Market Speed

Overview

This project studies how liquid markets respond to scheduled macroeconomic releases using free, institutional-grade public data.
Macro events (CPI, Non-Farm Payrolls) are treated as exogenous, time-stamped shocks to the market. The system response is measured through changes in returns, volatility, and cross-asset correlations.

The main purpose of this project was to learn core quant research skills by building an event‑driven market study with free data, clean code, and clear visuals.

---

Research Questions

1. Do scheduled macro releases materially change market volatility?
2. Is the event-day move statistically large relative to recent price behavior?
3. Do macro events increase or decrease cross-asset co-movement?
4. Are effects symmetric across asset classes and event types?

---

Data Sources

• Daily prices: Stooq via pandas-datareader
• Macro event calendar: FRED via pandas-datareader
  • CPI
  • Non-Farm Payrolls (NFP)

All data is pulled programmatically and cached locally to ensure full reproducibility.

---

Methodology

Each macro release is treated as a known event date. For each asset and event:

1. Define a pre-event window and post-event window (±5 trading days).
2. Compute return-based statistics before and after the release.
3. Compare event-day returns to a rolling historical baseline.
4. Measure changes in cross-asset correlation structure.

No model fitting or forecasting is performed at this stage. The focus is on empirical response diagnostics, not prediction.

---

Key Metrics

• vol_delta
  Post − pre volatility, where volatility is the standard deviation of daily returns over the event window.

  • > 0 → higher noise after the event
  • < 0 → volatility compression after the event

• ret_delta
  Post − pre mean daily return (drift shift).

• impact_ratio

impact_ratio = abs(event_day_return) / median(abs(returns[-20:]))

• ≈ 1 → event day is typical
• > 1 → outsized macro impact
• < 1 → muted reaction

---

Repository Structure

market-speed/
├─ README.md
├─ environment.yml
├─ .gitignore
├─ src/
│  ├─ data_loader.py         # Stooq daily loader + returns
│  ├─ event_loader.py        # CPI/NFP events from FRED → data/events.parquet
│  ├─ daily_event_min.py     # Pre/post (±5d) stats → CSV + console medians
│  ├─ event_day_impact.py    # Event-day impact ratios → CSV + console medians
│  ├─ daily_event_study.py   # Volatility boxplot + correlation heatmap → PNGs
│  └─ basic_tests.py         # t-tests + written summary → TXT
├─ data/                     # Cached Parquet/CSV (generated)
└─ figures/                  # PNGs/CSVs/TXT outputs (generated)