Earnings Revisions Stock Screener Strategy

What do Wall Street legends Martin Zweig, Peter Lynch, and William O’Neil have in common?

It’s their belief that share prices are driven by the improvement, or momentum, of earnings.

Read these articles for a deep-dive into a Martin Zweig stock screener strategy, and Peter Lynch stock screener strategy.

Each have built enviable careers around this philosophy and made an enormous amount of money in doing so.

Do you want to know an insanely profitable stock screener strategy that captures this in a simple way?

Great, because I’m about to show you…

It comes from the idea that stocks whose earnings have been positively revised by analysts recently are likely to exhibit this earnings momentum.

Read my ChartMill review to see this screen in action.

Earnings Surprises

Remember, share prices reflect the prevailing expectations of investors. Therefore, they will only react to information that differs from these expectations. The stock market is forward-looking after all.

In other words, it doesn’t matter what a company’s earnings are in absolute terms; it matters what they are relative to expectations.

When actual earnings differ from consensus forecasts, we call this an earnings surprise.

For example, if a company grows earnings by 50% when they were expected to grow 60%, then all else equal, the market will view this unfavourably and punish the share price. This is an example of a negative earnings surprise.

In contrast, a positive surprise is one where actual earnings exceed estimates.

Post-Earnings Announcement Drift

Studies show that stocks announcing an earnings surprise will drift in that direction for months. In the academic literature, this is a well-known anomaly called the “post-earnings announcement drift”.

The post-earnings announcement drift goes against the “efficient markets hypothesis”, as under this theory we should expect an instantaneous jump to the new fair price after an earnings surprise – and therefore no opportunity to profit from it.

The reason it doesn’t is because of a behavioural bias known as “conservatism bias”. This is where investors tend to put too much weight on old information and are therefore slow to react to new information.

This systematic bias in humans is what creates an opportunity for YOU, as it causes the share price to gradually ‘drift’ towards the new fair price – creating an opportunity to profit from it in the meantime.

Earnings Revisions Lead to Earnings Surprises

Ok, so now you understand why earnings surprises drive stock prices.

But you’re probably wondering where earnings revisions come into it.

To answer this, you need to understand the relationship between revisions and surprises. They are closely linked, which means they share the same logic in terms of their impact on share price movements.

Analysts generally revise earnings for one of two reasons:

1. The company has already announced an earnings surprise, which leaves the analyst scrambling to adjust their earnings estimates closer to reality.
2. They predict that an earnings surprise is coming and revise their earnings estimates leading up to the announcement.

In scenario 1, the analyst is behind the curve. Nevertheless, the revision usually happens straight after the announcement, which means it basically coincides with the surprise.

In the second scenario, the analyst is ahead of the curve. Academic research suggests that earnings revisions before an announcement are a good indicator of a subsequent earnings surprise.

What’s more, the same academic research suggests persistence in earnings revisions. In other words, one earnings revision usually begets several more in the same direction.

This links back to “conservatism bias”, which recall, is the slow processing of new information. In this context, analysts either underestimate future earnings for positive surprises, or overestimate them for negative surprises.

In summary, earnings revisions are either a reaction, or a precursor, to an earnings surprise. Which, as we’ve already established, lead to predictable and persistent movements in share prices that we can profit from.

The backtests below confirm this phenomenon still exists today. That is, picking stocks with positive earnings revisions perform exceptionally well in subsequent months.

AAII EPS Revisions Screen

In case you needed more convincing, AAII have also backtested different earnings revisions strategies, which you can see in the linked article.

As you can see from their research, stocks with positive estimate revisions have significantly outperformed the broader market, while stocks with negative revisions have underperformed.

How to Define Earnings Estimate Revisions?

While all this sounds great, you’re probably asking yourself how exactly to define an earnings revision.

Essentially, an earnings revision is a change in a company’s future earnings expectations. These expectations usually represent an average of sell-side analysts’ earnings forecasts, which you can get from investment services such as Zacks Investment Research or Koyfin.

When comparing these across many stocks, we are more interested in the percentage change in earnings forecasts over a certain period of time. This gives us a more standardised measure to compare and rank across a large group of companies.

This begs the question as to what is meant by future earnings though. Are we talking about next years earnings, or next quarters?

And over what historical time frame should we measure these earnings estimate revisions over? The last month, 3 months, 6 months…?

Earnings Estimate Revisions Backtest

To answer this, I have backtested different combinations of lookback and forecast periods. The aim here is to see whether certain combinations have more predictive power for future stock returns.

For example, with a lookback period of 1 month and forecast period of the current year, we measure the percentage change in current year earnings estimates over the previous month.

The notation for this strategy is 1mFY1. 1m means we are measuring revisions over the previous month, and FY1 is the standard notation for the current fiscal year. Similarly, FQ1 means the current fiscal quarter, and FY2 means the next fiscal year.

We then pick the companies with the greatest upward revisions. For each of the backtests below, I define this as the 10% of companies with the greatest earnings estimates revisions.

The backtests cover different regional markets to see if the strategy works universally. The regions covered are the US, UK, Europe, and Japan.

Due to differing data availability for these markets, the historical sample periods are of different length. In each case though, the hypothetical portfolio is rebalanced monthly, and as already mentioned, constructed of the 10% of companies with the greatest upward revisions.

The stock universes are broad market indices such as the S&P 500. This ensures that enough analysts cover the stocks for revisions to be meaningful.

The results show the cumulative return for each strategy, the corresponding compounded annual growth rate (CAGR), the average annual portfolio turnover, and the average holding period for each stock.

Returns are measured in local currency terms, and all of the data comes from Bloomberg.

As always, backtests come with the disclaimer that they are simulated results and not indicative of future performance.

US

The stock universe for the US is the S&P 500 and the backtest period is from January 1993 to May 2022.

As you can see from the table below, the best performing strategy was 3mFQ1. That is, picking stocks with the greatest percentage change in current quarter EPS estimates over the last 3 months had the greatest predictive power for returns.

Its cumulative return was 8602%, representing a CAGR of 16.3%. This outperformed the S&P 500 by an average annual rate of 6% – leading to a 5.5x greater cumulative return.

It’s also interesting to note that the 1mFQ1 strategy performed just as well as the 3mFQ1 strategy. In the US, it doesn’t seem to matter whether you measure the estimate revisions over a 1 month or 3 month lookback period, as long as the forecast period is for the current quarter.

All charts have a logarithmic scale for ease-of-comparison.

UK

The stock universe for the UK backtests was the FTSE 350, and the sample period was from May 2005 to May 2022.

Quarterly EPS estimates are extremely patchy on Bloomberg for stocks in the UK. Therefore I omit the quarterly strategies and just focus on the yearly ones. The same is true for Europe and Japan.

The best strategy in the UK over the sample period was 3mFY2, with a CAGR of 15.8%. This represents an average annual outperformance of 9% vs. the FTSE 350 – leading to an almost 5.5x greater cumulative return than the benchmark.

Europe ex. UK

The stock universe for the Europe ex. UK backtests was the Stoxx 600 index, excluding UK stocks. The sample period was from May 2005 to May 2022.

As you can see, the best performing strategy was 3mFY2 – with a CAGR of 12.6%. This was 6% better on average than the Stoxx 600 each year, leading to a 3.3x greater cumulative return.

Japan

The stock universe for Japan was the Nikkei 225 and the sample period was from May 2005 to May 2022.

The best performing strategy was 1mFY1, with a CAGR of 12.3%. The 1mFY2 performance was very similar, which shows that measuring revisions over a 1 month lookback period – whether for this year’s EPS estimates or next year’s – yields the best results in Japan.

Both strategies beat the Nikkei 225 by an average of around 5% per year – leading to a more than 3x greater cumulative return.