Futures Statistical Arb

No. While market-neutral, stat arb risks include: spread diverging further (blow-out), correlation breaking down, execution slippage, and liquidity issues. It's lower risk than directional trading but not risk-free.

SXF-SCF (S&P/TSX 60 vs Composite) is the most reliable: highest, most stable correlation and the deepest liquidity, with an ~80% spread margin credit on the Montreal Exchange. The 60-vs-Financials pair (SXF-SXA) is the truest sector pair but SXA trades thinly. Bank pairs such as RY-TD are the classic single-name trade, often executed in the underlying shares given thin share-futures liquidity.

You need margin for both legs, though the SXF/SCF spread receives ~80% margin relief on the Montreal Exchange. The standard SXF (C$200 x index) is institutional-scale (~C$300,000 notional per contract at a 1,500 index level); retail traders typically use the mini SXM (C$50) and SCF (C$5). Expect roughly C$15,000-25,000 of spread margin for a mini pair; C$50,000-100,000 of starting capital allows proper sizing across pairs.

For SXF-SCF at the Z > 2.0 threshold: typically 3-5 opportunities per month. Higher thresholds (Z > 2.5) mean fewer but higher-probability setups, and the quarterly roll week adds basis noise.

For broad-index pairs (SXF-SCF): 1-4 days. For bank-stock pairs (RY-TD): 3-8 days. Based on the half-life of the spread. If there is no reversion by 2-3x the half-life, consider exiting.

Regress daily returns of A on B over 60-90 days. Beta = slope coefficient. Alternatively: Beta = Covariance(A,B) / Variance(B). Update weekly as beta drifts.

Fundamental changes (takeover, sector rotation), regime shifts (risk-on/off), earnings surprises, or - very common in Canada - commodity-price moves (oil and gold) that hit the energy- and materials-weighted Composite differently from the bank-heavy 60.

Montreal Exchange index futures are quarterly only (Mar/Jun/Sep/Dec), settling on the third Friday and cash-settled. Roll the quarterly contract ~1 week before expiry and roll both legs together via a calendar or Inter-Group Strategy spread. Bond-futures legs (CGB/CGF) are physically deliverable - exit before the first notice day.

Ratio spread (A/B) is better here because the levels differ enormously (the 60 is ~1,500 while the Composite is ~25,900); the ratio normalizes the relationship. A simple spread (A-B) only works for similarly priced instruments.

Often used interchangeably. Traditionally, pairs trading is simpler (correlation-based), while stat arb uses more sophisticated statistics (cointegration, factor models, ML).

Model hedge ratio as a random walk state variable. Use Kalman filter to recursively estimate optimal ratio as new data arrives. This adapts to changing relationships better than fixed regression.

Z-score level and momentum, rolling correlation, volume ratio, days since signal, VIXC level, and sector momentum (energy and materials are dominant Canadian factors). Avoid too many features (overfitting). Feature-importance analysis is crucial.

Test against: spread blow-out (Z to 5), correlation break (to 0), liquidity crisis (5x wider spreads - more acute in Canada's thinner book), and historical events (the 2020 COVID crash, the 2022 rate shock, and the 2025 tariff/commodity shocks). Ensure survival in all scenarios.

Shorter half-life (1-2 days): Can use larger position (faster exit, lower holding risk). Longer half-life (5+ days): Smaller position (capital tied up, more exposure to regime change).

Monitor: rolling correlation (a drop signals a regime change), half-life changes (lengthening means slower reversion), win-rate decline, and Z-score distribution changes. Commodity-price regimes (oil, gold) drive Canadian cross-sector relationships - watch them. Use ML regime-detection models.