Quant models
Factor models built from statistics
Five classic factors scored every month across the S&P 500. Separate from our rules-based models.
Factor models built from statistics, scored monthly across the S&P 500. Backtests are hypothetical; the live record is kept apart.
The five factors
Each factor on its own: the top fifth of the S&P 500 by that factor, equal weight, 2016–2026 backtest.
| Factor | Per year | Max drop | Sharpe |
|---|---|---|---|
| Value | +13.1% | -36.6% | 0.72 |
| Momentum | +13.9% | -19.7% | 0.88 |
| Quality | +12.4% | -23.8% | 0.81 |
| Low volatility | +9.6% | -20.6% | 0.79 |
| Size | +12.1% | -38.4% | 0.64 |
| All stocks, equal weight | +13.2% | -26.5% | 0.84 |
| S&P 500 (SPY) | +15.6% | -23.9% | 1.05 |
How the quant models work
Value. Earnings, book value and sales compared with market value, each against the same sector.
Momentum. Return from 12 months ago to 1 month ago (the latest month is skipped).
Quality. Return on invested capital, operating margin and low debt, against the same sector. Banks and insurers use return on equity and net margin.
Low volatility. How little the stock's daily price moved over the past year.
Size. Smaller market value scores higher (within the S&P 500, so still large companies).
Each month every factor becomes a z-score; the model averages them and holds the top fifth. Rules were fixed before testing, not tuned on the results.
How these differ from Our models
Our models follow a few fixed screens (for example momentum or growth rules) and hold 10–20 stocks. The quant models score every stock on several statistical factors at once and hold about 95, so they behave more like a broad factor fund than a concentrated pick list.
Disclosures
Backtests are hypothetical. They apply the rules to past prices and SEC filings with trading costs of 10 bps per side; no taxes or other slippage. They are not real results.
Survivorship bias. Stocks count only while they were in the S&P 500, but companies that were later acquired or delisted have no price data here and are missing, which can flatter past returns.
Data limits. Fundamentals come from SEC XBRL filings, used from their first filing date; tagging gaps leave some companies without a value or quality score. Market values use filed diluted share counts.
Live record. Starts at the Oct 30, 2026 close and is reported separately; it is never blended with the backtest.
Research only: model portfolios for information, not personal advice and not accounts anyone manages. Past results don't predict future returns. See the disclaimer.
Independent research · Not investment advice · Hypothetical unless marked Live · Data: Yahoo Finance (15-min delayed), SEC EDGAR, House Clerk · Important disclosures · Contact desk@coteau.app
Beta 0.9 · Oct 21, 2026