The
Lab
Where theses go to be tested in public,
without a predetermined conclusion.
The default in this space is binary. You are either endorsed or destroyed. The cheerleader or the attacker. The Lab is a third position: here is the thesis as its author intended it, here is what the data says, here is what we still don’t know.
Every investigation begins with a claim that is specific enough to test. Every claim is attributed to its originator, presented fairly, and examined without a predetermined verdict. If the data supports the thesis, that gets documented. If it doesn’t, that gets documented. If the answer is genuinely uncertain, that is the finding.
This is collaborative forensic analysis. It produces work that neither side of a debate typically builds — because both sides already know what they want to find.
A thesis, not a topic. One falsifiable claim. The originator must be willing to be named and to engage with the findings.
The third position.
Neither cheerleader nor assassin.
Three conditions.
A thesis must meet all of them.
The Lab does not investigate topics. It investigates claims. The distinction matters. “Is stablecoin adoption a threat to Bitcoin?” is a topic. “Stablecoin share of on-chain transaction volume will remain above 70% for the next 36 months on Ethereum, Solana, Tron, and BNB Chain” is a thesis. The Lab only accepts the second type.
A threshold. A time window. A mechanism. A chain of causation. If there is no possible world in which the claim could fail, there is no test to run. The Lab requires a claim that could, in principle, produce the finding: this is false.
Every investigation credits its originator fully — name, title, the platform where the thesis appeared. The originator is invited to review the analysis before publication and to respond in writing if they disagree with the findings. Anonymous theses are not accepted.
Not agreement — engagement. The analysis improves when the person who built the thesis challenges the test design. The Lab is not a drive-by forensic. The originator sees the methodology before the verdict. If they find an error, it gets corrected and credited.
The Lab does not accept commissioned work. Investigations are opened at The Lab’s discretion. Submitting a thesis is not a purchase. The Lab cannot be retained, paid, or engaged to produce analysis — by the originator or by anyone acting on their behalf.
If you have a financial interest in the outcome of an investigation — you are long the asset, managing a fund exposed to it, or selling a product that benefits from a favourable finding — that must be disclosed in your submission.
Undisclosed conflicts of interest will result in the investigation being closed and the fact of non-disclosure published as part of the record. The disclosure itself is not disqualifying. The concealment is.
The Lab is not a commercial service. If you are looking for commissioned research, private intelligence, or consultancy — bespoke analysis built to your brief, delivered under engagement terms — that work is available separately.
It is governed by a different engagement model entirely, with different scope, confidentiality terms, and deliverables. The two are kept apart by design. The Lab’s independence depends on that separation remaining absolute.
To discuss commissioned work: paulfaulkner.com/contact or view services.
The 7/93 Model
Conclusion: The thesis fails on its own falsifiable conditions. The 7% figure does not hold under audit (derivation yields 0.5–10% depending on denominator). The stablecoin dominance claim holds directionally, but the central quantitative claim is unsupported.
Note on process: The originator was notified of the complete findings via email on 5 May 2026, in accordance with The Lab’s collaborative engagement policy. No response or substantive engagement was received. The investigation is therefore closed as Concluded — Thesis Failed on forensic evidence alone.
Cash Erosion Model
Conclusion: The thesis fails on Condition 5 at the first audit step. “UK M2” does not exist as an official aggregate — the Bank of England uses M4. The 8% monetary expansion figure is not supported by BoE M4 data over any defensible averaging period. The 11% combined erosion rate is not derivable from authoritative sources. The model has cited a non-existent aggregate to support a rate it cannot justify.
Note on process: The originator submitted anonymously and was not contactable for pre-publication engagement. The investigation proceeded on the forensic evidence alone and is closed as Concluded — Thesis Failed.
Bitcoin: Pricing a Probability Distribution
A sophisticated framework from an institutional analyst within corporate finance and capital markets. The originator is not disputing that Bitcoin could fail — they are arguing that the mechanism of failure is structural and trackable, defined by six interacting vectors (regulatory terminality, technical fragility, security sustainability, adoption durability, competitive displacement, geopolitical hostility). The thesis contends that markets treat survival probability as a static input rather than a dynamic variable, and that this gap defines a systematic mispricing.
The Lab is testing not whether the regime framework is appealing, but whether it is actually falsifiable. Can p_zero be derived from observable data? Do the six vectors produce measurable inputs with defined thresholds? Does the framework produce a price range with a confidence interval? And — most critically — does the evidence support the claim that institutional markets are systematically treating survival probability as static?
The 200-Week MA Conviction Model
Conclusion: The thesis fails on all eight forensic criteria. The 200-week MA is a descriptive statistic, not a predictive model — a category error the originator’s bond-ladder analogy makes precise but cannot resolve. The MA’s own growth rate has declined 74% from the early era (2.05%) to current (0.53%), cycling downward with each halving. The framework conflates the MA reaching $1M with spot price reaching $1M — a lagging average that reflects the price 1,400 days after the fact. And by the originator’s own framework, a spot price 1.28× above the MA is not an accumulation window. It is the sell signal. The thesis is its own refutation.
Note on process: The originator was contacted via LinkedIn prior to publication. The response to a direct question about whether the framework constituted a falsifiable price model produced a bond-ladder analogy, then a chart. No model was submitted. The investigation proceeded on the forensic evidence and the interactive lab tool. The originator’s response pattern is itself documented as Finding 8 of the investigation.
The Monetary Debasement Model
Conclusion: The thesis fails on all five forensic conditions. S&P ÷ M2 over 67 years is +51% — not almost flat. CPI-adjusted price return is +951% before dividends. The correct real-return deflator is CPI, not M2; the denominator was chosen because dividing any long-run return series by any large growing number compresses it. The post cites no data source, states no methodology, and carries an undisclosed commercial agenda. Disproved in under 30 minutes on primary data.
Note on process: The originator was not contacted prior to publication. The data is unambiguous and requires no originator input to resolve. The originator is invited to respond via thelab@paulfaulkner.com. Any substantive response will be published in full.
The USD Debasement Model
Conclusion: Partial. The nominal case holds across all prescribed start dates — the dollar is lower in trade-weighted terms today than in 1964, 1973, 1985, or 2002. But the BIS Real EER from January 1973 to March 2026 records +0.12%. The inflation adjustment eliminates the entire apparent nominal weakness from the analytically correct anchor. The current real EER of 103.21 sits 10.58 points above its 62-year mean of 92.63 — consistent with cyclical elevation, not structural erosion.
Note on process: This is The Lab’s first Partial verdict and its first investigation originating from a constructive analytical engagement rather than an adversarial one. The originator identified the correct methodological question. The data answered it. Tiaan Fourie is invited to respond via thelab@paulfaulkner.com. Any substantive response will be published in full.
The Gold Debasement Model
Conclusion: Partial. Gold outperforms interest-bearing cash from every anchor tested — 1971, 1975, 1980, and 2000 — in both nominal and real CPI-adjusted terms. The investment case for gold survives. The debasement interpretation does not. The $35 anchor was a government-enforced fixed price never available in the free market. The originator’s own 1980 comparison used the intraday peak price ($850) rather than the monthly average ($668) — corrected data shows gold winning that comparison too. And the Eurodollar market, operational since 1957, demolishes the causal architecture the claim requires.
Finding 6 — Structural: The debasement narrative requires a singular monetary break point in August 1971. The Eurodollar market — operational since 1957, built on Soviet USD holdings in London clearing banks placed outside Federal Reserve jurisdiction — demonstrates the dollar’s international monetary role was already operating outside Bretton Woods fourteen years before Nixon closed the gold window. The causal architecture fails before the data is examined.
Gold vs USD Cash — Interest, Ownership & the 1980 Exception
Conclusion: Partial — gold’s favour. The interest adjustment was already embedded in the total return series supplied by the originator; applying it changes nothing. Gold wins 1.95× from the first legal ownership date (Dec 1974), 7.64× from August 1971, and 2.68× from December 1977 — all after full monthly T-bill reinvestment. The ownership ban, correctly applied, removes the period most structurally unfavourable to gold and still leaves gold winning by 138bps p.a. The January 1980 exception is real — cash wins by 53bps p.a. — but requires selecting gold’s single most extreme month in history as an entry point. The AI-generated counter-thesis that framed the investigation (“compound interest frequently matches or outpaces gold”) is false for four of five periods tested.
Note on process: Tiaan Fourie supplied a 15-sheet dataset covering July 1944 to March 2026 and submitted the thesis directly following Lab Case 007. He is credited as data originator. The investigation notes that his own dataset, correctly read, contradicts the AI-generated counter-thesis he included with the submission. Tiaan is invited to respond via thelab@paulfaulkner.com. Any substantive response will be published in full.
The XCN Call — A Public Price Prediction Under Forensic Review
Conclusion: Fail on all four falsifiable conditions. Year-end 2025 price: $0.0044 — down 79% from video-date price. Post-video local peak: $0.0211 (+2.4%), then uninterrupted decline. The January pump described as “just a dip” was the local top; unrecovered in 13 months. The $1 target is 200× from current price with no observable trajectory. Structural finding: 15 revenue streams across two domains are structurally decoupled from the trade outcome. Three-layer disclaimer architecture documented. No contact mechanism exists by the subject’s own design.
Note on process: Self-initiated. No external thesis originator. The claims were made publicly to 54,931 viewers. No contact has been made and none is possible — the subject’s website explicitly states he cannot be reached directly. The claims were public. The data closes the case. This is the work. Part 2 — How Retail Gets Farmed — publishes separately.
The Fill-Down Fallacy
Conclusion: Fail on all three active findings, one structural. The table is one formula repeated 74 times: $100,000 × (1 − 0.042)ⁿ. FRED CPIAUCSL (post-Volcker, 1983–2026, n=520 months) shows 4.2% is the 84th percentile — only 15.8% of monthly readings have ever exceeded it; in the decade 2010–2019, zero months did. At the Fed’s stated target (2.0%), $100,000 reaches $22,420 by 2100 — 5.3× the published figure. The 10-year TIPS real yield (FRED DFII10, June 2026) is +2.15%: $100,000 in TIPS grows to $482,600 in real purchasing power over 74 years — 115× the implied cash terminal value. TIPS is the instrument class designed specifically for the problem described. It is not mentioned once. The cash-vs-Bitcoin binary is the product architecture of a Bitcoin-focused RIA, not an analytical conclusion. The rolling window failure was generated by the author in the same thread: after providing current-window data showing Bitcoin −9% (2yr) vs Gold +77%, he shifted to September 2025 to show Bitcoin +335% and described September as honest and the current window as cherry-picking.
Note on process: Self-initiated following two public challenges in the thread: “pull the FRED data it’s free, your thesis collapses” and “Model it, all scenarios — I challenged you a few backs to model it, you lack the skills and knowledge.” The author published a companion Substack article using the same single-assumption structure instead. FRED data was downloaded and analysis completed on 12 June 2026 — the same day both the post and companion article appeared. Charlie Andrys is invited to respond and submit a revised analysis via thelab@paulfaulkner.com. Any substantive response will be published in full.
While the first meeting was still running,
the analysis already existed.
The 7/93 investigation — from Raymond’s LinkedIn thread to specced, red-teamed, forensically audited, and published — took under one hour. Concept to live research page while the originator was still in the thread. This is not a claim about speed. It is a claim about what speed at this level makes possible — that by the time a conventional team has finished scoping, the intelligence product exists, the arithmetic is sourced, the methodology is documented, and the five falsifiable conditions are mapped to their data pipelines.
