Actionable insights — Another Golden Opportunity?

1. Name an asset's official drivers first — then treat a price that ignores them as evidence

1. Before forming a view, write down the two or three variables the asset is conventionally supposed to follow. Committing to them in advance is what makes the test honest — otherwise you rationalise whichever driver happens to fit afterwards.
2. Check the drivers' recent direction. You want a stretch where they have moved decisively against the asset — not ambiguously, but hard enough that the textbook answer is unambiguous.
3. Then look at the price. If it went where the drivers said, you have learned nothing. If it refused — held a level, went sideways, quietly firmed — you have found a discrepancy that has to be explained by someone on the other side of the tape.
4. Interpret the refusal as accumulation, not as strength. This is the Kovner reading: a consensus the market will not confirm means a lot of people are about to be wrong. It says nothing yet about timing.
5. Require a trigger before acting. The refusal identifies the setup; the entry is the break of an observable technical level — a moving average, a trendline — because that is what forces the offside crowd to buy back.
6. Expect the move to be violent and to have no fundamental news attached, since its fuel is short-covering rather than a change in the drivers. Do not wait for the drivers to turn before believing it.

• The floor giving way — the whole read is invalidated by a decisive close below the level that was refusing to break, which is also why that level makes a clean, pre-defined stop. Watch too for the drivers reversing after you are in: a falling dollar and falling real rates turn a contrarian trade into a consensus one, which is when you are being paid to think about trimming, not adding.

2. Disqualify a structural bid that has been present through the decline

1. List the "everyone knows" bullish supports for the asset — the ones cited in every article about it (central-bank buying, ETF flows, a supply deficit, an official mandate).
2. For each, ask one question: was it present during the fall? A support that was already in place while the price declined has been empirically shown to be insufficient, whatever its size.
3. Downgrade those supports from reasons to buy to reasons the downside was contained. That is a real and useful function — it shapes position sizing and stop placement — but it is not a catalyst.
4. Insist that the actual trigger come from a variable that has changed: positioning flushed, a price level defended, an option market repriced, an earnings estimate bottoming.
5. Apply the same scepticism to the data on the standing bid, especially when the buyer has an interest in concealing it. Treat official figures as a floor, not a measurement — and do not build a thesis on a number the reporter is motivated to understate.

• Reported official-sector purchases accelerating and the price responding — the combination, not either alone, is the confirmation. Conversely, a rally the standing buyer is likely to walk away from (per insight 3) can be self-limiting.

3. Model the strategic buyer's incentives, not just its size

1. When a market has a large non-financial participant (a state, a central bank, a strategic stockpiler), do not assume it behaves like a fund. Ask what it is optimising for and over what horizon.
2. A buyer accumulating a target quantity over a decade wants a low average price and is therefore an anti-momentum force: it steps back into strength and steps forward into weakness.
3. Infer the behavioural rule from that: the strategic buyer will not compete with a speculative mania and will not let speculators front-run it, because doing so raises its own cost and hands the profit to the crowd.
4. Invert the usual sentiment read. Widespread western enthusiasm is bearish in such a market, because it removes the biggest structural buyer; broad despair is bullish, because it brings that buyer back.
5. Use this to time alongside the strategic buyer rather than ahead of it — buy where its incentives say it is active, which is exactly where sentiment is worst.

• The mirror-image risk: if the price "runs hard, they would likely walk away." A sharp rally removes the very support that justified the entry, so the strategic-buyer leg of the thesis weakens precisely as the trade works — plan to be paid by something else (positioning, earnings revisions) by then.

4. Measure sentiment with something people have to pay for

1. Prefer sentiment gauges backed by money at risk over surveys and commentary. Options pricing, positioning data and fund flows cost their participants something to be wrong about; opinions do not.
2. Learn the asset's structural normal before reading any extreme. Equity indexes pay up for puts (crash risk is down); gold pays up for calls, because the perceived tail is up. Comparing across assets without that adjustment is meaningless.
3. Pick a stable, quotable construct and stick with it — here the 1-yr 25-delta call skew — and establish its ordinary range and its historic extreme from a long chart.
4. Read a collapse in the premium paid for the upside tail as apathy in its purest form: not people betting against the asset, but nobody willing to pay anything for the possibility that it rips.
5. Treat that as a cheap-optionality condition as well as a sentiment one — when nobody wants right-tail exposure, the leveraged expression of the trade is unusually inexpensive.
6. Cross-check against attention: what the crowd is looking at instead. A crowded rival theme is corroboration that the neglect is real and not merely seasonal.

• The skew normalising back toward its 3-4 vol-point average — that is the sentiment leg being paid out, and it happens well before the crowd's attention returns. Watch also for the opposite extreme (10 vol points) as the marker that the mania has re-formed and the position is now the consensus.

5. Only take a view on what is on its way to the front page — and refuse to have one on manias

1. Grade every candidate by where it sits in the attention cycle, not by whether you think it is cheap. The target is the asset on "page 17 on its way to page 1"; the anti-target is the one already on page 1.
2. When something is on page 1 every day, accept that you have no edge and say so. "I didn't have any value to add" is a complete and disciplined answer, and it protects the credibility you will need when you do have one.
3. Specifically, do not short a mania on valuation. "It was a mania, and it would end when it ended, and not a moment sooner" — the absence of a timing mechanism is the reason to stay out on both sides.
4. Keep the discarded candidate on a watchlist and re-examine it once the attention has moved on. The interesting moment is the same asset, opposite coverage.
5. Use the crowd's current object as your clock: when the enthusiasm has clearly relocated to something else, your old page-1 asset is now on page 17.

• Your own asset returning to page 1 — magazine covers, retail inflows, the skew back at 10 vol points. That is the signal to become a seller, and by this method it arrives before any deterioration in the fundamentals.

6. Defuse the cheap-cyclical trap by asking whether the earnings have already fallen

1. Whenever a cyclical screens cheap on P/E, state the standard objection out loud before anyone else does: the multiple is low because the E is about to collapse. A cheap cyclical at the top of its cycle is the classic value trap.
2. Then test it empirically rather than dismissing it. Ask whether the commodity has already fallen and whether reported earnings have already followed it down. If both are true, the reset is behind you, not ahead.
3. Check where the analysts' commodity deck sits versus spot. Sell-side models are famously slow to move their price assumptions, so a lagging deck means published estimates are stale in your favour once the commodity stabilises.
4. Note the asymmetry that creates: the asset does not have to rally for estimates to rise. Mere stabilisation at the current level lifts forecasts, which bottoms EPS, which is what makes the multiple real.
5. Frame the payoff as an estimate-revision event, since that is the mechanism that re-rates a stock, and put the historic-cheapness figure in context by naming the only prior episodes that were cheaper.
6. Size for the possibility you are early: a bottoming process has a duration, and this method gives you a valuation floor, not a date.

• Forward-EPS revision breadth turning up while the multiple stays put — the cleanest confirmation the reset is complete. The falsifier is equally clean: estimates continuing to fall with gold flat, which would mean cost inflation, not price, is eating the margin.

7. Attach an exit rule to a contrarian entry — and notice when the guest's rule isn't yours

1. Every contrarian entry needs a stated answer to "what if the refusal-to-fall was just a pause?" Decide it before the position exists, because the same argument that justifies buying weakness will justify buying more of it.
2. Pick one of the two coherent rules and commit: average down into a valuation thesis with a defined range, or cut on the invalidation of a price-action thesis. Never switch rules mid-position — that is how a trade becomes an investment by accident.
3. Match the rule to the evidence you actually entered on. A trade triggered by a technical break and a short-covering run is a price-action thesis, so it is falsified by price and belongs under "losers average losers."
4. When you republish someone else's research, read their rule as carefully as their argument, and note where it differs from your own house discipline. Two people can share a thesis and still need opposite behaviour when it goes against them.
5. When endorsing outside work publicly, add the caveat you would want a reader to have — most importantly what has already moved since the piece was written — and mark your own related position honestly, gain and entry price both.
6. Keep the conditional in the instruction. "Any pull-back should render it an accumulation candidate" is a different order from "buy it," and the difference is the whole risk-management content of the note.

• Whether the next POW! portfolio tables convert this endorsement into a rating or a new position — that is where a broadcast thesis either becomes an action or stays commentary. Watch too for the gap between the guest's entry price and yours: republished research is always at least a few sessions stale, and here the caveat says so explicitly.