LaunchTower — Momentum + Quality Factor Dataset (2026-09-14)

Independent market-data research. A fully reproducible momentum + quality factor screen on 151 US large-caps, computed from public Yahoo Finance price data. Every number in this CSV can be regenerated from scratch with the included script.

⚠️ Disclaimer: This is research/educational output from public market data. Not personalized investment advice, not a recommendation to buy or sell any security. Past performance does not guarantee future results. No return or performance promises are made.

---

📦 What's in This Repo

File	Description
data/factors_2026-09-14.csv	Full 151-stock factor table — 14 columns, ranked by composite score
data/factors_sample_2026-09-14.csv	15-row sample (top 10 + bottom 5) for quick inspection
scripts/reproduce.py	Exact Python script to regenerate the dataset from scratch
README.md	This file — full methodology, sample output, and how to use the data

---

🔬 Methodology (Transparent & Reproducible)

Universe

151 US large/mid-cap equities spanning tech, healthcare, financials, energy, industrials, consumer staples, and materials. Tickers are listed in data/factors_2026-09-14.csv.

Data Source

• Yahoo Finance via yfinance (free, public, auto-adjusted prices)
• Window: 252 trading days (2024-09-16 → 2026-09-14)
• No paid data, no proprietary feeds.

Factor Definitions

Column	Definition
ret_1m	Simple return over the last ~21 trading days
ret_3m	Simple return over the last ~63 trading days
ret_6m	Simple return over the last ~126 trading days
ret_12m	Simple return over the last ~252 trading days
vol_ann	Annualized volatility = std(daily_returns) × √252
max_drawdown	Maximum peak-to-trough decline over the 12-month window (negative)
from_52w_high	last_price / 52w_high − 1 (≤ 0)
momentum_score	Cross-sectional z-score of ret_12m (mean 0, std 1, ddof=1)
quality_score	Cross-sectional z-score of −vol_ann (lower vol → higher score)
composite_score	0.5 × momentum_score + 0.5 × quality_score

Composite Score Formula

composite = 0.5 × z(ret_12m) + 0.5 × z(−vol_ann)

Where z(x) = (x − mean(x)) / std(x) computed cross-sectionally across all 151 tickers (ddof=1).

Interpretation:

• Positive composite → strong 12-month momentum and/or low volatility
• Negative composite → weak momentum and/or high volatility
• The score is a ranking tool, not a signal. It tells you where each stock sits relative to its peers on these two dimensions.

What This Is NOT

• ❌ Not a trade signal or buy/sell recommendation
• ❌ Not a prediction of future returns
• ❌ Not a backtest (no out-of-sample validation included)
• ❌ Not a substitute for your own due diligence

---

📊 Sample Output (Top 10)

Rank	Ticker	Last Price	12M Return	Ann. Vol	Max DD	Composite
1	MU	$975.26	+548.8%	81.4%	−39.1%	+2.204
2	VLO	$390.42	+153.0%	36.2%	−12.1%	+0.824
3	INTC	$102.94	+318.3%	79.6%	−42.0%	+0.786
4	MPC	$395.93	+120.8%	34.3%	−18.3%	+0.669
5	PSX	$259.47	+101.6%	30.9%	−17.3%	+0.630
6	JNJ	$265.58	+52.1%	19.1%	−11.0%	+0.611
7	CSX	$48.95	+50.9%	22.2%	−11.6%	+0.526
8	TTE	$91.89	+56.1%	24.6%	−20.1%	+0.499
9	TGT	$155.83	+77.2%	30.6%	−13.3%	+0.483
10	TRV	$375.20	+36.3%	20.8%	−8.8%	+0.469

Bottom 5 (Weakest Composite)

Rank	Ticker	Last Price	12M Return	Ann. Vol	Max DD	Composite
147	HOOD	$112.57	−4.4%	72.2%	−57.3%	−1.075
148	ZS	$164.54	−42.6%	63.2%	−64.9%	−1.092
149	MRNA	$143.97	+467.0%	192.6%	−34.2%	−1.095
150	COIN	$175.26	−45.9%	70.8%	−63.6%	−1.303
151	SMCI	$40.10	−8.8%	91.6%	−65.0%	−1.587

Note: MRNA ranks near the bottom despite +467% 12M return because its 192.6% annualized volatility dominates the quality penalty. This is exactly what the composite is designed to surface.

---

📈 Distribution Statistics

Statistic	Value
N tickers	151
Composite mean	0.000
Composite std	0.463
Composite range	−1.587 → +2.204
Median composite	+0.099
N positive	89 (59%)
N negative	62 (41%)
12M return range	−58.8% → +548.8%
Ann. vol range	18.4% → 192.6%

---

🛠️ Reproduce It Yourself

pip install yfinance pandas numpy
python scripts/reproduce.py

The script:

1. Downloads 252 days of daily adjusted close prices for all 151 tickers from Yahoo Finance
2. Computes all factor columns exactly as described above
3. Writes factors_YYYY-MM-DD.csv to the current directory

Expected runtime: ~30–60 seconds (network-bound). No API key required.

Key Code (Inline)

import yfinance as yf
import pandas as pd
import numpy as np

TICKERS = [/* 151 tickers — see full list in the CSV */]
LOOKBACK = 252  # trading days

prices = yf.download(TICKERS, period="2y", interval="1d", auto_adjust=True)["Close"]
prices = prices.dropna(axis=1, thresh=int(LOOKBACK * 0.8))  # drop tickers with >20% missing

def compute_factors(prices: pd.DataFrame) -> pd.DataFrame:
    last = prices.iloc[-1]
    ret_1m  = last / prices.iloc[-22]  - 1
    ret_3m  = last / prices.iloc[-64]  - 1
    ret_6m  = last / prices.iloc[-127] - 1
    ret_12m = last / prices.iloc[-253] - 1

    daily = prices.pct_change().dropna()
    vol_ann = daily.std() * np.sqrt(252)

    cummax = prices.cummax()
    drawdown = prices / cummax - 1
    max_dd = drawdown.min()

    hi_52w = prices.max()
    from_high = last / hi_52w - 1

    z = lambda s: (s - s.mean()) / s.std(ddof=1)
    momentum = z(ret_12m)
    quality  = z(-vol_ann)
    composite = 0.5 * momentum + 0.5 * quality

    df = pd.DataFrame({
        "last_price": last, "ret_1m": ret_1m, "ret_3m": ret_3m,
        "ret_6m": ret_6m, "ret_12m": ret_12m, "vol_ann": vol_ann,
        "max_drawdown": max_dd, "from_52w_high": from_high,
        "momentum_score": momentum, "quality_score": quality,
        "composite_score": composite,
    }).sort_values("composite_score", ascending=False).reset_index()
    df.insert(1, "rank", range(1, len(df) + 1))
    df.insert(2, "date", prices.index[-1].strftime("%Y-%m-%d"))
    return df[["ticker", "rank", "date", "last_price", "ret_1m", "ret_3m",
               "ret_6m", "ret_12m", "vol_ann", "max_drawdown", "from_52w_high",
               "momentum_score", "quality_score", "composite_score"]]

---

🎯 How to Use This Data

• Screening: Filter composite_score > 0.5 for a momentum + low-vol shortlist
• Relative value: Compare ret_12m vs vol_ann to find stocks that are "cheap" on risk-adjusted momentum
• Portfolio construction: Use composite_score as a weighting input in your own optimizer
• Research: Extend the model with additional factors (value, size, liquidity) using the same z-score framework
• Backtesting: Use the 12-month return windows as a starting point for out-of-sample validation

---

📧 Get the Full Pack

This repo contains the free sample (151-stock CSV + methodology + reproducible script).

Want more? The full LaunchTower Factor Pack includes:

• Weekly updated factor screens (every Friday)
• Extended universe (300+ tickers)
• Additional factor columns (value, size, liquidity)
• Backtest results and performance attribution
• Python notebook with interactive exploration

👉 Get the Full Factor Pack on Whop — one-time purchase, instant delivery.

---

📄 License

MIT License — use, modify, and redistribute freely. Attribution appreciated but not required.

📬 Contact

Questions or feedback? Open an issue on this repo or reach out via the Whop product page.

---

Generated by LaunchTower — independent market-data research. Not affiliated with Yahoo Finance or any exchange.