Research/Education
EducationMonday, November 9, 2026

Position Sizing by Conviction: Sizing COT Extremes Right

Position sizing is the skill that separates survivors from blow-ups. Here's how to scale exposure to COT extremes like bonds (z +3.18) and copper (z -2.96) in an inflation-shock regime.

Everyone obsesses over entries. Pros obsess over *size*. You can be right on direction and still get carried out if you bet too big into the wrong regime. Today we'll walk through position sizing — the least glamorous, most important lever in quant trading — using RetailVest's live positioning data as the raw material.

Why sizing beats signal

A great signal sized wrong is a bad trade. The math is brutal: a 50% drawdown requires a 100% gain just to break even. That asymmetry is why professional desks anchor on *risk per trade* rather than dollar notional. The core idea: decide how much of your account you're willing to lose if you're wrong, then let that — plus volatility and conviction — dictate the size.

Right now the macro backdrop demands discipline. We're in an INFLATION_SHOCK regime: the VIX sits at 25.64, S&P 20-day momentum is -5.6%, and the 2s10s spread is a flat 0.38%. Per FRED, PPI (All Commodities) is running at 267.85 (+5.46) and the 10Y breakeven inflation is 2.34. Elevated vol means every position you take should be *smaller* than it would be in a calm tape — volatility is the denominator in any sane sizing formula.

Step 1: Size inversely to volatility

The simplest robust rule is volatility-targeting: the more a market moves, the fewer contracts you hold. With VIX at 25.64 — well above the low-20s baseline — a volatility-targeted book automatically trims notional. If you'd hold 10 units at VIX 15, you'd hold closer to 6 at today's reading. No discretion required; the regime does the de-risking for you.

Step 2: Scale to conviction, not just direction

This is where RetailVest's proprietary edge comes in. CFTC COT speculator positioning gives you a *conviction dial* via z-scores. A |z| >= 2 is statistically extreme — the crowd is stretched and prone to snapping back. The current board (CFTC COT):

  • **30-Year US Treasury Bond: z +3.18** — extreme long. Specs are maximally crowded into bonds.
  • **Copper: z -2.96** — extreme short.
  • **Palladium: z -2.61** — extreme short.
  • **Platinum: z -2.44** — extreme short.
  • Silver z -1.25 (bullish), WTI Crude z -1.25 (bullish), Gold z -0.77 (bearish).
  • The sizing logic: your *largest* positions should go toward the most statistically extreme, mean-reversion-friendly setups — but only after the volatility haircut from Step 1. Copper at z -2.96 and the PGMs (palladium z -2.61, platinum z -2.44) are three independent extreme shorts. That's a cluster worth a bigger risk budget than a middling signal like gold at z -0.77. Check each name on RetailVest's per-commodity COT pages before you size.

    Step 3: Don't double-count correlated bets

    Here's the trap beginners fall into: platinum and palladium are both PGMs driven by similar auto-demand dynamics. US Total Vehicle Sales (FRED) printed 16.508 (+0.11) — a modest demand tailwind for both. If you go long both against their extreme shorts, that's not two trades; it's one trade in 2x size. Treat correlated extremes as a *single* risk bucket when allocating. The 30Y bond extreme long (z +3.18) is your diversifier here — it moves on rates (10Y yield 4.84%, 2Y at 4.46 per FRED), not metals.

    Step 4: Let backtests calibrate, not seduce you

    RetailVest's Strategy Builder shows the gold_silver_ratio strategy at 1008.37% total backtested return and gold_200ma_trend at 623.88%. Monster numbers — but note silver_rsi_bounce is -10.0% over the past month. Backtested returns tell you a strategy's *edge*; they don't tell you how much to risk on the next trade. Size to the strategy's drawdown profile, not its headline return. A high-return strategy with deep historical drawdowns deserves a *smaller* per-trade risk allocation, not a bigger one.

    Putting it together

    In an inflation shock, your sizing checklist:

    1. Haircut everything for VIX 25.64 — smaller than normal.

    2. Concentrate risk budget on statistical extremes (copper z -2.96, PGMs).

    3. Collapse correlated bets (platinum + palladium) into one bucket.

    4. Keep the bond extreme long (z +3.18) as a rate-driven diversifier.

    The takeaway

    Pick one extreme — say copper's z -2.96 — and size it so that a stop at a sensible technical level costs you no more than 1% of your account at today's VIX of 25.64. Then halve any position in the correlated PGM bucket so you're not secretly betting 2x. Ask Tara, our AI analyst, to pull the per-commodity COT z-scores and current volatility before you commit a single contract. Conviction tells you *what*; position sizing tells you *how much* — and how much is what keeps you in the game.

    #position sizing#risk management#cot#copper#treasuries#inflation

    Market data for informational purposes only. Not financial advice. Past performance does not guarantee future results.

    How are we doing?