Skip to content

Stock selection in bear markets

Steven Carroll
Steven Carroll
Global Head of Sales Strategy and Execution, Investment Solutions and Data Science

As bear markets emerge in the U.S., stock pickers need to focus on capital preservation. How can Refinitiv StarMine help investment professionals optimise their stock selection in the current dynamic macro environment?

  1. With the current macro environment affected by factors such as the war in Ukraine and rising inflation, stockpickers need to adapt to investing in bear markets.
  2. Refinitiv StarMine Text Mining model leverages artificial intelligence (AI) to analyse different text sources and pinpoint problematic words and phrases that are likely to be an advanced indicator of business distress, and to help forecast probability of default.
  3. StarMine models consistently perform well during heavy periods of volatility and other market dislocations, and can add significant value throughout the market cycle, by providing unique analytics that provides additional context and idea generation.

For more data-driven insights in your Inbox, subscribe to the Refinitiv Perspectives weekly newsletter.

It has been a challenging year for stockpickers.

Numerous factors are making forecasting particularly challenging on the macro front – the duration of the war in Ukraine, the degree to which inflation is transient and supply-led, versus the demand side pressure driven by the U.S. consumer.

The Fed put is obviously nowhere to be seen.

Focusing on capital preservation in bear markets

In such an environment it is crucial that institutional asset managers focus on capital preservation, identifying stocks with a high likelihood of missing and those misnamed as safe-havens, with heightened market expectations and plenty of downside in the event of a disappointment. Refinitiv has numerous unique tools to assist in this regard, all part of our StarMine content.

Figures 1 and 2 give a sense of the decline we’ve already seen in Tech valuations; we can already see the NASDAQ starting to return to the top end of the valuation range it has inhabited throughout the last decade.

Unfortunately, figure 2 highlights the degree to which inflation (and potentially inflation expectations) have slipped their moorings.

In such an environment, it seems premature to assume a bottom, as there seems a much higher danger of interest rates moving beyond the range of the last decade.

No one is completely clear on how to deal with a bond bear market, with few employed managers having worked through the last one under the Volcker Fed.

NASDAQ Comp versus Bond market. Stock selection in bear markets

Inflation genie escapes bottle. Stock selection in bear markets

StarMine: identify opportunities, save time, and zero in on the most viable investment ideas

How can StarMine help?

StarMine is a suite of equity and credit analytics that support institutional asset managers with stock selection and risk management. Its initial focus was on rating and ranking sell-side research analysts and this DNA still informs several of the analytics we offer.

Let’s start with the Smart Estimate and Predicted Surprise.

The Smart Estimate is a differentially weighted estimate, based on IBES data, that incorporates a number of factors to improve the performance relative to the more common equal-weighted forecasts.

The first major difference is the scanning for clusters of revisions. When new information hits the market, be it a changed commodity price, a competitor’s announcement and commentary, a buyback announcement, or commentary from the supply chain (to name a few) the more skilled analysts immediately make changes to their numbers.

StarMine detects these clusters and, until an analyst makes a change subsequent to this cluster, they are temporarily excluded. Quite often this means the Smart Estimate is only working with a small subset of analysts.

With those that are left, their contribution is weighted based on their historical accuracy and the number of days since their estimate(s). The Smart Estimate thus puts greater weight on analysts with a strong track record and recent estimates, and excludes those whose estimate doesn’t incorporate the latest market data.

The difference between the Smart Estimate and the consensus I/B/E/S estimate is called the Predicted Surprise – and when it is negative, there is a significant chance of a company missing estimates and experiencing further negative revisions (as the slower analysts catch up with the faster ones).

Here is a quick sample of large-cap Tech stocks with negative predicted surprises (using data from 10 June).

I’ve included both sales and EPS, and the ARM score (Analyst Revisions Model), which is a percentile rating measuring change in analyst sentiment based on both mean revisions in estimates (across the income statement) and current predicted surprises.

It is an excellent tool for forecasting changes in future revisions and this is, of course, highly correlated with future price movements.

I’ve also included one of our valuation models. In the current environment, high valuations combined with misses are a heightened risk for the portfolio (or a potential short…).

Refinitiv StarMine

On the topic of heightened expectations, I should also measure some of the data exhaust from another of our valuation models.

Our IV model is a standard DDM but with a few twists. It does not rely on sell-side forecasts in the long term, which are notoriously bullish. Rather, it scales them back on the basis of our StarMine Research team’s work on the degree to which these forecasts are inaccurate.

Fortunately, the behavioural biases that drive these errors are persistent, so we can produce a series of longer-term EPS forecasts, called Smart Growth. We then use Smart Growth to make long-term dividend assumptions to determine a fair value.

The model can be of interest for measuring expectations where we switch from determining fair value to assuming the market value is fair. Now we calculate what stream of EPS growth would equate to that value. The market is implying the 5- and 10-year EPS CAGR based on the last close.

Here are the most bullishly priced Large Cap Tech stocks based on our Market Implied Growth measure.

No surprises on some of the names given the enthusiasm for Cyber stocks, but still, in a rising rate environment, these are stocks where disappointment is going to hurt and one’s optimism needs to be tempered.

As you can see from the ARM score, these stocks continue to be underpinned by positive revisions.

Analysts are very much on board with these names (but it is important to note that ARM is a measure of change in sentiment, not a measure of the absolute level). However, there’s a pretty clear correlation between these punchy valuations and the ARM score.

Refinitiv StarMine

The Bold Estimate

Another area of interest based on our analyst work is the Bold Estimate.

We rate and rank five-star analysts on a curve and the top tier is rated ‘5 Stars’. These analysts have a strong track record of forecasting earnings – and of doing so early. Being right on the day is great for headlines, but you need to be right early enough that your customers can position.

We then look for areas where these five-star analysts are making differentiated calls. We call these Bold Estimates. So we have an analyst making a big call (positive or negative) and that analyst has a strong track record of being right.

As an example, the mean estimate for Zoom is $3.71. One analyst has a five-star track record and has made a very bullish call; their EPS number is $4.46. This is the highest estimate of the 28 analysts in I/B/E/S tracking Zoom – but with the crucial context that they have historically been extremely accurate on calling Zoom’s EPS numbers.

Our users can quickly scan across their investable universe, looking for these differentiated calls from leading analysts, which can provide both long and short ideas.

Earnings Quality

So enough of the risk management. What about finding long ideas?

There are a few models here that perform extremely well during heavy periods of volatility and other market dislocations. One is Earnings Quality (EQ).

This is a measure of the sustainability of current earnings by decomposing those earnings into more and less sustainable factors.

EQ identifies companies with conservative reporting (looking at changes in accruals), where earnings are well backed by cash flow and it also uses a Dupont decomposition to look at its profitability (turnover is a more sustainable source of profit than margins, unsurprisingly).

All of this together gives us a nice “safe haven” model that helps both highlight potential harbours but also stocks where earnings are at risk of decelerating as they revert to a more sustainable level.

Below is a selection of Large Cap U.S. stocks with high EQ. The 10-year Market Implied Growth number is also included to give a sense of whether the company is already priced to perfection.

It is interesting to compare some of the new economy names. For example, Zoom with almost 12 percent EPS expected, but against the bullish expectations on Costco at almost 17 percent.

Refinitiv StarMine

Probability of Default

There is a final analytic that is probably of most interest if you do believe the economic backdrop will continue to deteriorate.

StarMine has a variety of credit models designed to help forecast the Probability of Default.

One of the more unusual models is our Text Mining Model which looks at various text sources with some clever AI to identify problematic words and phrases likely to be an advanced indicator of business distress.

A phrase in an earnings call can portend doom much earlier than most credit ratings or market signals, so this is an interesting early signal that can give you an early indication of smoke.

It is noteworthy that the absolute level of PD for all these companies remains low, but the model does move much more quickly than traditional sources and can be usefully incorporated into risk management and investment processes.

Below is a list of Large Cap names where the Text Mining Model is presently scoring at a low level, and probably worth a check to see what text source is generating the concern.

Large cap names where the Text Mining Model is scoring at a low level

StarMine has many other models that I haven’t discussed here. This blog is simply to provide you with a cross-section of some of our unique analytics that can provide additional context and idea generation during what is a very difficult period in the market.

While these analytics are of value throughout the market cycle, in this type of risk-off phase where missteps are punished much more severely, StarMine can add significant value on top of your current provider.

StarMine: identify opportunities, save time, and zero in on the most viable investment ideas


What is the current forecast for U.S. bear markets?

As bear markets emerge in the U.S., stock pickers need to focus on capital preservation.