Nowcast Before Forecast
Establish current conditions before predicting what comes next
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 97%
Begin by determining the economy's current state rather than jumping directly to a distant forecast. Examine contemporaneous inflation, employment, housing, spending, and financial conditions, then compare that evidence with the assumptions embedded in consensus expectations. Convert disagreements into probabilities instead of treating outcomes as binary certainties. Because markets are often broadly right, take a contrarian position only when several relevant indicators reveal a meaningful gap between present reality and prevailing expectations. The output is a conditional forecast grounded in a defensible nowcast: if current momentum persists, one path becomes more likely; if the underlying evidence changes, revise the forecast. This process reduces dependence on stale narratives and makes forecasts easier to update and communicate.
Origin
Neil Dutta describes learning probability-weighted forecasting and selective disagreement with consensus from economist Ethan Harris. Extracted from Masters in Business.
Core principles
- 01Diagnose the present before predicting the future
- 02Weight evidence by its current relevance
- 03Treat forecasts as probabilities rather than certainties
- 04Challenge consensus selectively, not reflexively
- 05Use multiple indicators to confirm the prevailing regime
How to run it
- 1
Frame the Decision
Specify the forecast or investment decision and the time horizon it must address. Separate what must be known now from what must be estimated later.
Pro tip Express the decision as a question with observable implications.
Watch out A vague question encourages selective use of evidence.
- 2
Build the Nowcast
Assess current inflation, employment, housing, spending, production, and financial conditions. Prioritize evidence that describes present momentum over backward-looking headlines.
Pro tip Start with indicators closest to the mechanism behind the forecast.
Watch out Do not mistake one volatile release for a durable regime change.
- 3
Map Consensus Assumptions
Identify what markets and forecasters currently expect. Note which assumptions would have to hold for that consensus outcome to occur.
Pro tip Distinguish the stated consensus from what asset prices actually imply.
Watch out Do not oppose consensus merely because contrarian views attract attention.
- 4
Weight the Scenarios
Assign relative probabilities to plausible outcomes based on the nowcast and the mechanism connecting current conditions to each outcome. Include evidence that could falsify the leading scenario.
Pro tip Use conditional language that connects each scenario to measurable developments.
Watch out Avoid presenting a low-probability possibility as the base case.
- 5
Pick the Battle
Depart from consensus only where the evidentiary gap is large enough to matter. Translate the disagreement into a clear forecast or decision.
Pro tip Concentrate on the few calls where your information advantage is strongest.
Watch out Constant contrarianism puts you repeatedly on the low-probability side.
- 6
Update Continuously
Reassess the nowcast as new data arrive and revise probabilities when the underlying mechanism changes. Preserve the reasoning trail so changes are disciplined rather than reactive.
Pro tip Predefine the observations that would cause you to change your view.
Watch out Do not defend an outdated forecast after its premises fail.
In the wild
In late 2021, Dutta observed firmer inflation, unemployment falling faster than expected, and a powerful housing recovery. Those current conditions suggested that the Federal Reserve was behind the economy and would need to raise rates more aggressively than the sanguine consensus expected. He forecast at least four increases, an out-of-consensus call that ultimately proved conservative.
→ A present-state diagnosis produced a more accurate directional policy forecast than the prevailing consensus.
An analyst can assess recession odds by first checking employment, claims, household income, housing, and company behavior. If the economy is still creating jobs, claims remain low, and no credible transmission mechanism explains a contraction, the analyst can assign less probability to recession despite widespread forecasts.
→ The forecast becomes conditional on observable mechanisms rather than recession rhetoric.
Common mistakes
Forecasting Before Diagnosing
Starting with a preferred future narrative encourages analysts to force current data into that conclusion. Establish the nowcast first.
Being Contrarian for Attention
Consensus is often directionally correct, so indiscriminate opposition creates repeated low-probability bets. Disagree only when the evidence and mechanism justify it.
Treating Outcomes as Binary
Economic outcomes usually combine competing forces and evolve over time. Use probabilities and conditional scenarios instead of absolute declarations.
Is it for you?
Best for
Economic researchers, investors, and strategists making decisions under uncertain and rapidly changing conditions.
Not ideal for
Situations where reliable contemporaneous evidence is unavailable or the decision depends primarily on non-economic factors.
From the transcript
“So I think the first thing to do in this business is you want to make sure you have the now cast right right, forget…”
“in this business, it's about picking, like weighing probabilities and then picking your battles with the consensus wisely.”
“I'm not the kind of person that's just going to be contrarian for the sake of being so”
From the episode
Neil Dutta on Economic Research