Competing Estimate Simulation¶
Simulation model — instantiates Winner-Conditioned Valuation Correction
Simulates the whole field of rival estimates to see where the winning bid lands in that distribution — quantifying how much winning implies you overshot, and flagging when correlated information makes the overshoot worse.
A shading rule asserts how much to discount; Competing Estimate Simulation derives it. Its distinguishing move is to build the rival field explicitly — draw many synthetic competitor estimates from an assumed error distribution, run the contest thousands of times, and read off, from the runs you win, how far above true value the winner typically lands. The output is a selection-adjusted valuation: the expected worth of the object conditional on your being the one who won it, rather than its worth in the abstract. Because it simulates the whole field rather than applying a factor, it can capture what a flat rule cannot — fat tails, asymmetry, and above all correlation among rivals' errors, which is the quiet multiplier of the curse.
Example¶
A carrier is preparing for a spectrum auction in which several national rivals will bid for the same license blocks — a common-value setting, since the licenses are worth roughly the same discounted future cash flows to whoever holds them. Analysts have a valuation, but so does everyone else, and all of them rest on the same demand forecast. The simulation draws each rival's estimate as true value plus noise, repeats the auction ≈100,000 times, and in each run records who wins and how the winner's estimate compares to truth. It reports that with five bidders the winner overshoots true value by, illustratively, ≈15% on average — and by far more once the rivals' errors are modeled as correlated, because then everyone is optimistic together and the winner is the most optimistic of an already-inflated crowd. That selection-adjusted number, not the raw valuation, is what the bid should be built on. Under the hood the simulation is estimating the expected maximum of several noisy draws — an order-statistics problem.[n1]
How it works¶
The engine is a Monte Carlo over rivals. Assume a distribution for competitors' estimate errors and a correlation structure among them; simulate the field; condition on the runs in which you win; and read the winner's average overshoot and its spread. The result is a distribution, not a point — you can vary the number of bidders and the correlation and watch the correction move, which is exactly the sensitivity a fixed rule hides.
Tuning parameters¶
- Error-distribution assumption — the spread and shape of rivals' estimate noise. Wider or fatter-tailed distributions produce a larger simulated overshoot.
- Correlation structure — how correlated rivals' errors are. This is the single biggest driver of curse severity and the main reason to simulate rather than apply a rule; understating it makes the curse look mild.
- Number of competitors — more bidders make the winner a more extreme order statistic, enlarging the correction.
- Runs and resolution — Monte Carlo sample size, trading precision against compute.
When it helps, and when it misleads¶
Its strength is that it captures what a flat shade misses: it prices the effect of many bidders, of skewed value distributions, and especially of correlated signals — the case where everyone reads the same optimistic report and the curse compounds. It hands the downstream bid a defensible, distribution-based correction instead of a guessed factor.
Its weakness is the weakness of any simulation: the overshoot it reports is only as good as the assumed error and correlation, and those are hard to know and easy to set optimistically. Understate correlation and the model blesses a bid that will still be cursed. Its classic misuse is tuning the input distribution until the simulation returns the number you wanted. The discipline that guards against it is to set the error and correlation assumptions from outside evidence — past auctions, how much rivals' data overlaps yours — before running, and to stress the correlation upward rather than down.
How it implements the components¶
Competing Estimate Simulation fills the modeling subset — the components that quantify the selection effect itself:
winner_signal_model— the simulation is the model of what winning signals: by conditioning on the runs you win, it measures how optimistic the winner's estimate must have been.selection_adjusted_valuation— it derives this by simulation: the distribution of the object's value conditional on winning, from which the corrected estimate is read; the Winner's-Curse-Adjusted Bid Model is what commits that figure analytically into a bid.correlated_signal_warning— it models correlation among rivals' errors directly and surfaces when shared information turns a mild curse into a severe one.
It does NOT convert that number into a bid — the fast rule is Common-Value Bid Shading Rule and the full derived bid is Winner's-Curse-Adjusted Bid Model — and it does not decide entry (the Bid/No-Bid Gate).
Related¶
- Instantiates: Winner-Conditioned Valuation Correction — it produces the quantified selection correction the rest of the appraisal consumes.
- Sibling mechanisms: Common-Value Bid Shading Rule · Winner's-Curse-Adjusted Bid Model · Bid/No-Bid Gate · Reserve Price or Walkaway Limit · Sealed-Bid Premortem · Independent Valuation Panel
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Simulates the whole field of rival estimates to see where the winning bid lands in that distribution — quantifying how much winning implies you overshot, and flagging when correlated information makes the overshoot worse, making its operative form a computation, comparison, model, or analytic representation used to infer, estimate, or choose.
Independent corroboration: The frozen evidence defines Competing Estimate Simulation as 'Simulates the whole field of rival estimates to see where the winning bid lands in that distribution — quantifying how much winning implies you overshot, and flagging when correlated information makes the overshoot worse', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Economics & Finance
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: Auction theory cohered simulation of rival estimates to estimate winner's-curse bias conditional on winning a common-value contest.
Related originating lineages:
- Statistics & Experimental Design — Monte Carlo simulation and order statistics supply the computational estimator and sensitivity analysis.
Review resolution: Both reviewers agree on economics_finance as primary. Reading the source mechanism confirms that its defining operation belongs to that lineage; the final record retains statistics_experimental_design only where it materially formed the mechanism and keeps present-day application breadth separate from provenance.
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] Order statistics — the winning bid is the maximum of the bidders' estimates, and the expected value of the maximum of several noisy draws sits above the common mean by an amount that grows with the number of draws and their spread. Simulating the field estimates exactly this quantity, which is why more bidders and more noise both enlarge the correction. ↩