Actionable insights — AIA Free Weekly Email 7.23.26

1. Falsify a commodity thesis with a named physical signal, not with price

1. State the bullish (or bearish) commodity view together with the single observable event that would end it — a physical supply fact, not a price level and not a headline.
2. Pick the signal at the choke point of the thesis: if the case rests on barrels that cannot reach the market, the falsifier is the return of those specific barrels — named producers back at full export capacity, not partial, not announced.
3. Hold the position through official reassurance and through weak spot prices, because neither is the agreed test. Only the named event closes the trade.
4. Write the canary down once, so the exit is not renegotiated in the middle of a drawdown.

• Iraqi and Kuwaiti export terminals returning to full nameplate capacity (loadings and shipments, not statements); SPR and commercial crude draws continuing past the stated operational minimum; official reassurance that contradicts the physical inventory trend.

2. Audit a backtested timing tool before you let it move money

1. Separate what is defensible in the construction from what is proven in the record. A model that blends contrarian sentiment indicators with fundamental data (here mining margins) is structurally more defensible than one leaning on a single technical indicator — but that is a point about design, not about alpha.
2. Deliberately seek an independent critique instead of grading it yourself; Polomny outsourced the review ("I ran it through AI") and then published the negative verdict alongside the tool.
3. Refuse to convert a reported CAGR into an expectation until the evidence supports it. The correct default: "internally consistent and economically plausible backtest results — but not yet an independently verified source of alpha."
4. Write down, in advance, the exact evidence that would upgrade the verdict — then you know what you are waiting for rather than drifting into belief.

• The named verification bar: a timestamped monthly signal history, exact execution rules, a complete net (not gross) return series, rolling-period rather than single-window results, and several more years of genuinely out-of-sample performance. Short track records that begin at a convenient start date; volatility and Sharpe quoted without net-of-cost returns.

3. Judge a strategy against the right comparator — risk-adjusted, in one currency

1. Never accept a headline return alone. Demand the same window's volatility and Sharpe against both the underlying asset and the leveraged expression of it — for miners that means gold and GDX, not one or the other.
2. Check the currency the record is measured in (this one is gold in euro terms); a currency move can manufacture or erase most of an apparent edge.
3. Ask whether the outperformance came from the signal or from simply carrying less risk — a strategy running near the underlying's volatility while beating a far more volatile benchmark is claiming a genuinely different thing from one that just levered up.

• Returns quoted without volatility; a benchmark chosen to flatter (gold when the vehicle is miners, or vice versa); base-currency switches between the strategy and its comparator.

4. Price in the miner leverage before the drawdown, then read the drawdown as dislocation or verdict

1. Assume mining equities move roughly 2× the metal in both directions — the March move (gold −11.52%, GDX −20.78%) is the normal case, not an anomaly. Size the position to a drawdown you have pre-agreed to.
2. When it happens, ask the diagnostic question explicitly: is this a verdict on the companies or a dislocation? Answer it with balance sheets and margins, not with the chart.
3. If the businesses' fundamentals are intact while the price is not, the drawdown is an entry: hold cash beforehand so you can "act while others digest the loss," then rebuild cautiously and selectively rather than all at once.
4. Pair the fundamental read with a positioning read — overbought → oversold is the condition that turns a dislocation into a purchase.

• A metal drawdown amplified ~2× in the miners while sector margins and balance sheets stay intact; a swing from overbought to short-term oversold; dry powder raised before the fall rather than after it.

5. What a generalist does with a specialist's un-refutable claim

1. When a deep specialist presents a contrarian forecast you cannot evaluate on the merits, say so plainly instead of adopting or dismissing it — "I don't have the expertise to refute this."
2. Grade the work instead of the conclusion: how long, how bottom-up, how many links in the chain were checked (here 18 months across producing basins, pipelines, processing, LNG exports and AI power projects). Depth of process is the part a generalist can assess.
3. File it on the radar screen with a dated marker rather than trading it, and re-test as the forecast's near-term waypoints arrive (the 2028 storage-draw inflection long before the 2030 exhaustion claim).
4. Check whether it strengthens or contradicts the frame you already own; a claim that fits your durable thesis earns shelf space, it does not earn a position by itself.

• The 2028 waypoint — storage draws accelerating beyond seasonal norms; LNG export and AI-datacenter power demand landing on schedule; the specialist's own updates and whether an equally credentialed rebuttal appears.