekly stock screens.

Second, although our stock screens are designed to be a source of ideas rather than ready-made portfolios, I still want them to broadly reflect real-life investing conditions. Given this, I am not too keen on the idea of refreshing a portfolio every quarter. Not only does this risk over-trading – and the associated costs that come with this – but it sets the tone.

Of course, refreshing the screen every three months might allow me to quickly change tack, should the performance prove a disaster. It may also be possible to run quarterly and annual versions of the screen simultaneously, to experiment with the drift effect. I’ll have to give it some more thought.

The inaugural screen saw nine stocks pass, including two from Aim. While this is a wider selection than some of our recent refreshes, it’s not a huge haul from the 1,259 listed companies screened. However, because the screen is merely looking to maximise one variable (nearness to the one-year high) and pass one filter (a 3 per cent one-month earnings upgrade), it can easily be adapted to widen the results in future refreshes.

The stocks that make the cut, which are detailed in the table below and ranked by proximity to their 52-week price high, include a few classic momentum names such as InterContinental Hotels (IHG) and London's two super majors, Shell (SHEL) and BP (BP.), which have recently rallied in line with a spiking oil price.

Both have a chequered history of earnings, largely due to the unpredictability of oil and gas prices. Indeed, because the investment case for Shell and BP depends so much on factors beyond the companies’ control, it could be that investors incorporate news in a different way to other stocks.

I’m also interested to find out whether smaller names in the pack might be subject to lower price discovery, and thus more susceptible to the slower-digestion theory. We’ll take a look at how the Double Top is faring in a few months’ time.