Actionable insights — The Second Mouse Gets the Cheese

1. "The Second Mouse Gets the Cheese" — a re-entry checklist for a trade that already killed its first crowd

1. Start from a trade that blew up recently — one where a large speculative long (or short) was stopped out in a violent move. That is the candidate, not a disqualifier.
2. Test the three conditions, all of which must hold: (a) the fundamental picture is unchanged or improved since the blow-up; (b) positioning is materially cleaner than it was going in; (c) the prior participants are gone or shrunken — the survivors are running "a fraction of their position sizes."
3. Treat fear as the fuel, not the veto: "the positioning is cleaner and investors are more scared the second time around, which means the trade works in a classic wall of worry" — nobody is left to sell, and few can hold it "for the meat of a move."
4. Size it knowing you will be uncomfortable ("do the hard trade") and buy time — staggered expiries rather than one dated bet — because a wall-of-worry move is slow to start.

• A market that just carried out its own bulls; peers who "won't touch it"; a thesis you can restate unchanged with a straight face. If the fundamentals genuinely broke, it is not a second mouse — it is a falling knife.

2. Score positioning against price level, not against its own history — in dollars, not contracts

1. Pull the weekly Commitment of Traders managed-money / non-commercial net position and convert it to notional US$ (contracts × contract size × flat price). Contract counts lie when the price has moved 30%.
2. Ask the level question: "what price was this much length consistent with historically?" If today's net long matches what the market carried when price was 20% lower, positioning has not caught up to price — the move is unowned.
3. Do the same for the short side: a short that is "-30% by contract count" can be unchanged in dollars if flat price rallied the same amount. The short has not actually been cut.
4. Read the result through Kovner: a consensus the market is not confirming — a lot of people positioned to be wrong.

• The catch-up: fresh length arriving (net long rising toward the level the price implies) is the trade working and, eventually, the exit signal. Positioning that stays light while price grinds higher keeps the wall of worry intact.

3. Normalize by open interest — and use OI to tell short-covering apart from fresh money

1. Chart open interest alongside price. Collapsing OI means the market has shrunk — less commercial activity, thinner liquidity, more convexity per dollar of new flow.
2. Compute net long as a % of open interest, not just the absolute number, so the reading is comparable across eras of different market size. Then ask again what price that percentage historically implied.
3. Use the mechanics to identify who is transacting: buying either takes an existing long's exit (OI flat) or draws a fresh short (OI rises). A rally on falling or flat OI = underwater longs selling to shorts who are covering — no new money.

• OI expanding on a rally — the signature of new longs entering, i.e. the crowd finally returning. Falling OI into higher prices keeps the setup convex but also means thin, gappy markets in both directions.

4. Let the physical market adjudicate the narrative — prompt price, regional swaps, and the time-spread curve

1. Track the prompt physical benchmark (Dated Brent for immediate delivery) separately from the futures flat price. When physical leads the paper market down, an unseen physical buyer has stepped away — that is the tell to find and name.
2. Cross-check with a regional differential (Dubai 2m swap over Brent) to test whether an alleged supply-side story (e.g. blocked or "dark" transits) is actually binding, versus a demand story elsewhere.
3. Read the time-spread curve — 2-3m, 2-5m and 1yr — and anchor its level to a historical episode: "what was flat price doing the last time spreads were this backwardated?" Steep backwardation is the market "screaming for crude to come to market," i.e. inventories short.
4. Trade the gap: if spreads and physical say one price and the flat price/positioning say a much lower one, the flat price is the thing that is wrong.

• Backwardation flattening or flipping to contango (the shortage resolving) — the signal to stand down. Prompt physical rolling over ahead of futures again = a new demand disappearance, as in 2Q.

5. The "Jaws of Death" — cracks lead, crude follows: trade the second derivative's lead over the flat price

1. Chart refining margins (crack spreads) against crude flat price. Cracks are downstream demand for the product; crude is the input. When the two diverge into open "jaws," the gap is the trade.
2. Assume the direction of resolution runs from products to crude — "cracks lead, crude inevitably follows" — because refiners earning outsized margins pull barrels through: "as long as refiners are earning a $50+ crack, they will run every last barrel as hard as they can, stressing equipment and skipping maintenance if they have to."
3. Check the inventory backdrop that sustains the crack (global + US product stocks). Brutally tight product inventories mean the margin is not a one-week refinery-outage artifact.
4. Look for the precedent close: the last time the jaws shut, did crude merely catch up, or catch up and keep going? Size for the latter.

• Cracks compressing without crude rising (demand destruction, the bearish resolution) versus crude rising into a still-elevated crack (the bullish one). Refinery run rates and maintenance deferrals confirm the pull-through.

6. Keep an honest scorecard — name the wins and the mechanism of each loss

1. When re-entering a trade that hurt you, publish the full record first: which calls worked, which did not, and what specifically you got wrong — not "I was early."
2. Separate thesis error from expression error. A right view lost through "crappy options and cash trades" is an instrument/timing problem; a missed demand shift is an analytical problem. They demand different fixes.
3. Name the variable you under-weighted and re-rank it in the model for the next iteration (here: how far positioning can reset, and a single buyer's ability to vanish).
4. Use the survivors' behaviour as data: if peers who traded it are gone or de-sized, that is the positioning input for insight #1 — your own drawdown is evidence the mouse trap did its work.

• A commentator re-upping a losing trade without the post-mortem — the tell that the model has not been updated. And note the delta/vol-adjusted framing: "largest position" in an options book means risk, not dollars deployed.