# **Appendix — Methodological Notes**

---

## **Why Elasticity Is Observational**

Retail prices are not randomized.
Observed price–quantity relationships reflect correlation, not causal response.

This system does not attempt causal identification.
It focuses on robust decision-making given observed behavior.

---

## **Why Bootstrap Is Used**

Closed-form uncertainty assumptions are fragile in pricing contexts.

Bootstrap resampling:

* captures parameter uncertainty
* avoids distributional assumptions
* supports downside-aware evaluation

---

## **Why No Machine Learning Models Are Used**

The pricing decision is low-dimensional.

Additional model complexity:

* increases opacity
* complicates governance
* does not improve decision quality at this stage

ML pricing belongs to later integration phases.

---

## **Out-of-Scope Extensions**

The following are intentionally excluded:

* causal pricing experiments
* promotion-response modeling
* multi-SKU or portfolio pricing
* inventory-constrained pricing
* dynamic or reinforcement learning pricing

These extensions require additional data and governance structures.

---

## **Closing Note**

The system is designed to answer one question well:

> **What price can be deployed with confidence under uncertainty?**

Everything else is deliberately deferred.

---