TSX Commodity ETF Rotation

While the top momentum ETF might have the highest return potential, it also carries concentration risk. If that ETF reverses, your entire portfolio suffers. Selecting 2-4 ETFs provides diversification - some protection if one pick underperforms. Momentum leadership can change quickly, and having multiple positions reduces the impact of being slightly late on rotations.

Because TSX ETFs trade in single shares with no futures margin or lot minimums, you can start much smaller than the MCX futures version. A comfortable diversified rotation across 3-4 ETFs runs well with roughly C$25,000-50,000, but a 2-3 ETF rotation is practical with as little as C$5,000-10,000. The main constraint is keeping bid-ask spreads and commissions small relative to your position sizes, so favour the more liquid ETFs when your account is smaller.

Monthly rotation reviews don't mean mandatory changes. If your current holdings still rank in the top 3-4, keep them. Rotation only occurs when rankings change significantly - an ETF drops out of the top tier or a new leader emerges. Forced rotation regardless of rankings just adds unnecessary turnover costs (spreads plus commissions).

If the entire commodity complex is falling, rotation becomes less effective as you're picking the 'best of the bad.' In such environments the trend filter (50-day EMA) will likely exclude most or all ETFs. This is when holding cash, or a broad diversified holding like COMX, is appropriate - wait until clear leaders emerge with uptrends before deploying capital.

HUN is a futures-based ETF, so its return reflects the underlying futures curve, not just the spot price. When the curve is in contango (later-dated futures more expensive than near-dated), the fund loses a little each time it rolls forward - this 'roll decay' can erode the ETF even when spot prices hold steady. Natural gas is especially prone to this. It's a structural feature of price-tracking commodity ETFs, and a reason to watch carry and to consider producer ETFs (like XEG) when curves are steeply in contango.

Calculate each ETF's volatility (20-day standard deviation of daily returns). Invert each volatility: 1/Vol. Sum all inverted values. Weight for each = (1/Its Vol) / (Sum of all 1/Vols). Example: if HUG vol=1%, HUZ vol=2%, HUC vol=3%, inverse vols are 100, 50, 33.3 (sum=183.3). Weights: HUG 54.5%, HUZ 27.3%, HUC 18.2%. Lower vol = higher weight.

First, calculate sector momentum (weighted average of constituent ETFs' returns: precious = HUG/HUZ/XGD, energy = HUC/HUN/XEG, base metals = COPP/XBM, materials-agri = XMA/COW). Rank the sectors. Allocate across sectors based on rank: top sector 40-50%, second 25-35%, third 15-25%, weakest 0-10%. Then within each allocated sector, select the top momentum ETF. This two-tier approach captures both sector and individual ETF momentum.

Target an average pairwise correlation below 0.5 among your selections. Cross-sector selections (e.g. HUG, HUC, COPP) naturally achieve this. Same-sector selections (HUG + HUZ) will have high correlation (0.8+). A Canadian wrinkle: two producer-equity ETFs (say XEG + COPP) share a common TSX equity-beta factor, so their correlation is higher than the underlying commodities would suggest - pairing a producer ETF with a price ETF generally diversifies better.

Overnight gaps are a reality here because TSX ETFs only price during the 9:30 AM - 4:00 PM ET session while the underlying commodities trade nearly around the clock globally. The ETF therefore gaps at the open to reflect overnight moves in oil, gold, etc. Individual stops set at 2x ATR typically accommodate normal gaps; for extreme gaps beyond stops, exit at the first opportunity rather than hoping for recovery. Consider smaller positions in the most gap-prone funds (HUN, HUC around inventory reports). Diversification across ETFs also helps - a gap in one won't destroy the whole portfolio.

It depends on what you want to capture. Price-tracking ETFs (HUG, HUC) follow the commodity itself but embed futures roll decay. Producer ETFs (XGD for gold, XEG for energy) follow the equities, which add operating leverage and equity beta - they can outperform in a strong commodity-plus-equity tape and underperform when the broad market sells off. Many traders compute momentum on the price ETF for the clean commodity signal, then choose the implementation (price vs producer) based on the macro regime and the futures-curve shape.

Calculate cross-sectional volatility (standard deviation of all ETFs' returns). Compare to the historical distribution. If in the top quartile (high vol), use a 15-day lookback and consider bi-weekly rebalancing. If in the bottom quartile (low vol), use a 30-day lookback. For the holding period, calculate the month-over-month autocorrelation of momentum rankings. High autocorrelation (>0.7) = extend holding; low (<0.3) = shorten. Implement these as rules with thresholds in your algorithm, using VIXC and the underlying commodities' implied vol as the regime gauge.

CFTC Commitment of Traders data on the underlying COMEX/NYMEX futures (which HUG, HUZ, HUC and HUN track) shows positioning of commercial hedgers vs speculators. Extreme speculator long positioning (>90th percentile historically) can precede reversals. Use COT as a filter/warning, not a primary signal. If an ETF has top momentum but its underlying future shows extreme speculator positioning, consider reducing its weight by 20-30%, setting tighter stops, or skipping it if other factors are also weak. COT is a contrarian indicator best used at extremes.

Build a regression of portfolio returns against factor returns: Portfolio = alpha + b1xMomentum_Factor + b2xTrend_Factor + b3xCarry_Factor + e. Momentum factor = return of the top momentum quintile minus the bottom quintile; similarly for other factors. Coefficients show factor exposures. Run rolling 12-month regressions to track how exposures change. Crucially for ETFs, add a 'wrapper' term capturing MER, bid-ask spread, embedded futures roll and USD/CAD translation, so you can separate signal quality from implementation drag. Attribution = Factor Return x Factor Exposure; alternatively use Brinson-style attribution decomposing into selection, allocation and interaction effects.

Create a macro regime indicator: Score = w1xEquityMomentum + w2x(-VIXC_level) + w3x(-DXY) + w4xPMI_momentum, and optionally a USD/CAD term for currency drag. Positive score = risk-on, negative = risk-off. In risk-on: increase the energy and base-metals sector caps (to 45-50%), reduce the precious-metals cap (to 20-25%), and lean toward producer ETFs when the TSX is trending. In risk-off: increase the precious-metals cap (to 45-50%), reduce industrial sectors (to 25%), and prefer CAD-hedged or price-tracking gold (HUG/CGL). Apply these caps to sector allocation before individual ETF selection. Review the macro regime weekly, sector caps monthly.

Key pitfalls: (1) Survivorship bias - include ETFs that were delisted, merged or renamed (several Horizons funds were rebranded Global X). (2) Look-ahead bias - ensure rankings use only data available at decision time. (3) Unrealistic execution - add bid-ask spread (wider for the smaller funds) and commissions. (4) Ignoring the MER and embedded futures roll - price-tracking funds quietly leak return; use total-return (distribution-adjusted) NAV series, not price-only. (5) Currency assumption errors - decide explicitly whether you model returns in CAD (the real experience) and account for USD/CAD on unhedged funds. (6) Over-optimization - limit parameters tested, require out-of-sample validation. (7) Short history and capacity - several of these ETFs are small, so verify liquidity for your size and use 5+ years covering different regimes.