YWR: Why Estimate Revisions Work

I talk about EPS estimate revisions a lot and estimate revision momentum is 60% of the Global Factor Model ranking.

But why is it important, why does it work, and how do we use it?

It’s time I explain my thinking behind the estimate revision score and how I use it.

But first let me tell you a story about Muppet Capital, a London hedge fund I worked for early in my career.

• Muppet Capital and Earnings Surprise Day

• Anchoring Bias and Autocorrelation

• Academic Literature

• Nvidia Example

• The Global Factor Model and Finding the Story

Muppet Capital LLC

I used to work as a fundamental analyst at a $3bn hedge fund which loved to trade earnings day surprises.

Every day the PM would look for chart patterns which roughly resemble the drawing below.

A stock bops along for a few months, generally trending upwards, has a bit of a sell off, stabilises for a few months, then starts to trade up into an earnings release. Then on the day of the earnings release the stock goes up 5-15%. The more the better.

I was a complete monkey at this firm.

When the PM found one of these set ups (and he had computer system for finding them) I would receive an email with the ticker and a form to fill out. I was supposed to write a quick paragraph of why the results surprised and whether earnings estimate were likely to go up.

I wasn’t supposed to say what I thought about the company, or if I liked management, etc. I was the human in the loop to just double check that an earnings surprise had happened. I was also supposed to call 2-3 sell side analysts and get their view on what they thought about the results and their earnings estimates.

After I had quickly gone over the earnings release, spoken to 2-3 analysts to confirm there was a fundamental earnings surprise and estimates were going to have to increase, I would fill out the email form. All of this was supposed to happen in under 2 hours. Once the process was complete the fund would buy the stock. +5%, +15%, it didn’t matter.

The expected pattern was that the stock would likely tread water at this new higher level for a few days. It might even give back 2-3%, which was fine, but after 5-10 days it would start to trade up and make new highs. The positive earnings release was the kick off for a new multi-month trend higher.

Over 6 months I probably went through this process over 100 times across every sector in Europe. The pattern was the same. It didn’t matter the company, sector, country, or which analysts covered the company.

But why did this stupid pattern work? What was going on?