Skip to main content

Consensus vs Actual & the Surprise Reaction

Updated Jun 23, 2026 at 8:47pm

Research Draft Medium 1,314 words

Markets price expectations, not facts. By the time an economic release or an earnings report is published, the consensus forecast is already embedded in the price — so what moves the tape is the surprise: the gap between the actual number and what was expected (surprise = actual − consensus). A "strong" print can sell off if the strength was already anticipated and positioned for; a "weak" miss can rally if it was less bad than feared. The central, counterintuitive lesson is that the level of a number tells you little; the deviation from consensus and the market's reaction to it are what is tradeable, because the reaction reveals how the market was positioned going in.

How expectations are formed

Two distinct expectation tracks exist, and confusing them is a classic error:

  • Published consensus (the official number). For macro data, this is the median of an economist survey — vendors poll forecasters and publish a consensus (Bloomberg, Reuters/LSEG, and others run these surveys; the figure on a data calendar is that median). For company earnings, it is the aggregated mean/median of sell-side analyst EPS and revenue estimates (FactSet, Refinitiv/LSEG, Zacks, Bloomberg). This is the visible bar a print "beats" or "misses."
  • The whisper number (the unofficial number). Analysts continuously update internal models but do not re-publish a report for every revision, so their true current expectation often drifts away from the stale published estimate. That unpublished expectation — passed among trading desks and favored clients — is the whisper number. A Bloomberg News study cited by reference sources found whisper numbers missed actual reported earnings by ~21% versus ~44% for published consensus, i.e. the whisper is often closer to reality. The practical consequence: a stock can beat published consensus yet fall, because it missed the higher whisper the active money was actually trading against. (This node's sibling "Whisper Numbers & Positioning Into Events" covers the whisper/positioning mechanics in depth.)

"Priced in" is the bridge between the two: if a strong number was widely expected, traders have already bought ahead of it, so the good news produces little or negative follow-through — there is no one left to buy, and pre-positioned longs sell the event to lock in the move.

How the surprise reaction works

When the number hits, traders instantly compute actual − consensus and re-price. The reaction has two parts worth separating:

  • The initial move — the gap/spike in the seconds-to-minutes after release, scaling roughly with the magnitude of the surprise. High-impact macro prints (CPI, NFP, FOMC) routinely move index futures, rates, and FX in one tick.
  • The follow-through — whether the move holds and extends over the session and following days. Follow-through is the more informative part: it reveals whether the surprise actually shifted the consensus narrative or was a one-print blip that gets faded. A large surprise that fails to follow through (reverses intraday) is itself a positioning signal — the news was already in. The decision of whether to ride the release or fade it is its own sibling node, "Trading the Release vs Fading the Move."

Reactions are frequently counterintuitive because the interpretation of a number depends on the macro regime. The canonical case is "good news is bad news": when the market's dominant fear is tight monetary policy, a strong jobs or hot inflation print is read as "the Fed stays higher for longer," and equities fall on good economic data. In a growth-scare regime the same hot print is bullish (recession fears ease). The data didn't change meaning — the Fed reaction function the market is pricing did. (Both have dedicated siblings: "The Fed Reaction Function" and "'Good News Is Bad News' Regimes.")

How it's used in practice

  • Know the calendar. Treat scheduled high-impact releases (FOMC, CPI, PCE, NFP/jobs, and a company's earnings date) as binary, time-stamped volatility events. Mark them; never be surprised by one. (Sibling: "The High-Impact Release Calendar.")
  • Trade the surprise, not the headline. Frame every print as actual vs consensus (and vs whisper), then watch follow-through to read positioning. The reaction is the signal; the absolute number is context.
  • Aggregate surprise as a regime read. The Citigroup Economic Surprise Index (CESI) aggregates many macro surprises into one series (weighted standard deviations of actual-vs-survey over a rolling ~3-month window, with time decay). Positive = data broadly beating; negative = broadly missing. By construction it is mean-reverting — it cycles, so an extreme reading flags that expectations have over- or under-shot reality and are due to recalibrate, not that a trend will persist.
  • For swing positions, the binary-event rule dominates: holding a multi-day swing into a scheduled binary event (earnings, FOMC, CPI) exposes you to gap risk where direction is largely a coin flip on positioning you cannot observe. The standard discipline is to size down, hedge, or be flat into the event — capturing a clean directional swing usually means not betting the house on a number whose reaction you can't predict.

Standing & honest caveats

This is a robust, widely accepted description of how markets process scheduled information — the surprise-driven, expectations-relative nature of reactions is well documented across macro and earnings research, and post-earnings drift is a long-studied (and debated, possibly decaying) anomaly. But several honest limits apply:

  • The reaction is regime-dependent and reflexive. The same surprise can produce opposite reactions in different regimes (growth-scare vs inflation-scare). There is no stable mechanical map from "beat → up." Anyone selling a fixed rule is overfitting.
  • Pre-event positioning is unobservable. "Priced in" and the whisper number are inferences, not measured inputs. You only confirm positioning after the reaction, by reading follow-through — which is exactly why this is expectation-management, not a clean signal.
  • Surprise indices have weak/contested predictive value for equities. Aggregate-surprise series like CESI describe sentiment vs forecasts well and track bond yields reasonably, but their direct edge for forecasting equity returns is limited and disputed; treat them as a regime gauge, not a trade trigger.
  • The first move can be a head-fake. Algorithmic spikes on the headline number frequently reverse once the details (revisions, internals, guidance) are digested. Reacting to the headline tick is a known way to get whipsawed.

System relevance

This node is the conceptual hub of the Economic Data Calendar & High-Impact Events branch; its specialized mechanics live in siblings — Whisper Numbers & Positioning, The Fed Reaction Function, "Good News Is Bad News" Regimes, and Trading the Release vs Fading the Move — and it cross-links the Earnings & Guidance → Earnings Surprises & Revisions node, the swing-trading Holding Through Earnings or Not and Post-Earnings-Announcement Drift (PEAD) nodes, and the options Vol-Crush Around Earnings node. For Delvantic's Augustus swing-setup agent, the operative caveat is hard: do not treat a beat/miss as directional on its own, and flag any candidate swing that would be held into a scheduled binary event (earnings, FOMC, CPI) as carrying un-modelable gap risk — the regime context (supplied by the regime engine) determines whether a given surprise is bullish or bearish, and pre-event positioning is not observable in advance.

Sources