Who calls the turn?
Wall Street's consensus earnings forecast treats every analyst the same. I built a Reversal Index (Ri) that scores analysts on the quarters where earnings change direction, added it to an analyst-ranking model, turned the result into a 20-year trading backtest, and then re-ran everything out of sample to see what held up.
The consensus is reversal-blind
Forecasts are easy when earnings keep trending. They break down at the turn.
Tesla is headed for $2,600 a share.
Tesla has been wildly overvalued for years.
Analysts disagree all the time. The useful question is whose forecasts deserve more weight, especially when the trend is about to break.
IBES collects every sell-side analyst's EPS forecast. Averaging them gives the market consensus that investors, traders and the financial press quote. Shoorvazi (2025) improved on the plain average by scoring each analyst on accuracy, information diversity and consistency, then averaging only the top 25%.
That framework treats every quarter alike. But the quarters where EPS reverses direction are where most analysts get caught extrapolating the old trend, and those are the quarters that matter most for prices. My idea was to measure, analyst by analyst, who is good at exactly those quarters.
What counts as a reversal
A mechanical rule with two conditions: the direction flips, and the move is large enough not to be noise.
- 01Direction flips:
sign(ΔEPSt) ≠ sign(ΔEPSt−1) - 02Move is material:
|ΔEPSt| / |EPSt−1| > 10%
Applied to Apple, 65 of 156 quarters (42%) qualify. Reversals are not rare events.
Quarterly EPS has a calendar. A holiday-quarter spike followed by a normal quarter looks like a "reversal" every year. Summing to annual EPS strips that seasonality out. Try it below.
The Reversal Index
How good is an analyst on the hard quarters, relative to how good they are on ordinary ones?
Better on reversals than usual. A comparative advantage on the quarters that are hardest to call.
No difference. The analyst is as good, or as bad, at the turn as anywhere else.
Worse on reversals. The analyst keeps extrapolating the old trend after it breaks.
From analyst scores to a trade
Ri becomes the fourth factor in Shoorvazi's scoring model. The top-scoring analysts form a refined consensus, and the gap between it and the market consensus becomes a trading signal.
Average of 1 − |forecast − actual| / |actual|, clipped to [0, 1].
How far the analyst's forecast typically sits from the crowd.
Inverse-variance share of the analyst's forecast errors over time.
Reversal-quarter accuracy divided by normal-quarter accuracy.
The pipeline, with a real event: Apple, 29 Oct 2020
Score analysts
Grid-search w₁…w₄ in 0.2 steps per stock; score every analyst.
Keep the top 25%
Rank by score and keep the top quartile.
Refined consensus
Average the selected analysts' EPS forecasts.
$0.837Compare to market
Market consensus averages everyone.
$0.781Signal
(refined − market) / |market|
+7.2%Sort & hold
Each quarter, split stocks into terciles by signal; hold 60 trading days.
- Signal
- +7.2%
- Actual EPS
- $0.73
- 60-day return
- +29.0%
- Tercile
- Top
Two answers to the same question
I first evaluated the model in-sample. An independent review then rebuilt the pipeline walk-forward, using only information available at each point in time. Switch between the two below; every chart on the page follows the switch.
In-sample, as submitted
Weights are grid-searched on all of a stock's history, including the quarters being scored. This is the version in the final report and deck.
Walk-forward, audited
Features, Ri and weights use only prior periods. Forecasts must be dated before the announcement. The 3-factor baseline is optimised independently.
Phase 3: does Ri improve the forecast? 15-stock pilot
Reversal-quarter forecast error: 4-factor vs 3-factor
% reduction in MAE, right is betterData table
Phase 4: annual or quarterly Ri?
Does the signal make money?
68 stocks, 2005–2024, 4,837 earnings announcements. The top tercile is held long; the bottom tercile is also held long as a control; the long-short is top minus bottom, dollar-neutral.
Growth of $1, 60-day holding window
log scale60-day results
Sharpe ratio by holding horizon
trading daysLong-short return by sub-period
annualised, 60-dayWhat survived the audit
Every headline claim from the final report, re-tested walk-forward.
| Claim | In-sample | Walk-forward | Verdict |
|---|
What the walk-forward rebuild changed
- No look-aheadAnalyst features, Ri and weights are estimated only on periods before each test quarter.
- Forecast timingOnly forecasts activated before the announcement date count.
- Fair baselineThe 3-factor model is optimised on its own, not derived from 4-factor weights.
- Overlapping positionsDaily accounting treats overlapping 60-day windows as real concurrent positions.
- True dollar-neutralLong-short is long basket minus short basket; cash when one side is empty.
What I take away
Analyst heterogeneity is real
Ri splits analysts almost evenly, so it captures something accuracy alone does not. Turning that into a better forecast is a separate, harder problem.
A grid search can invent an edge
Picking the best of dozens of weight combinations per stock, on the same data used to score it, was enough to produce a +12% "improvement" that disappears out of sample.
Beating SPY is not alpha
The top tercile beat the S&P 500, but so did the bottom tercile. When the control leg drifts up just as much, the return is market exposure, not signal.