Pair Trading Strategy

Random stock pairs lack the statistical relationship needed for mean reversion. You need stocks that are fundamentally related (same sector, similar business) and statistically proven to have a cointegrated relationship. Without this, the spread can diverge permanently and never revert, causing losses.

For cash/stock-based pairs, roughly CAD $50,000-75,000 lets you trade 2-3 pairs with proper position sizing - and you need a margin account to short the second leg. Using options can lower the cash requirement through leverage. The key is having enough to diversify across multiple pairs while keeping each pair's risk near 2% of capital. Note: you cannot short inside a registered account such as a TFSA or RRSP, and the CRA may reassess frequent trading in a TFSA as business income, so an active long/short pair book generally belongs in a non-registered margin account.

This is the main risk in pair trading. You have a stop loss at Z = 3.5 - if the spread widens beyond this, exit the trade and accept the loss. Also investigate why the spread diverged - it could be a structural break (M&A news, a spinoff, a dividend cut, an asset write-down) that means the pair is no longer valid.

Yes - and in Canada most retail pair trading is done with shares, since single-stock futures aren't available here. Shorting the second leg requires a margin account and the shares being available to borrow (most Big Six banks, railways, pipelines and large producers are easy to borrow). Alternatives for the short leg are equity options on the Montreal Exchange or shorting the interlisted US listing (which adds currency risk). Start with liquid, easy-to-borrow names.

Daily monitoring of Z-score and correlation is recommended. Weekly deep review of cointegration status. Set alerts for Z-score approaching exit targets (0.5) or stop levels (3.5). Pair trades are not set-and-forget - active monitoring is essential, and on the short leg keep an eye on borrow availability/cost.

Avoid new entries in the few days before either name reports. A handy feature of the Canadian market: the Big Six banks all report within roughly the same one-to-two-week window each quarter (their fiscal year ends October 31, so results land in late February, late May, late August and early December) - plan bank-pair entries and exits around it. For existing positions, consider reducing size or tightening stops. Results can cause temporary divergence that reverts, but can also trigger stops; some traders close pairs over peak reporting weeks.

Stocks: no expiry, hold indefinitely, but shorting needs a margin account and borrow. Index futures (SXF on the Montreal Exchange): leverage and easy index exposure, but quarterly expiry/rolling and only at the index (not single-stock) level. Options: defined risk and leverage, 100 shares per contract on the MX, but theta decay and Greeks. Remember single-stock futures are not available in Canada, so for an individual-name short leg the choice is shares (margin), options, or the interlisted US listing.

Investigate the cause: stock-specific news (a dividend cut, a failed acquisition, a major mine or asset write-down, a management change) may be permanent for that name. Market-wide fear causing divergence is usually temporary. If correlation drops below 0.60 for more than two weeks without an obvious temporary cause, assume the relationship may have broken and exit.

Ratio spread is generally better - it's normalized for different price levels and more stable over time. Price spread works for similar-priced stocks. Beta-adjusted spread is most accurate for proper hedging. Start with ratio spread, graduate to beta-adjusted for precision.

Optimal is 5-8 pairs for diversification without over-complexity. Too few (1-2) = concentrated risk. Too many (>10) = difficult to monitor and manage, with diminishing marginal benefit. Ensure sector diversification - a maximum of 2 pairs from the same sector.

Generate all sector pairs from liquid stocks. Filter sequentially: correlation > 0.70, cointegration p < 0.10, half-life 3-25 days, stable relationship. Rank remaining pairs by a composite score (cointegration strength, half-life, stability, liquidity). Walk-forward validate. Rerun quarterly.

For pair selection: XGBoost/Random Forest classification (profitable vs not). For entry timing: gradient boosting with Z-score features + momentum + regime. For exit: reinforcement learning or supervised regression. Ensemble ML with traditional rules - use ML to filter, not replace, fundamental logic.

Calculate volatility (std dev of daily P&L) for each pair. Weight = 1/Volatility, normalized to sum to 100%. Lower-volatility pairs (banks) get more capital, higher-volatility ones (energy, miners) get less. Rebalance monthly. This equalizes the risk contribution across pairs.

Track: (1) average correlation among S&P/TSX Composite constituents (high = trending), (2) VIXC and the US VIX (>25 = elevated risk), (3) market breadth (extreme = trending, mixed = mean-reverting). Combine into a regime score. Adapt allocation and thresholds based on the regime.

Execute the less liquid leg first. Use algorithmic execution (TWAP/VWAP) for large sizes. Place limit orders to capture the bid-ask. Track slippage and execution cost - it should be <20% of expected profit. For very large positions, consider spreading execution over multiple days or using options.