Analyst Calls Digest: Noise or Signal? Block, Equifax, Lockheed—Checking the Backtests.
# Analyst Calls Are Just Unsolicited Signals — Treat Them Like Any Other Noisy Feature
Monday’s upgrade-downgrade carousel — Block, Boise Cascade, Equifax, FirstEnergy, Lockheed Martin, Lumentum, NetApp, Roper, Spotify — arrived on schedule, as it does every trading day. Wall Street research calls are a data stream like any other. The problem is most retail traders ingest them as truth rather than as features with documented predictive decay and a hit rate that rarely survives transaction costs.
The quantitative record on analyst revisions is actually decent — that’s the part the skeptics get wrong. Post-earnings revision momentum is one of the better-documented post-earnings announcement drift companions: upward estimate revisions clustered after a print tend to drift positive over the following 4–12 weeks. But a Monday-morning price target hike with no earnings catalyst behind it is a different signal class entirely. It’s correlated with recent price action, which means you’re mostly buying momentum that already happened, at a spread, after the move.
If you insist on trading these, build the pipeline properly. Pull the calls via a structured feed, normalize them into an ordinal scale (upgrade > PT raise > reiterate > downgrade), then condition on whether the analyst acted before or after the relevant information event. An upgrade following a guidance pre-announcement is stale; an upgrade against recent negative price momentum is the rare case worth flagging, because it implies a fundamental view diverging from price — occasionally the seed of genuine alpha, more often a lagged reaction to a 13F you already saw.
The names on today’s list illustrate the problem with headline-driven entries. Equifax and NetApp get calls that hinge on macro rate paths and enterprise IT spend cycles — inputs you’d want to model explicitly, not inherit from a sell-side narrative. Lumentum’s calls historically track optical transceiver order flow, which is a genuinely leading indicator if you can get the data; the analyst note is a delayed, smoothed copy of it. Lockheed’s calls are essentially defense budget beta with extra steps.
Risk treatment matters more than the signal itself. Position-size these trades as you would any low-Sharpe overlay: small, stop-managed, and never the reason a book has drawdown. If your backtest can’t show the call-event strategy beating a simple 20-day momentum filter on the same universe net of slippage, the analyst call added nothing but latency.
The efficient move: log every Monday call into a local dataframe, tag the catalyst class, and let twelve months of forward returns tell you which analysts — by name — carry real information. Some do. Most are narrative dressed as research, and no price target ever survived contact with an actual order book.