Market-Making Decision Game

The scenario

This is the take-home form of the interview market-making game. You quote a two-sided market on a security whose value is hidden until settlement. A counterparty trades against the side that favours them — so every fill is a clue about the true value. Quote around your estimate, manage inventory, and learn the value from the order flow.

Where this shows up

Task brief

# Market-Making Decision Game

**Role relevance:** The take-home form of the interview market-making game.
**Estimated time:** 50-60 minutes  ·  **Difficulty:** Intermediate  ·  **Format:** Python (.py), standard library only

## What you are given
- `market_making_decision_game_starter.py` - a fixed game engine (hidden value, counterparty flow, settlement) with the `MarketMaker` class left to you.

## What you must deliver
1. A two-sided quote around a fair estimate, skewed against inventory
2. An order-flow update rule (a fill is information)
3. A run comparing learning vs no-learning
4. One line on the adverse selection you observe

## Constraints
- Standard library only; the engine and fill model are fixed.

## Submission note
Implement `MarketMaker.quote` and `on_fill`, run it, then check the mark scheme and solution.

Your tasks

1. 01Implement a quote: a reservation price (fair − skew·inventory), then bid/ask at ± a half-spread.
2. 02Skew the quotes against inventory so your position mean-reverts toward flat.
3. 03Update your fair estimate from each fill — an ask lifted means the value is likely higher; a bid hit, lower.
4. 04Run the game and compare learning vs no-learning; explain the adverse selection you see.

How you're assessed

The full points-based mark scheme is included with the pack.

What you'll learn

• ▸Why a fill is information, not just a trade — order flow reveals value.
• ▸How adverse selection runs over a stale quote, and how learning limits it.
• ▸How skewing quotes manages inventory while you provide liquidity.

Study alongside