Reference-Class Bid Review¶
Review — instantiates Winner-Conditioned Valuation Correction
Places a pending bid's estimate inside a class of comparable past contests and reads off the base-rate outcome and the typical field of rivals, producing a debiased, outside-view input before any winning-conditional correction.
Before you can correct an estimate for the fact that you might win, the estimate itself has to be honest — and the deal team's inside-view number rarely is. Reference-Class Bid Review takes the raw appraisal and refuses to argue about it on its own terms; instead it assembles a class of comparable past contests — same asset type, same rough size, same kind of buyer — and asks what actually happened to the people who valued those objects and to the fields they bid into. Its one defining move is substituting the outside view for the inside one: rather than reasoning up from this deal's specifics, it reads the base rate off the class. Two things fall out that the archetype needs downstream: a debiased unconditional estimate stripped of this-time-is-different optimism, and a grounded assumption about how many informed rivals a contest like this usually draws. It is not the winner's-curse correction itself — it produces the clean inputs that correction runs on.
Example¶
A company plans to acquire a competitor and its deal team projects ≈$300M in synergies, justifying a ≈40% premium. The Reference-Class Bid Review sets the projection aside and builds the class: roughly thirty comparable acquisitions in the same sector over the past decade. The pattern is sobering — realized synergies in that class landed a long way below what the acquirers projected, and the winning bidders' premiums were, more often than not, later judged to have been overpayment. The same class shows these deals typically drew three to four serious, well-advised bidders.
The review doesn't declare the deal bad. It hands two corrected inputs to the valuation: an unconditional synergy estimate marked down toward the class base rate (nearer ≈$120M than $300M), and a competitor assumption of ≈4 informed rivals. Those become the raw material for Winner's-Curse-Adjusted Bid Model — which will then shade further for the fact that winning against those four would itself be bad news.
How it works¶
- Define the class before looking at the answer. Fix the set of comparable past contests by objective criteria — asset type, size band, buyer type — while the desired number is still unknown, so the class can't be gerrymandered to flatter the deal.
- Read the base-rate outcome. Pull what winners in the class actually realized, not what they projected, and locate this deal's inside-view estimate against that distribution.
- Extract the field. Record how many informed bidders contests in the class typically attracted, and how sophisticated they were.
- Hand off debiased inputs. Emit a regressed-toward-base-rate unconditional estimate and a grounded competitor assumption for the valuation model to condition on.
Tuning parameters¶
- Class breadth — narrow and highly similar versus broad and loosely comparable. Narrow classes are more relevant but thin and easy to cherry-pick; broad ones are robust but dilute relevance.
- Recency weighting — how much to favor recent contests over older ones when the market has shifted.
- Idiosyncrasy adjustment — how far to let genuine, defensible differences pull the estimate back toward the inside view. Set it high and "this time is different" quietly reappears.
- Outcome metric — which realized quantity defines the base rate: post-deal returns, realized-versus-projected synergy, or observed overpayment.
When it helps, and when it misleads¶
Its strength is that it attacks the optimism before it compounds. The winner's-curse correction shades an estimate for selection; this review makes sure the estimate being shaded wasn't already inflated by the inside view.[n1] It also supplies the field size the whole correction scales on, sourced from history rather than from the deal team's hopes.
Its characteristic failure is the self-serving reference class: draw the boundary narrowly enough and you can assemble a class of flattering comparables, or dismiss every sobering one as "not really like us." That is the review run backwards — the class chosen to license the number instead of test it. It also breaks when the class members aren't genuinely comparable, or when a truly novel deal has no honest class at all. The discipline that guards against this is to fix the class definition before the base rate is known, and to make idiosyncrasy adjustments explicit and defended rather than assumed.
How it implements the components¶
This review realizes the outside-view input stage of the archetype — the debiasing that happens before any conditioning on winning:
unconditional_estimate_trace— it surfaces the deal team's raw inside-view estimate as an explicit object of review and regresses it toward the class base rate, producing the traced, debiased unconditional figure.competitor_information_assumption— it reads the typical number and sophistication of rivals off the class, grounding the field-size assumption the correction scales on.
It does not compute the winning-conditional correction on those inputs — that is Winner's-Curse-Adjusted Bid Model's work — and it does not itself maintain the record of realized outcomes; it consumes that record, which Post-Auction Loss Review builds. The fresh-panel route to challenging an estimate is Independent Valuation Panel.
Related¶
- Instantiates: Winner-Conditioned Valuation Correction — this review supplies the debiased, outside-view inputs the rest of the appraisal conditions on.
- Consumes: Post-Auction Loss Review provides the realized-outcome record that the reference class is read from.
- Sibling mechanisms: Winner's-Curse-Adjusted Bid Model · Independent Valuation Panel · Competing Estimate Simulation · Post-Auction Loss Review · Sealed-Bid Premortem · Bid/No-Bid Gate
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Reference-Class Bid Review operates as an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution because it places a pending bid's estimate inside a class of comparable past contests and reads off the base-rate outcome and the typical field of rivals, producing a debiased, outside-view input before any winning-conditional correction.
Independent corroboration: The frozen evidence defines Reference-Class Bid Review as 'Places a pending bid's estimate inside a class of comparable past contests and reads off the base-rate outcome and the typical field of rivals, producing a debiased, outside-view input before any winning-conditional correction', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Behavioral Economics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Reference-class correction of inside-view estimates is canonical behavioral decision research.
Related originating lineages:
- Economics & Finance — Auction and winner's-curse theory materially supplies the bid-specific conditional valuation problem.
Review resolution: Both blind reviewers agree that behavioral_economics is the primary origin. Explicit reconciliation of alternate origin disagreement, domain reach disagreement adopts reviewer_a's classification because reference-class correction of inside-view estimates is canonical behavioral decision research. The resulting lineage records alternates=economics_finance, origin_mode=cross_disciplinary_synthesis, and domain_reach=specialized; these describe formative provenance separately from later applicability.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] The outside view / reference-class forecasting — estimating an outcome by comparing the case to a class of similar past cases and using their statistical distribution, rather than reasoning from the case's own particulars (the "inside view"). The distinction is due to Kahneman and Tversky; the forecasting application to large projects was developed by Bent Flyvbjerg. ↩