Files
zmVault/bid-price-modeling.md
T

124 lines
3.8 KiB
Markdown

---
id:
aliases: []
tags:
- authorship/llm
- destiny/fleeting
- status/complete
- topic/estimating
- topic/risk
- type/idea
title: Modeling Bid Prices Under Intrinsic Cost Uncertainty
---
# Modeling Bid Prices Under Intrinsic Cost Uncertainty
> [!warning]
> This text is almost entirely LLM output.
> I don't intend to keep or use any significant portions of it.
Consider a construction project characterized by an intrinsic but unknown final cost $C$.
Prior to project completion, $C$ cannot be observed directly;
instead the estimator possesses only a probability distribution over feasible outcomes.
Let $(\Omega, \mathcal{F}, \mathbb{P})$ be a [probability space](https://en.wikipedia.org/wiki/Probability_space)
describing the estimator's uncertainty,
and define the true cost model as a non-negative random variable
$$
C : \Omega \to [0,\infty)
$$
with [distribution](https://en.wikipedia.org/wiki/Probability_distribution) $\mu_C$.
The distribution $\mu_C$ summarizes all available information at the time of bidding,
including quantities, labor productivity uncertainty,
market conditions, subcontractor pricing variability,
and correlation structures inherent to the estimator's assumptions.
Although the natural mathematical representation of cost is thus a distribution,
procurement mechanisms typically require that each bidder submit a single deterministic price.
Denote this bid by
$$
B \in [0,\infty).
$$
> Read as
> "$B$ is an [element](https://en.wikipedia.org/wiki/Element_(mathematics))
> of the interval from zero to infinity, including zero."
A bid $B$ may be viewed as the output
of a pricing [functional](https://en.wikipedia.org/wiki/Functional_(mathematics))
$$
\Phi : \mathcal{P}([0,\infty)) \to [0,\infty),
$$
mapping a cost distribution $\mu_C$ to a scalar.
Examples of such functionals include:
### 1. Risk-neutral expectation
$$
\Phi(\mu_C) = \mathbb{E}[C],
$$
> Read as "Phi of mu sub C equals the expected value of C."
where $\mathbb{E}[\cdot]$ denotes the [expected value](https://en.wikipedia.org/wiki/Expected_value).
### 2. Risk-adjusted expectation
$$
\Phi(\mu_C) = \mathbb{E}[C] + \lambda\sqrt{\mathrm{Var}[C]},
$$
> Read as "Phi of mu sub C equals the expected value of C plus lambda times the square root of the variance of C."
where $\mathrm{Var}[C]$ is the [variance](https://en.wikipedia.org/wiki/Variance)
and $\lambda>0$ is a risk-loading parameter.
> This mirrors mean--variance pricing common in portfolio theory.
### 3. Quantile-based pricing
$$
\Phi(\mu_C) = Q_p(C),
$$
> Read as "Phi of mu sub C equals the p-quantile of C."
where $Q_p$ is the $p$-[quantile](https://en.wikipedia.org/wiki/Quantile)
of the distribution.
### 4. Utility-maximizing bid
Under a bidder [utility](https://en.wikipedia.org/wiki/Utility) function $U$,
$$
\Phi(\mu_C) = \arg\max_{b\ge0} \; \mathbb{E}[\,U(b - C)\,].
$$
> Read as "Phi of mu sub C equals the argument b greater than or equal to zero
> that maximizes the expected value of U of b minus C."
> [$\arg\max$](https://en.wikipedia.org/wiki/Arg_max) is the value of $b$ that maximizes the expression.
***
The central tension is:
* The ontologically correct representation of project cost prior to execution is a **probability distribution**, whereas
* The procurement mechanism requires a **deterministic scalar**.
The study of such pricing functionals $\Phi$ sits within
[stochastic optimization](https://en.wikipedia.org/wiki/Stochastic_optimization),
[risk measures](https://en.wikipedia.org/wiki/Risk_measure),
and [mechanism design](https://en.wikipedia.org/wiki/Mechanism_design).
Understanding how different choices of $\Phi$
compress and distort the underlying uncertainty $\mu_C$
has direct implications for bidder profitability, competitive strategy,
and how risk is allocated across the construction market.