Trading Analyst Estimate Revisions
The direction sell-side analysts are revising their earnings estimates — not the level of the estimate itself — has historically predicted returns, because a wave of upgrades often reflects real information diffusing through the analyst community before it's fully priced in.
Prerequisites: How Short Selling Works
Sell-side analysts publish earnings estimates that everyone can see, so trading on the level of an estimate — is this stock cheap relative to consensus earnings — is already priced in by the time it's public. What's harder to arbitrage away quickly is the direction of change: when a cluster of analysts start raising their estimates for a stock in the same week, that revision often reflects real information (a supplier's earnings call hinting at strong demand, a channel check) diffusing through the analyst community gradually rather than all at once, and price tends to keep drifting in the direction of the revisions for weeks afterward.
The signal: revision breadth and magnitude
The standard construction is an earnings estimate revision score, combining how many analysts moved their estimate and by how much over a trailing window, often standardized as:
aggregated across all covering analysts, and frequently paired with a simpler "breadth" measure: the number of analysts raising estimates minus the number cutting them, divided by total analysts covering the stock. In words: is the whole analyst community leaning the same direction, and how far.
Worked example. A stock covered by 20 analysts has a consensus FY1 EPS estimate of $2.00 at the start of the month. Over the following four weeks, 14 analysts raise their estimates (averaging a 6% increase each) and only 2 cut theirs, moving the consensus to $2.09 — a revision score of +4.5% with strong breadth (14 up vs. 2 down, or +60% net breadth). Historical studies of this signal find stocks in the top decile of revision breadth and magnitude have outperformed the bottom decile by roughly 5–10% annualized, with the effect concentrated in the first one to three months after the revision wave, consistent with gradual information diffusion rather than an instantaneous repricing.
It's the direction and breadth of estimate changes across the analyst community, not the absolute estimate level, that predicts subsequent returns — the level is already public and priced.
What this means in practice, and what erodes it
Revision signals are a standard input in quantitative "fundamental momentum" or "estimate revision" factor sleeves, often combined with price momentum since the two are correlated but not identical (price can move ahead of or lag estimate changes). The edge decays with speed of access: funds paying for real-time consensus-estimate feeds (I/B/E/S and similar) can trade the signal within hours of a revision; funds working off monthly data capture a much-decayed version of the same effect. The signal is also weaker or reversed around known bias periods — analysts are documented to be systematically more bullish on stocks their firm has an investment-banking relationship with, which can distort breadth measures for widely-banked large caps.
Don't treat an individual analyst's single estimate change as the signal — one analyst moving alone is noisy and often idiosyncratic to that analyst's model updates. The predictive power comes from breadth: multiple analysts moving the same direction independently, which is much harder to explain as one person's modeling quirk.
Practice in interviews
Further reading
- Womack, Do Brokerage Analysts' Recommendations Have Investment Value? (1996)