When official data overrules the rule

There is a geopolitical analysis error that is very easy to make and nearly impossible to spot from the inside: being right about the event and wrong about the market.

A concrete example. Tension builds around the Strait of Hormuz — a third of the world's seaborne oil passes through it. The reasoning writes itself: supply risk, therefore a risk premium, therefore oil up. The reasoning is sound. And it can be entirely wrong, for a reason that has nothing to do with geopolitics: if US commercial inventories have just risen sharply, the market has a buffer, it knows it, and it does not pay the premium.

So we gave three public datasets the right to cancel our own rules. Not to weight them: to cancel them.

The hierarchy, and why it runs this way

A rule is a hypothesis: "this type of event has this effect on this asset". A fundamental datum is a fact: inventories on 17 July stood at this level. When the two contradict each other, letting the hypothesis win would be an odd choice — yet that is the default behaviour of any rules engine that ignores fundamentals.

Here, the contradiction triggers a veto and the signal is not emitted.

1. EIA — energy inventories

The US Energy Information Administration publishes crude and natural gas inventories weekly. We hold 63 weekly readings, from 9 May 2025 to 17 July 2026. A bullish WTI signal is checked against the most recent change: if inventories have clearly rebuilt, the scarcity thesis does not hold that week, whatever the headlines say.

It is the most intuitive veto, and the easiest to audit afterwards: the datum is public, dated, and nobody can rewrite it.

2. WASDE — agricultural balance sheets

The USDA's monthly report gives ending stocks for wheat and corn. Same logic: a bullish wheat signal driven by a conflict is cancelled if the balance sheet shows comfortable stocks.

Let us be honest about this veto's maturity: we currently hold a single WASDE reading (July 2026). The mechanism is wired and tested, but it has no history yet. The next report, around 12 August, will be the second data point. A veto with one data point is a veto that has demonstrated nothing — we would rather write that than imply a depth we do not have.

3. CFTC — who is already positioned

The third is of a different nature. Every week the CFTC publishes speculators' net positions on futures contracts. We hold 14 series with roughly 110 readings each, from 18 June 2024 to 21 July 2026 — a little over two years, enough to place the current position in its percentile over 104 weeks.

What this measures is not supply or demand but crowding. If bullish speculative positioning on Bitcoin sits at the 98th percentile of two years, the question is no longer "is the news good?" but "who is left to buy?". A trade everyone has already taken carries an asymmetric risk profile: the favourable news is in the price, the exit is not.

That veto blocks nothing, for now

This is the point we want to stress, because it matters more than the three mechanisms combined.

The positioning veto runs in observation mode. It detects, it writes, it does not block. Its firings carry an [observation, non-blocking] marker and are published as-is among our abstentions:

[observation, non-blocking] Extreme speculative positioning (COT BTC net +3500, P98 over 104 weeks) — crowded trade on the up side

The reason is simple: we do not yet know whether it is right. A gate that blocks signals must first demonstrate that it blocks the right ones. In observation mode we measure what it would have prevented and what those signals actually returned; it will only switch to blocking if that measurement justifies it, and the switch is automatic, nightly, outcome-driven.

The "crowded trade" reasoning is appealing. It is widespread in the literature. None of that makes it true on our assets, at our horizon, with our rules. We have already seen two hypotheses of this kind — solid on paper, tested with a pre-registered plan — abandoned after measurement. A filter added "because it seems prudent" degrades an engine just as surely as a missing filter, except nobody notices.

What this hierarchy taught us

Three things, two of them uncomfortable:

1. Sound geopolitical reasoning is not enough. The event can be correctly read, correctly scored, and the market can fail to move — because a material fact, published and verifiable, says otherwise. That is a limit of the method, not a bug.
2. Official data has a quality our rules will never have: it is dated by its issuer. An EIA report from 17 July is from 17 July, full stop. That makes it a defensible arbiter, including when it contradicts us — especially then.
3. Maturity must be stated. Two years of COT, one year of EIA, a single WASDE reading: these three vetoes do not carry the same weight, and presenting them as equals would sell a depth we do not have.

Where to check

The vetoes that block, those that observe, and each one's quantified reason are published on the Truth Table, under "What we refused to trade". Every source is public and free: you can redo the calculation.

---

GeoPulse observes geopolitical events and measures what they do to markets. We do not promise prediction: we publish our decisions, our abstentions and our mistakes, timestamped and sealed. Create a free account.