Read the source · this Framework is our reading of their public work · Doomberg →
Synthos Research · Frameworks · Thesis Snapshot
Doomberg: how the voice actually thinks
This is not a profile. It is a working model of Doomberg's worldview — the causal
beliefs, the stated falsifiable gates, what the voice changed its mind about (dated), and where it is silent —
reconstructed from dated claims made between April 2025 and April 2026, and tested against claims the model never
saw. Doomberg is team-written and pseudonymous; we treat it as one coherent voice, exactly as it presents itself.
The record behind this Framework is the thinnest of the four published this round, and that is disclosed in every
section rather than papered over. Every quote on this page is a dated, verbatim fragment from our claim record.
11-node causal graph, 24 edges — the thinnest of the four published this round
durable beliefs
2
both fall just short of the 12-month invariant bar, kept anyway and flagged
recorded mind-changes
2
both undated for cause — disclosed, not hidden
direction fidelity
68.0%
13 of 19 out-of-sample tests
Model fidelity: 68.0% direction on 19 out-of-sample tests
What that number means: we froze this model on claims through April 17, 2026, asked it to
predict how Doomberg would react to real energy and geopolitical events from the following weeks, and graded
those predictions against the claims actually made in that window — claims the model had never seen.
13 of 19 gradeable predictions got the direction right. The pass bar, set before grading began, was
60%.
Zero invented mechanisms: the model never fired a causal link that had failed adversarial verification
against the record.
Correct silences: 71% on this eval round.
The thin-record caveat, stated plainly: with only 11 nodes, 24 edges and 2 confirmed multi-year
invariants behind it, this is the smallest causal graph of the four Frameworks published this round. Two genuinely
durable-looking beliefs (co-production economics and fade-the-spikes) fall a few weeks short of the 12-month bar
we require before calling something an invariant — we kept them, and we say so rather than quietly promoting
them.
What holds up despite the thin sample: the long-term bearish energy anchor survived a real stress test.
Mid-spike, during the 2026 Hormuz closure, the model correctly kept the structural glut/equilibrium thesis running
underneath the war-premium overlay, rather than treating the spike as a thesis change.
1 · How the world works, according to Doomberg
Only two beliefs in the record clear even a near-invariant bar — this voice's
claim history is short and dense rather than long and slow. Both are given in full, followed by the hub the
newest claims route through.
Co-production economics: expensive gas is structurally bearish for oil
Held April 2025 → March 2026 · ~11.5 months — a few weeks short of our 12-month bar,
kept and flagged rather than silently promoted
When gas drilling is profitable, the oil that comes up alongside it is a near-costless byproduct — producers
will sell that associated oil at almost any price, because the gas economics already paid for the well.
dirt-cheap co-produced US natural gas are all bearish for oil— 2025
high gas induces gas drilling that yields associated oil as a byproduct— 2025
Fade the spikes — never bet on or time them
Held April 2025 → February 2026 · ~10 months — reads as much like temperament as a
market view, and is disclosed as such
Geopolitical price spikes — "icarus prints" — inevitably fade. The discipline is to buy the put after the
spike has already happened, never to try to time or front-run the spike itself.
'icarus prints' inevitably fade— 2025
rather we wait to fade them— 2026
The hub: a persistent bearish energy anchor, with a war-premium overlay on top
Current through April 2026 — the frame that survived a live stress test
Underneath everything sits an equilibrium price for oil and gas — roughly $55 as an anchor, $70–75 as a
long-run range — built on the view that "peak resource" fears are a recurring myth: US hydrocarbon abundance
keeps reasserting itself. The 2026 Hormuz closure layered a genuine higher-for-longer war premium on top, but the
long-term glut thesis was explicitly maintained mid-spike, not abandoned. Two threads run on top of that same
anchor: Europe as structurally the biggest loser of any hydrocarbon-short world, and a long five-to-fifteen-year
reconstruction/pipeline-buildout opportunity on the other side of the current disruption.
top is likely already in— March 24, 2026
as long as Iran wants it closed it stays closed— 2026
2 · Its highest-conviction causal chains
Chains stated as mechanisms, not co-occurrences we inferred. With only 24 edges in
the whole record, these four chains carry most of the model's weight.
The co-production chain: cheap gas → byproduct oil → oil stays capped
Cheap, co-produced natural gas is structurally bearish for oil.
dirt-cheap co-produced US natural gas are all bearish for oil— 2025
Because the associated oil comes up almost free, as a byproduct of drilling for gas.
shale oil gets given away as a byproduct— 2025
So producers overproduce oil at nearly any price.
co-product producers overproduce oil at nearly any price— 2025
The AI-power chain: datacenter demand → grid bottlenecks → off-grid gas colocation
AI datacenter demand is projected to double US natural-gas demand within about a decade.
natural gas demand doubling within about 10 years— 2025
US natural gas is already getting a bid on that expected demand.
US natural gas getting a bid on AI-driven datacenter demand— 2025
Grid buildout can't keep pace, so datacenters colocate directly at the gas well.
bypassed by off-grid, made-to-order power plants— 2025
gas in and data out— 2025
The tactical spike chain: geopolitical shock → icarus print → the trade is the fade, not the spike
A geopolitical supply shock produces a price spike.
'icarus prints' inevitably fade— 2025
The discipline is to trade only after the spike, never to bet on its arrival.
rather we wait to fade them— 2026
And the bigger the spike, the bigger the eventual crash.
the bigger the spike the bigger the crash— 2026
The de-dollarization chain: a multipolar shift → sanctioned powers sell Treasuries, buy gold
The world is shifting from unipolar to multipolar.
sell US Treasuries and replace them with gold— 2026
Sanctioned and sanctions-wary powers are the ones making the swap.
oil is worth about $10 a barrel less— on a hypothetical Russia sanctions-relief deal, 2026
Worth knowing — the mid-crisis test it passed
On March 24, 2026, mid-Hormuz-crisis, the record shows the structural view holding even as the tactical spike ran
— top is likely already in — while a separate hormuz_closure → reconstruction_opportunity edge kept
a five-to-fifteen-year rebuild thesis running underneath the same disruption.
3 · What would change its mind — its own tests
Doomberg states gates as prices, dates and named actors rather than vague hedges.
These are its own conditionals, in its own words.
If
A US-Russia deal ends the war and lifts sanctions
A stated, priced downside test on oil — not a vague "geopolitics matters" hedge.
oil is worth about $10 a barrel less— 2026
If
Trump gates LNG exports under an emergency declaration
Henry Hub gas would tank — a named policy trigger with a named consequence.
if Trump gates LNG exports under emergency it would tank— 2026
As long as
Iran wants the Strait of Hormuz closed
It stays closed — described as near-impossible to reopen by force. The gate is Iranian intent, not military
capability.
as long as Iran wants it closed it stays closed— 2026
If
A geopolitical shortage spike has already occurred
The trade is a nine-month put after the fact — never a bet that the shortage arrives.
buy a 9-month put after a geopolitical spike— 2025
If
The US turns away Russian tankers bound for Cuba
A stated, specific precedent-setting test for whether China could embargo Taiwan without firing a shot.
embargo Taiwan without firing a shot— 2026
Worth knowing — it names its own low-confidence calls
The record self-flags speculative claims rather than presenting everything at equal confidence — on the Taiwan
scenario specifically, the honest label attached was mostly hope, not analysis.
4 · What it recently changed its mind about
We log mind-changes as a feature, not an embarrassment. Both recorded revisions in
this record are undated for cause — disclosed here exactly as they appear, rather than assigning a trigger we
cannot see.
February 12, 2026 — reversed course on Middle East military conflict
Old (January 25, 2026, conviction 55, explicitly hedged)Doesn't expect Middle East military action anytime soon; admits could be wrong
→
New (conviction 55)Some form of US military conflict with Iran is likely and would escalate quickly; markets aren't pricing it
Honest detail: no trigger is stated in the record for this reversal. The conviction level, 55,
did not change even as the direction flipped — the record keeps both calls rather than smoothing over the
reversal.
April 1, 2026 — Taiwan went from "not his base case" to "almost a certainty" in one week
Old (March 24, 2026, conviction 30)China-embargoes-Taiwan scenario is speculative; China could set a precedent to embargo Taiwan without firing a shot
→
New (conviction 85)forcing a reunification-tantamount political arrangement within 5 years
Honest detail: conviction jumped from 30 to 85 in a single week with no trigger stated in
the record. We flag the magnitude and speed of the jump rather than treating a one-week, 55-point conviction swing
as ordinary.
5 · Where it is silent
Two kinds of silence, labeled: mechanisms actively argued against on the record, and
mechanisms that simply never appear in this thinner claim history. With only 11 nodes total, this list is short by
construction, not by omission.
Actively rejected — it has said no, on the record
Fed rate decisions driven by inflation or employment data. The record asserts the opposite: Fed moves
are politically and electorally timed, purely on the electoral calendar, and are not justified by the macro data.
Renewables substituting for oil-demand growth. Explicitly denied.
A conventional carry-trade / rate-differential framework for the dollar. The dollar framework running
throughout the record is de-dollarization and gold-anchored multipolarity instead.
Never asserted — the whole record, zero mechanism
Broad equity index valuation tied to Fed liquidity or rate policy. No mechanism connecting index
multiples to liquidity or rates appears anywhere in the record.
Labor-market data (jobs, unemployment) as a forecast driver. No linkage in any claim.
Routine central-bank QE/QT as a commodity-price driver. The "money printer" mechanism that does appear
is tied specifically to war-related force-majeure events and insurer failures, never to routine balance-sheet
policy.
OPEC+ supply-cut decisions as the controlling factor on oil price. The supply narrative in this record
runs entirely through non-OPEC, Western-Hemisphere, shale-and-gas economics instead.
ESG or carbon-policy mechanisms (stranded assets, carbon pricing). Absent from every commodity and
energy thesis in the record.
Why this matters: ask this Framework about Fed policy driven by jobs data or an
OPEC-controlled oil price, and the honest answer is "this model is silent here." Refusal with receipts beats
confident invention — and on a record this thin, that discipline matters more, not less.