SYNTHOS RESEARCH

Synthos Research

Frameworks — how our tracked voices actually think

A Framework is a working model of one person's worldview. From every dated claim a tracked voice has made across years of talks, interviews and writing, we reconstruct how they actually think: which beliefs are load-bearing, which causal mechanisms they genuinely assert, what they say would change their mind, what they have already changed their mind about (dated), and where they are simply silent. Every belief is backed by dated, verbatim quotes from our claim record — nothing is inferred from reputation, and a Framework may only believe what the claim record can prove.

Then we test it. Before a Framework is published, we freeze it and ask it to predict how the person would react to real events — and grade those predictions against what the person actually said next, using claims the model never saw. A Framework that fails that test is not published. The scores, including the misses, the partial credits and the small samples, are printed on every page.

educational only, not investment advice · how voices earn tracking · methodology

The synthesis

The Synthos Framework · v0

Where markets are, and where they're heading

The whole council, read against live hard data — a probabilistic view at short, medium and long horizons

7 frameworksfidelity-weighted into one house view
Causal chainsevery link a dated verbatim receipt
Pre-registeredhorizon odds, gradeable as events resolve

Today's market reads as one specific configuration the council can name — an oil-shock liquidity event. Four causal chains, a mechanism ledger, and a probabilistic read of the next six months to the multi-year destination, with the falsifiers that would change it.

Open the Synthos Framework →

Live

Live · eval passed

Jordi Visser

Macro · AI · Bitcoin — one of the highest-volume voices in our claim record

87%direction fidelity, 39 out-of-sample tests
Dec 2, 2025worldview pivot, dated & documented
Since 2021deepest belief: AI is structurally deflationary

1,730 dated claims, 2020→2026. One full worldview pivot with the old beliefs' death dates on the record — AI operating leverage lifts equities and pushes wealth into Bitcoin — plus his own stated falsifiers, and the places he is silent.

Open the Visser Framework →
Live · eval passed

Bill Gurley

Venture · marketplaces · AI market structure — the longest span in our claim record

13.2 yrshis two oldest beliefs, held without reversal
90.9%direction fidelity — small sample: 11 tests, disclosed
19 / 19correct silences on Fed, inflation & war

176 dated claims in his own voice, 2012→2026 (guest voices excluded). The bubble template with a rotating substrate — There is an AI bubble and essentially everyone is in it — a robotaxi timeline that collapsed from 25 years to one on a trigger he named, and a dormant rates belief the model flags as stale rather than fires.

Open the Gurley Framework →
Live · eval passed

Andreas Steno

Tactical macro · liquidity plumbing · positioning — the fastest-rotating Framework in our record

73.5%strict direction fidelity, 34 out-of-sample tests
6 flipson the dollar while his structural spine never moved
116 / 116test claims verified as his own voice — a first

1,041 dated claims in his own voice, 2023→2026 (co-host and guest voices excluded). A no-era operating system — watch Yellen's issuance, not Powell, for liquidity — quantified transmission lags that survived adversarial audit, a falsifier that fired on the date he named, and a recession retreat that became a new belief.

Open the Steno Framework →
Live · eval passed

Luke Gromen

Macro · fiscal dominance · gold and Treasuries — the longest span of the four newest Frameworks

65.7%direction fidelity, 35 out-of-sample tests
Since 2018oldest belief: the dollar as the sacrificial release valve
100%correct silences this eval round

A fiscal-dominance spine held since 2018 — MMT wins because there is no other politically viable option — the oil-breaks-the-Treasury-market chain, his own priced thresholds, and a Bitcoin call that flipped bearish→bullish twice inside 30 days while the gold thesis under it never moved.

Open the Gromen Framework →
Live · eval passed

Lyn Alden

Macro · fiscal dominance · energy and Bitcoin — the largest out-of-sample test run on any Framework so far

63.0%direction fidelity, 46 out-of-sample tests
10durable, principle-horizon beliefs, never reversed
2026conceded the halving cycle no longer matters

Her fiscal-dominance framework read in rate-of-change terms — Nothing stops this train — explicit if-else conditionals in place of price targets, and a record that updates on observed data (the failed Bitcoin "barbell") rather than on expectation.

Open the Alden Framework →
Live · eval passed

Mike Green

Market structure · passive flows · index mechanics — published with one gap disclosed, not hidden

62.5%direction fidelity, 16 out-of-sample tests
7durable beliefs, unchanged since 2019
1 disclosed gapthe dealer-hedging mechanism is not published as grounded

The passive-flow framework that concentrates markets into mega-caps — liquidity doesn't scale with market cap — his own falsifiable thresholds (roughly 83% passive share), and an explicit disclosure: his most famous mechanism, dealer hedging, did not survive adversarial verification and is not published as grounded.

Open the Green Framework →
Live · eval passed

Doomberg

Energy · geopolitics · commodities — team-written, pseudonymous, the thinnest corpus published this round

68.0%direction fidelity, 19 out-of-sample tests
11 nodesthe smallest causal graph of the four newest Frameworks, disclosed as such
Held mid-crisisthe bearish energy anchor survived a live Hormuz spike

Team-written and pseudonymous, treated as one voice. A co-production-economics spine — dirt-cheap co-produced US natural gas are all bearish for oil — and a fade-the-spikes discipline, tested against a real war and disclosed as the thinnest record we have published.

Open the Doomberg Framework →
Live · eval passed

Brent Johnson

Macro · dollar mechanics · Santiago Capital — the council's structural dollar-bull dissent

75.0%direction fidelity — 16 gradeable tests from a thin 29-claim holdout
2026stablecoins become a second engine of the milkshake, re-dollarizing from below
0invented mechanisms fired across the whole eval round

The dollar milkshake spine — a rising DXY forces dedollarization by breaking access — a structural dollar-bull case against a council of mostly weak-dollar voices, and a disclosed gap: the model has no node for his actual Hormuz-shock trade.

Open the Johnson Framework →
Live · eval passed

Eurodollar University (Jeff Snider)

Eurodollar plumbing · deflation-first macro — the council's deepest dissent from the inflation consensus

90.9%direction fidelity — small sample: 11 tests, disclosed; the strongest first eval of any macro Framework
100%correct silences — no invented answers on equities, AI or labor
Monthsthe youngest corpus we publish, disclosed as such

The dollar rises on shortage, not strength — An energy shock becoming a dollar shock is squeezing the global economy — energy spikes are a deflationary tax, rate moves are reaction not stimulus, and the private-credit bust is a staged recognition cascade.

Open the Eurodollar University Framework →
Live · eval passed

Rhonda Patrick

Tactical macro · liquidity plumbing · positioning — the fastest-rotating Framework in our record

71%direction fidelity, 31 out-of-sample health-reasoning tests
8 durable beliefsdose-response gates that held across her record, receipt-verified
116 / 116test claims verified as his own voice — a first

1,041 dated claims in his own voice, 2023→2026 (co-host and guest voices excluded). A no-era operating system — watch Yellen's issuance, not Powell, for liquidity — quantified transmission lags that survived adversarial audit, a falsifier that fired on the date he named, and a recession retreat that became a new belief.

Open the Patrick Framework →
Live · eval passed

Niall Ferguson · geniuses lane

Financial history · geopolitics — the first Framework built from a thinker, not a markets voice

77.8%direction fidelity — small sample: 9 tests, disclosed
18 yrsclaim record 2008 → 2026, the longest span we publish
Since 2011Ferguson's Law on the record — and fired near-verbatim on 2026 fiscal news

The fiscal-decline law — when interest payments exceed defense spending a great power declines — the Cold War II chain from Ukraine to Taiwan, six dated reversals including flipping on China's AI capability, and one false silence the eval caught: his AI-optimist register, disclosed as the model's gap.

Open the Ferguson Framework →
Live · eval passed

Joscha Bach · geniuses lane

Minds, machines and AGI — a Framework with no markets machinery at all, graded on AI events

85.7%direction fidelity — small sample: 7 tests on AI-capability events, disclosed
100%correct silences — refused to invent a single markets view
Since 2013mind-is-computation axioms, 13 years unbroken

The master axiom — computation is necessary and sufficient to produce a mind — the AGI inevitability chain (once AGI is buildable it will be built), coexistence through shared purpose rather than control, and the famous scaling capitulation on the record, dated, with its trigger and its residue.

Open the Bach Framework →
Live · eval passed

Gavin Baker

AI infrastructure · semiconductors — the Framework that failed twice before it passed, all on the record

70.6%direction fidelity, 17 out-of-sample tests — third eval, two disclosed failures first
Two lanesdoctrine gates the pass (64.3%); fast-flipping vendor calls disclosed separately (3/3)
1 honest misstook a lower score rather than fire a mechanism verification had killed

The scaling-laws worldview — so the largest companies with the most compute and data win, you can't out-Nvidia Nvidia in a head-on assault — with a value investor's price discipline, his own pre-registered warning lights, and dated full reversals on Nvidia and Google that taught us how to grade a fast revisionist.

Open the Baker Framework →
Live · eval passed

Michael Howell · interview-derived

Global liquidity · CrossBorder Capital; the flows-and-refinancing lens on every asset

88%direction fidelity, 34 out-of-sample tests · silences 100% correct
~3 mo / 6-9 mohis quantified lags: liquidity leads Bitcoin, then equities
0 inventedevery edge a verbatim span of what he said; killed paths never fired

The global-liquidity spine · Liquidity and balance-sheet capacity, not interest rates, dominate markets · rebuilt from a deep interview backlog, with one edge (oil-to-liquidity) we flag as ours, not his.

Open the Howell Framework →
Live · eval passed

Ray Dalio

The economic machine · debt cycles, the Big Cycle of empires, money and gold

87.5%direction fidelity, 16 out-of-sample tests · principle/cycle-level
21 excludedguest/relay/cohost voices dropped; only his 19 own-voice claims graded
0 inventeddeclined the AI-capex-bubble call three times

The long-term debt cycle · with rates hitting zero the government prints money and the central bank buys its debt · to gold and a shifting world order, with a Bitcoin-calibration miss disclosed.

Open the Dalio Framework →
Live · held-out-interview eval passed

Raoul Pal · interview-derived

Macro · the Exponential Age — built from interviews where he is the guest: his voice alone, by design

72%direction fidelity — tested against a held-out 2026 interview (18 tests)
43stated invariants — the most credo-dense Framework we publish
2 dated shiftsincl. the four-year cycle loosening between 2024 and 2026

The debasement loop — That's you taking a credit card to pay off your credit card payments — the AI-builds-AI recursion, his own stated falsifiers, and the war-inflation mind-change the eval caught mid-flight and scored honestly as a miss.

Open the Pal Framework →
Live · eval passed

Nassim Taleb · geniuses lane · interview-derived

Tail risk · fragility — the Framework our claims pipeline failed on, rebuilt from his conversations

66.7%direction vs his real 2026 claims, which the interview-built model never saw — 6 tests, disclosed
11 yrsspine mechanisms unchanged across conversations 2013 → 2024
14xthe mechanism density of the failed claims build — one conversation vs 440 distilled claims

The Incerto with receipts — in the real world you're not paid in frequency right you're paid in dollars and cents, if you go bust you can never recover — the hidden-risk blowup chain, and a model that refuses point predictions because he does.

Open the Taleb Framework →

In the lab

Next Frameworks under construction. Each publishes only after it passes the same out-of-sample test — and when one falls short, we say so here rather than shipping it.

In the labJim BiancoRates, bonds and the macro plumbing. First eval came in below the bar — his AI flip happened in a gap in our claim record; we're rebuilding the record and the model before retesting.
In the labArthur HayesCrypto macro and liquidity cycles. Two evals below the bar — and the second taught us why: he doesn't just flip trades, he flips his whole regime read (the model was structurally bearish-crypto in the exact window he called the bottom in). The two-lane grading that rescued Baker can't fix that; a thinker whose worldview layer moves in weeks needs a faster-folding model than a quarterly freeze. His AI-capex machinery graded near-verbatim, so the corpus is good — the schema isn't, yet.
In the labGraham AllisonGeopolitics — the Thucydides Trap. The model built clean (zero invented mechanisms, perfect silences on markets it has no view on), but even a full-year test window held too few of his claims to grade — a scholar's cadence, not a model failure. The bar requires the score be demonstrated, not assumed. Next attempt: the interview-derived path that Howell and Pal just passed.
In the labMichael LevinBiology, agency and bioelectric intelligence. The model built clean and its one gradeable test was a near-verbatim hit — but one test is not a demonstration, and we don't publish a "100%" off a sample of one. Queued for the held-out-interview eval his long-form record is well suited to.