Index Arbitrage Strategy

Theoretically yes, but in practice there are risks: execution risk (the TSX cash leg and the MX futures leg may not fill at expected prices), tracking error (a basket may not perfectly match the index, though XIU nearly eliminates this), dividend risk (especially important in Canada because dividends nearly offset carry), and margin risk (CDCC can call variation margin on the futures leg despite the hedge). With careful execution the risks are manageable but not zero.

For SXF calendar spreads (futures only): roughly C$15,000-25,000 of margin per spread - the most accessible route. For XIU-vs-futures using SXM minis (C$50/pt, ~C$100k notional/contract): about C$100,000+. Using standard SXF (C$200/pt, ~C$410k notional): about C$400,000+ to hedge one contract with XIU. For an optimized 20-stock basket: ~C$500,000+; for a full 60-stock basket: ~C$1M+ for meaningful positions. Minis and XIU lower the entry size considerably.

The basis reflects net cost of carry = financing rate minus dividend yield. Canada's 3-month financing rate (~2.3% in mid-2026) is close to the S&P/TSX 60 dividend yield (~2.5%), so the two roughly cancel. That leaves a small net carry that can be slightly negative, so SXF frequently trades at or just below the cash index (backwardation) - the opposite of high-rate markets where futures carry a clear premium.

Modest. Because the basis is thin, costs are low and the market is efficient, after-cost arbitrage returns are typically around 0.5-3% annualized in normal conditions - and they are anchored by a risk-free rate that is itself only ~2.3%. During dislocations (quarterly roll squeezes, dividend-season distortions, or stress events such as tariff/policy shocks) returns can be higher. The appeal is low risk and consistency, not large absolute returns.

Yes, via put-call parity on SXO (S&P/TSX 60 index options): a synthetic future = Call - Put + Spot should equal the SXF forward; if they differ beyond costs, an arbitrage exists. Options add complexity (Greeks, liquidity) and are generally used by more sophisticated traders. SXO liquidity is thinner than SXF, so spreads can be wider.

Track every constituent's ex-date and amount and recompute fair value daily as ex-dates approach - this matters more in Canada than in high-carry markets because dividends and carry nearly cancel. The Big Six banks (Oct fiscal year-end) cluster ex-dates around late Jan/Apr/Jul/Oct, and pipelines/telecoms/utilities add steady quarterly flow, while Shopify and most miners pay little. You can also hedge dividend-forecast risk with the SDV (S&P/TSX 60 Dividend index future).

Cash-futures pits the spot basket (or XIU) against SXF - a cross-venue trade (TSX cash + MX futures) with low cost (no STT) and near-zero tracking error via XIU. A calendar spread pits the near quarter against the far quarter SXF - both legs on MX, the simplest and cheapest route, with no tracking error, but a different (roll) risk profile. Because expiries are only quarterly, each calendar spread can span ~3 months and a full dividend season.

Select the top ~20 names by index weight and liquidity; together they cover roughly 70-75% of the index (the Big Six banks alone are ~30%). Capture the remaining weight by slightly overweighting the included names, targeting tracking error below ~0.3%. Monitor daily and rebalance if it drifts - and re-check weights around the semi-annual April/October index rebalances. In practice, many desks skip the basket entirely and use XIU because its spread is so tight.

Around the four quarterly ROLL windows (roll-related basis distortions are larger but rarer than in markets with weekly expiries), during DIVIDEND-season complexity, at the semi-annual REBALANCE (April/October flows), and during STRESS dislocations - Canada's smaller pool of arbitrage capital lets the basis stretch further in shocks (e.g. tariff/policy events). HFT captures the small intraday mispricings; focus on these structural, analysis-heavy opportunities.

The cash leg trades on TSX and the futures leg on MX, so coordinate them tightly: use basket/list orders for the cash side, route the futures simultaneously, and use an XIU-vs-SXF spread order where your platform supports it. Trade in liquid hours (avoid the open/close), use realistic limits, and accept a little slippage as a cost of business. Remember SXF settles to the official OPEN, so settlement-day open auctions matter for carried positions.

Use the CORRA-based OIS curve matched to each quarterly expiry (CDOR was discontinued in 2024). Model each constituent's ex-date and amount with present-value discounting - dividend error dominates fair-value error in Canada because dividends nearly equal carry. Treat fair value as a confidence band (propagating rate, dividend and execution uncertainty) and trade only when SXF is outside it. Backtest predicted vs realised convergence at each settlement and recalibrate around the April/October rebalances.

60-vs-Composite (SXF-vs-SCF) ratio mean reversion via z-scores; residual arbitrage on constituents (watch the bank bloc, which moves together around earnings and BoC decisions); index-rebalancing arbitrage around the April/October changes; and a Canada-specific dividend-basis trade using the SDV dividend future. These lack guaranteed convergence but offer higher expected returns than the thin pure-arbitrage basis - size them as statistical bets, not locks.

Do not race on speed. Canada's low costs compress everyday SXF-vs-XIU mispricing to a point or two, captured by co-located HFT in microseconds (cross-venue TSX-MX latency is itself their edge). Instead, trade the structural opportunities they cannot easily exploit: the four quarterly roll windows, dividend-season fair-value complexity, rebalance flows, and stress dislocations. These reward analysis and patience over latency.

Minimum: real-time TSX and MX data, an index-calculation and fair-value engine, a CORRA-OIS feed, a dividend/corporate-action database, broker/exchange APIs, and a position monitor reconciling across CDS (cash) and CDCC (futures). Advanced: low-latency execution and co-location at the TMX data centre, automated basket/spread generation, and a real-time risk/margin system. Build incrementally - Python plus a broker API first, then a full cross-venue stack - and keep audit-ready logs because CIRO performs next-day cross-asset surveillance.

Test: correlation breakdown (basket diverges from the index by 0.3%+); a dividend surprise (a bank cuts or pays a special, or your forecast is off by 5%+) - critical in Canada given the thin net carry; a one-leg execution failure across TSX and MX; a CDCC margin spike on a gap move; thin back-quarter SXF or SCF liquidity in stress; and an XIU/stock borrow recall on a short-side (backwardation) trade. For each, quantify the P&L impact and define a contingency. Re-run quarterly and after any system change.