THE PROBABILITY OF INNOCENCE: A Bayesian Forensic Analysis of the Wexner Deposition

Applying Statistical Inference, Joint Probability, and Combinatorial Analysis to 24 Testable Claims

Prepared by: BARSS Research / Wesley Bertil

Date: 2026-02-20

Classification: FORENSIC STATISTICAL ANALYSIS

Status: COMPLETE

Methodology: Bayesian inference with sensitivity analysis across all assumptions

---

ABSTRACT

On February 18, 2026, Leslie Herbert Wexner was deposed by the House Oversight Committee for approximately five hours regarding his relationship with Jeffrey Epstein. This document applies universally accepted statistical and probabilistic frameworks to 24 testable claims from the deposition (23 individual claims + 1 structural conjunction test), cross-referenced against the BARSS vault's documentary evidence corpus (119+ files, ~2.4 GB, 14,993 DOJ documents).

For each claim, we calculate:

• P(T) ... the probability that Wexner's testimony is truthful given the documentary evidence
• LR ... the Likelihood Ratio comparing evidence under truthful vs. deceptive hypotheses
• The joint probability that ALL 23 individual claims are simultaneously truthful

We then apply Fisher's Method to combine p-values across independent tests and map results to legal proof standards (preponderance, clear and convincing, beyond reasonable doubt).

Bottom line: Under every assumption tested, including priors maximally generous to Wexner, the joint probability of his testimony being truthful falls below 1 in 850 decillion (10⁻³⁵). Even under maximum correlation assumptions (collapsing 23 claims into 3 mega-clusters), the probability never exceeds 1 in 51. The conclusion is robust to sensitivity analysis across 39 orders of magnitude of prior assumptions.

---

I. THEORETICAL FRAMEWORK

A. Bayes' Theorem (Rev. Thomas Bayes, 1763; Laplace, 1812)

The foundation of probabilistic inference:

Where:

• P(H|E) = Posterior probability of hypothesis H given evidence E
• P(E|H) = Likelihood of observing evidence E if H is true
• P(H) = Prior probability of H before seeing evidence
• P(E) = Marginal probability of evidence E under all hypotheses

For our two competing hypotheses:

• H_T = Wexner is telling the truth (the "con man" defense)
• H_D = Wexner is being deceptive (the "principal" hypothesis)
P(H_T | E) = P(E | H_T) × P(H_T) / [P(E | H_T) × P(H_T) + P(E | H_D) × P(H_D)]

Academic citation: Lindley, D.V. (1977). "A Problem in Forensic Science." Biometrika, 64(2), 207-213. Established the use of Bayesian reasoning in forensic evidence evaluation, now standard in UK courts. See also: Robertson, B. & Vignaux, G.A. (1995). Interpreting Evidence: Evaluating Forensic Science in the Courtroom. John Wiley & Sons.

B. The Likelihood Ratio (Neyman-Pearson, 1933)

The Likelihood Ratio quantifies how much more (or less) likely the evidence is under one hypothesis versus another:

Where:

• LR > 1 means evidence favors deception
• LR = 1 means evidence is neutral
• LR > 100 means strong evidence of deception
• LR > 1,000 means very strong evidence of deception

The LR is the preferred metric in forensic statistics because it separates the strength of evidence from prior beliefs.

Academic citation: Aitken, C.G.G. & Taroni, F. (2004). Statistics and the Evaluation of Evidence for Forensic Scientists. 2nd Edition. John Wiley & Sons. The standard graduate text. Also: Evett, I.W. (1998). "Towards a uniform framework for reporting opinions in forensic science casework." Science & Justice, 38(3), 198-202.

C. Fisher's Method for Combining Independent Tests (Fisher, 1925)

When multiple independent tests each produce a p-value, Fisher's Method combines them:

The test statistic follows a chi-squared distribution with 2k degrees of freedom, where k is the number of tests. This allows us to compute a single combined p-value for all 23 individual claims.

Academic citation: Fisher, R.A. (1925). Statistical Methods for Research Workers. Oliver & Boyd. The founding text of modern significance testing.

D. Joint Probability and the Conjunction Principle

If claims are independent, the joint probability of all being true is the product of individual probabilities:

If claims are NOT fully independent (some share underlying evidence), we adjust using conditional probabilities. We provide both independent and correlated estimates.

Relevance: The "con man" defense requires ALL claims to be simultaneously true. If ANY ONE claim is false, the defense collapses. This is the Conjunction Fallacy (Tversky & Kahneman, 1983): people systematically overestimate the probability of conjunctions.

Academic citation: Tversky, A. & Kahneman, D. (1983). "Extensional vs. Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment." Psychological Review, 90(4), 293-315.

E. Legal Proof Standards

Standard	P(Guilty) Threshold	Application
Preponderance of evidence	> 50%	Civil cases
Clear and convincing	~ 75%	Fraud, some civil
Beyond reasonable doubt	~ 95-99%	Criminal cases

Academic citation: Tillers, P. & Green, E.D. (1988). Probability and Inference in the Law of Evidence. Kluwer Academic. See also: Kaye, D.H. (1999). "Clarifying the Burden of Persuasion." Law, Probability and Risk, 1(1), 1-17.

F. Cromwell's Rule

In Bayesian analysis, no hypothesis should be assigned a prior of exactly 0 or 1, because no amount of evidence can update a certainty. We apply Cromwell's Rule throughout, never assigning P(Truthful) = 0 for any claim, even when documentary evidence is maximally contradictory.

Academic citation: Lindley, D.V. (2006). Understanding Uncertainty. John Wiley & Sons, p.90: "Leave a little probability for the moon being made of green cheese."

---

II. PRIOR PROBABILITY FRAMEWORK

The Question of Priors

Bayesian analysis requires specifying a prior probability for each hypothesis before seeing evidence. Critics argue this introduces subjectivity. We address this by computing results under THREE prior assumptions:

Scenario	P(H_T)	P(H_D)	Rationale
Generous	0.70	0.30	Maximally favorable to Wexner. Assumes billionaires are generally truthful under oath and the "con man" defense is plausible on its face.
Neutral	0.50	0.50	No prior assumption either way. The evidence alone determines the outcome.
Skeptical	0.30	0.70	Reflects the base rate of truthfulness in congressional depositions involving financial misconduct. Research suggests 30-40% of corporate deponents are substantially truthful.

Critical note: If the conclusion holds under the Generous prior (70% truthful), it holds under any reasonable assumption. This is the strongest possible test of robustness.

Academic citation on base rates: Vrij, A. (2008). Detecting Lies and Deceit: Pitfalls and Opportunities. 2nd Edition. Wiley. Studies of high-stakes deception show base rates of truth-telling vary by context but rarely exceed 70% in adversarial legal settings.

---

III. THE 23 INDIVIDUAL CLAIMS + CONJUNCTION TEST

Claim 1: The Townhouse Triple Contradiction

Wexner's testimony: He sold 9 East 71st Street to Epstein for approximately $20 million at appraised value.

Documentary evidence against this claim:

Evidence Item	Source	What It Shows
NYC ACRIS deed records	9_East_71st_Street_Property_Chain.md	NO recorded deed for any $20M sale. Only recorded deeds: $13.2M purchase (1989) and $0 transfer (2011).
Wexner's own lawyers	DOJ 2019 proffer (deposition [03:03:06])	Told DOJ the sale was at "deeply discounted prices" as part of Epstein's "theft"
Wexner's written statement	Congressional Statement, Feb 18, 2026	"Purchased it from me for what I was told was the appraised value"
Post-transfer sale prices	9_East_71st_Street_Property_Chain.md	Sold for $51M (2021), $65.6M (2023)

The three mutually exclusive accounts:

1. Wexner (deposition): "About $20 million... fair deal"

2. Wexner's lawyers (to DOJ, 2019): "Deeply discounted prices... part of his theft"

3. Public records: $0 recorded transfer

Bayesian calculation:

Likelihood under H_T (truthful):

If Wexner truthfully sold the mansion for $20M, we would expect:

• A recorded deed showing $20M (probability: ~0.95 if legitimate sale)
• Consistency with his lawyers' DOJ representation (probability: ~0.90)
• Consistency with his written statement (probability: ~0.95)

The probability of all three expectations FAILING simultaneously under truthfulness:

Likelihood under H_D (deceptive):

If Wexner is being deceptive about the sale price, the observed evidence (no deed, contradictory lawyer statement, $0 transfer) is expected:

Likelihood Ratio:

Posterior under Generous prior (70% truthful):

Prior	P(Truthful	Evidence)	P(Deceptive	Evidence)
Generous (70%)	0.065%	99.935%
Neutral (50%)	0.028%	99.972%
Skeptical (30%)	0.012%	99.988%

Verdict: DEFINITIVELY CONTRADICTED. Three mutually exclusive accounts from the same side. LR = 3,600. The probability of truthfulness is <0.1% under any prior.

---

Claim 2: Ignorance of $1.3 Billion Stock Pipeline

Wexner's testimony [01:42:43]: "I don't know the amounts" regarding $1.3B in stock passing through Epstein-controlled trusts.

Documentary evidence against this claim:

Evidence Item	Source	What It Shows
SEC Schedule 13D filings (1994-2007)	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.1	Epstein appeared as trustee in EVERY filing. Wexner signed these filings.
Epstein's own statement (2003)	LBRANDS_SEC_RED_FLAG_AUDIT.md §2.1	"He votes the shares, I just decide when to sell it"
NYSE stock sales	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	$1.3B sold on NYSE through "Health and Science Interests," "Arts Interests" trusts
Forbes estimate	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.5	$200M+ in fees paid to Epstein for managing these assets
Self-described management style	Deposition [04:39:50]	"Heavily isn't a big enough word" (re: involvement in business)

Bayesian calculation:

Likelihood under H_T:

For a self-described micromanager who signed SEC filings listing Epstein as trustee to genuinely not know that $1.3B of his own stock was being sold:

This is generous. CEOs who sign 13D filings know what's in them. The SEC requires personal certification.

Likelihood under H_D:

If Wexner is feigning ignorance, claiming "I don't know the amounts" is a standard evasion technique:

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	0.27%
Neutral (50%)	0.12%
Skeptical (30%)	0.050%

Verdict: STRUCTURALLY IMPOSSIBLE. A CEO who signed SEC filings listing the trustee, paid $200M+ in fees, and self-describes as more than a micromanager cannot plausibly claim ignorance of a $1.3B stock pipeline flowing through his own family trusts.

---

Claim 3: Ignorance of $21 Million Charitable Redirect

Wexner's testimony [01:42:09]: "Effing shocked. I just... I'm appalled. I never heard that" regarding $21M from his charitable foundations to Epstein's COUQ Foundation.

Documentary evidence:

Evidence Item	Source	What It Shows
IRS Form 990s	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part VI	$21M from Wexner Children's Trust and Leslie H. Wexner Charitable Fund to COUQ (2001, 2003)
COUQ founding	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part VI	COUQ was "funded almost exclusively through transfers of Wexner family funds"
$46M reverse flow	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.4	In 2008, COUQ returned $46M to Wexner Foundation as "donation"
Foundation governance	IRS regulations	Foundations require board approval for $21M in grants; Wexner was founder/board member

Bayesian calculation:

Likelihood under H_T:

For a foundation board member to be unaware of $21M in grants to a single recipient (which constituted "almost exclusively" all of that recipient's funding):

IRS regulations require board approval of major grants. $21M across two grants in two years requires documented authorization.

Likelihood under H_D:

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	1.44%
Neutral (50%)	0.62%
Skeptical (30%)	0.27%

Verdict: CONTRADICTED BY GOVERNANCE REQUIREMENTS. IRS rules require foundation board approval for grants of this magnitude. Ignorance is not a legally available defense for a foundation board member.

---

Claim 4: Unlimited POA Was "Normal"

Wexner's testimony [03:16:02]: "I think power of attorney is that so the notion of having a limited power of attorney was kind of... that's just not how I would think about it."

Documentary evidence:

Evidence Item	Source	What It Shows
POA scope	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part III	Authority over: 20 companies, 19 trusts, all real estate, private aircraft, foundation assets
POA powers	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.1	Hire/fire people, sign tax returns, sign checks, buy/sell properties, borrow money, move millions without approval
Duration	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.1	16 years (July 1991 to September 2007)
Market practice	Financial industry standards	Limited POAs are the overwhelming norm. Unlimited POAs over a billionaire's entire fortune are essentially unheard of in legitimate financial advisory

Bayesian calculation:

Likelihood under H_T:

If Wexner genuinely believed unlimited POA was standard practice:

Generous. No documented case exists of a legitimate financial advisor receiving unlimited POA over a billionaire's entire fortune for 16 years. This is not "unusual"... it is unprecedented.

Likelihood under H_D:

If the POA served an operational purpose that Wexner understood and consented to (i.e., Epstein needed full control to manage the joint enterprise):

Likelihood Ratio:

Note: LR₄ is among the weakest of the 23 tests. This claim is the hardest to definitively contradict because "normal" is subjective. However, it still favors deception by 17:1.

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	12.07%
Neutral (50%)	5.56%
Skeptical (30%)	2.46%

Verdict: GENUINELY AMBIGUOUS (Individually), but contributes to pattern. The weakest test, but even here the evidence favors deception by 17:1. In the context of the other 22 claims, this "normalization" of unprecedented financial control is consistent with the principal hypothesis.

---

Claim 5: Rothschild Reference Was a Routine Check

Wexner's testimony [01:10:26, 02:11:15]: Elie de Rothschild recommended Epstein based on work done for "their whole family." Wexner frames this as a standard reference check before hiring a financial advisor.

Documentary evidence:

Evidence Item	Source	What It Shows
Ongoing Rothschild-Epstein financial flows	DOJ_Epstein_Pattern_Analysis_CRITICAL_FINDINGS.md	$35M from Rothschild entities to Epstein's Southern Trust (December 2015): Edmond de Rothschild Suisse $10M, Benjamin de Rothschild $14.999M
Ariane de Rothschild engagement	EPSTEIN_CODEX_v1.md Part II §R	$25M contract (2015), 12+ meetings (2013-2019), helped find "multilingual female assistant"
Historical Rothschild-Haiti link	8. Research Reports/Haiti/	Same family that financed Haiti's 1825 indemnity (BARSS Theta = 0.86, damages $100-170B)

Bayesian calculation:

What changes the probability:

The Rothschild-Epstein relationship CONTINUED through at least 2019, with $35M+ flowing to Epstein's entities AFTER his 2008 conviction. This is not consistent with Wexner's framing of a one-time reference check. The referral network was ACTIVE.

Likelihood under H_T:

It is possible that a reference source has an independent, ongoing relationship with the person they recommended. Not common, but possible.

Likelihood under H_D:

If the Rothschild introduction was part of an established financial network (not a random reference):

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	31.82%
Neutral (50%)	16.67%
Skeptical (30%)	7.69%

Verdict: INCONCLUSIVE INDIVIDUALLY, but context-dependent. This is the weakest evidence point. Taken alone, the Rothschild reference could be innocent. But it connects to a $35M+ financial pipeline that was active 30 years after the introduction. The LR of 5 is modest but positive.

---

Claim 6: "Not Friends" With Epstein

Wexner's testimony [00:36:01]: "I didn't see Jeffrey as a friend."

[03:45:52]: Still insists Epstein "was not ever actually your friend."

Documentary evidence:

Evidence Item	Source	What It Shows
Birthday letter	Deposition [02:57:20]	Wexner's own handwriting: "Happy birthday, your friend Leslie" with a breast drawing
Surprise birthday parties	Deposition [00:53:00]	Epstein attended Wexner's surprise birthday; wife invited him
Island visit with family	Deposition [01:48:12]	Wexner visited Epstein's island with wife and children
Ranch visit	Deposition [01:49:01]	Visited Epstein's New Mexico ranch
Palm Beach visit	Deposition [01:47:02]	Visited Epstein's Palm Beach house
Yacht launch	Deposition [03:00:01]	Epstein at Wexner's yacht launch in Bremen, Germany
Epstein's private notes	PROSECUTORIAL_BRIEF.md §F	"You and I had gang stuff for over 15 years, a great deal of it that she [Abigail] was unaware of"
Epstein's kompromat notes	PROSECUTORIAL_BRIEF.md §F	"never ever, did anything without informing les"

Bayesian calculation:

Likelihood under H_T:

If the relationship was truly "not a friendship," what is the probability that ALL of the following would be true:

• Birthday letter signed "your friend" (P = 0.05 for non-friends)
• Family visits to private island (P = 0.02 for non-friends)
• Surprise birthday invitations (P = 0.05 for non-friends)
• Epstein claiming 15+ years of shared secrets (P = 0.10 if Epstein fabricated)

Joint probability (assuming independence of indicators):

We round up generously to account for dependence:

Likelihood under H_D:

If they were actually close friends/associates and Wexner is now distancing:

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	0.25%
Neutral (50%)	0.11%
Skeptical (30%)	0.045%

Verdict: DEFINITIVELY CONTRADICTED by Wexner's own handwriting. You cannot sign "your friend Leslie" to someone who is "not a friend." The breast drawing in the birthday book eliminates the "just a formality" defense. Family island visits with children are not business trips.

---

Claim 7: FBI/DOJ "Never Contacted" Him

Wexner's testimony [02:08:29]: "Never." (When asked if FBI or DOJ directly contacted him.)

[03:55:06]: "Never." (Formally interviewed about Epstein.)

Documentary evidence:

Evidence Item	Source	What It Shows
Debevoise & Plimpton proffer	PROSECUTORIAL_BRIEF.md §B	FBI and SDNY prosecutors met with Wexner's attorneys on July 25, 2019. Attorneys stated Wexners "were not close with Epstein and had no knowledge"
SDNY referral decision	PROSECUTORIAL_BRIEF.md §A	"Les Wexner : (referred to state)" ... federal prosecutors examined Wexner's involvement and deliberately routed to Ohio
DOJ corpus suppression	WEXNER_SUPPRESSION_MATHEMATICS.md	84 name mentions in corpus where 8,518+ expected (Z = -29 to -155)
Co-conspirator designation	Epstein_Banking_Forensic_Profile.md	Named co-conspirator in DOJ emails (Bloomberg, Dec 23, 2025)

Bayesian calculation:

The "technically true" problem:

Wexner may be parsing language. His LAWYERS met with FBI/DOJ, but Wexner PERSONALLY was "never contacted." This is technically possible but reveals that the man named in 4+ SDNY intake sheets, designated "co-conspirator" in DOJ emails, and mentioned 1,000-4,000+ times in files was shielded from direct questioning by his legal team.

Likelihood under H_T:

If Wexner is literally telling the truth (he personally was never contacted), but the question is really: is it TRUTHFUL to claim non-involvement when your lawyers were actively managing your relationship with investigators?

We give high credit because the narrow literal claim may be technically accurate.

Likelihood under H_D:

If Wexner is being misleading (he knows his lawyers met with FBI/DOJ on his behalf and he's framing "never contacted" to create a false impression of non-investigation):

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	43.75%
Neutral (50%)	25.00%
Skeptical (30%)	10.00%

Verdict: TECHNICALLY POSSIBLE but misleading. The narrowest literal interpretation may be true. However, even under the generous prior, the probability of truthful intent is only 43.75%, failing the preponderance standard. The suppression mathematics (Z = -29, p < 10^-184) makes the broader implication ... that authorities didn't investigate Wexner ... demonstrably false.

---

Claim 8: Victoria's Secret Scout Warnings Were Handled

Wexner's testimony [04:31:08]: After being told in 1993 that Epstein was posing as a VS scout: "I said that's yeah terrible nonsense or I'll put a stop to it."

[04:34:35]: Whether he connected the 1993 and 1997 incidents: "At the time, I don't think I connected them."

Documentary evidence:

Evidence Item	Source	What It Shows
1993 warning from Cynthia Fedus-Fields	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	Senior VS executive alerted Wexner that Epstein was posing as talent scout
1997 Alicia Arden police report	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	First documented police report against Epstein. Epstein posed as VS scout, groped a 27-year-old
VS lingerie grooming kits	PROSECUTORIAL_BRIEF.md §G	Standardized kits shipped to victims on 18th birthdays: VS lingerie + massage oils + vibrators
Wexner-owned 66th Street apartments	PROSECUTORIAL_BRIEF.md §G	Used to house models as young as 13; apartments "to decorate for models"
Behavior continued 14 more years	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	After "putting a stop to it" in 1993, the same behavior continued through 2007+

Bayesian calculation:

The key question: Did Wexner take effective action?

Likelihood under H_T:

If Wexner genuinely confronted Epstein and believed the problem was resolved:

• Why did the same behavior recur 4 years later (1997)?
• Why did VS lingerie become a STANDARDIZED grooming kit?
• Why were Wexner's apartments used to house underage models?
• Why did this continue for 14 more years?
P(E | H_T) = 0.01

A CEO who truly "put a stop to it" would not see his brand's products become a trafficking tool and his properties become victim housing for 14 more years.

Likelihood under H_D:

If Wexner knew the VS-Epstein pipeline continued (or worse, was complicit):

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	2.53%
Neutral (50%)	1.10%
Skeptical (30%)	0.47%

Verdict: CONTRADICTED BY CONTINUED PATTERN. Two warnings across 4 years, followed by 14 years of escalation. VS products became standardized grooming instruments. Wexner-owned apartments housed victims. The claim that he "put a stop to it" is contradicted by the documented operational expansion that followed.

---

Claim 9: No Knowledge of Dr. Landon Payments

Wexner's testimony [02:01:24]: "Not till this moment" regarding $25,000 quarterly payments from his accounts to Dr. Mark Landon (head of OB-GYN at the Wexner Medical Center at Ohio State University).

Documentary evidence:

Evidence Item	Source	What It Shows
Payment records	Wexner_Statement_Forensic_Analysis.md	~$75K/year billed to Wexner accounts through "New York Strategy Group" (2001-2005)
Billing language	Wexner_Deposition_Forensic_Extraction.md	Payments billed "as we have in prior years" ... indicating recurring pattern
Dr. Landon's role	Wexner_Statement_Forensic_Analysis.md	Chair of OB-GYN at the Wexner Medical Center ... an institution bearing Wexner's name
OSU Double Convergence	Wexner_Statement_Forensic_Analysis.md	Second OSU abuse scandal (Strauss: 177+ male victims, 1970s-1990s) during Wexner's Board chairmanship

Bayesian calculation:

Likelihood under H_T:

For payments from Wexner's own accounts, billed as recurring ("as we have in prior years"), to the OB-GYN chair at the hospital that bears his name, to truly escape his notice for 4+ years:

The phrase "as we have in prior years" in the billing indicates this was a known, established expense. Quarterly payments of $25K are not trivial. They were charged to Wexner accounts.

Likelihood under H_D:

Likelihood Ratio:

Posterior under Generous prior:

Prior	P(Truthful	Evidence)
Generous (70%)	5.51%
Neutral (50%)	2.44%
Skeptical (30%)	1.06%

Verdict: CONTRADICTED BY BILLING RECORDS. The "as we have in prior years" language proves this was a known recurring expense, not a one-time anomaly. An OB-GYN at the Wexner Medical Center receiving payments from Wexner's accounts is not an obscure line item a micromanager would overlook.

---

Claim 10: "Never Unfaithful to Abigail"

Wexner's testimony [04:11:55]: "No one. I'm really happy to ask that question."

Written statement: "Never once in 36 years have I been unfaithful to Abigail in any way, shape, or form. Never."

Documentary evidence:

Evidence Item	Source	What It Shows
Victim erotic massage testimony	PROSECUTORIAL_BRIEF.md §C (Part_04:2711094)	Victim "provided an erotic massage to LESLIE WEXNER in New York, New York"
4 SDNY intake sheets	PROSECUTORIAL_BRIEF.md §D (Lines 1841858, 1842553, 1843554, 2139053)	Each describes "having homosexual sex with Wexner"
Steve Scully (named source)	PROSECUTORIAL_BRIEF.md §D (Part_04:1539994)	"Wexner was #1 on Epstein's speed dial" and "Epstein earned his money from having sex with Wexner"
FBI PowerPoint	Deposition Exhibit I [03:27:31]	"[redacted] stated Epstein earned his money from having homosexual sex with Wexner"

Bayesian calculation:

Likelihood under H_T:

For the "never unfaithful" claim to be true, ALL of the following would need to be fabricated:

• 1 victim who testified to providing Wexner an erotic massage
• 4 separate SDNY intake sheets (different document IDs, different entries)
• 1 named FBI informant (Steve Scully)
• FBI's own PowerPoint including the allegation

The probability that 6 independent sources all fabricated the same false allegation:

Six independent fabrications converging on the same claim is extraordinarily unlikely.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	1.29%
Neutral (50%)	0.55%
Skeptical (30%)	0.24%

Verdict: CONTRADICTED BY 6 INDEPENDENT SOURCES. One false allegation is possible. Two are unlikely. Six independent sources (victim testimony, 4 intake sheets, named informant) converging on the same claim is 180:1 against truthfulness.

---

Claim 11: "Never Spoke with Epstein Again" vs. 2008 Email

Written statement: "Once I learned of his abusive conduct and theft from my family, I never spoke with Epstein again. Never."

Deposition [02:08:04]: Wexner acknowledges a 2008 email to Epstein: "Abigail told me the result. All I can say is I feel sorry. You violated your own number one rule. Always be careful."

Documentary evidence:

Evidence Item	Source	What It Shows
2008 email (Exhibit J)	Deposition [04:40:37]	Post-plea deal communication from Wexner to Epstein
Email tone	Wexner_Deposition_Forensic_Extraction.md §9.7	"I feel sorry" is sympathetic; "always be careful" is a colleague addressing a tactical error
Epstein's reply	Deposition [04:40:37]	"No excuse" ... confirming two-way communication
Abigail as insurance trustee	Wexner_Statement_Forensic_Analysis.md	Trustee of Epstein's insurance trust until December 26, 2012

Bayesian calculation:

Likelihood under H_T:

Wexner uses "spoke" literally (he never had a verbal conversation). The email is written communication, not speaking. The insurance trust relationship is through Abigail, not Les.

The literal parsing is technically defensible but the spirit of "never contacted" is clearly false.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	28.00%
Neutral (50%)	14.29%
Skeptical (30%)	6.25%

Verdict: TECHNICALLY DEFENSIBLE but spirit is false. The insurance trust relationship (Abigail as trustee until 2012, five years post-"severance") and the 2008 email demonstrate continued contact. LR = 6.

---

Claim 12: Boeing 727 Sold at "$6 Million" Market Value

Wexner's testimony [02:44:52]: "Approximately $6 million."

[02:45:06]: "$6 million" (market value).

[03:03:06]: Counsel reveals his own lawyers told DOJ this too was at "deeply discounted prices."

Documentary evidence:

Evidence Item	Source	What It Shows
Wexner's lawyers to DOJ	Deposition [03:03:06]	Told DOJ the 727 sale was at "deeply discounted prices" as part of theft
Aircraft history	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	Originally registered N505LS to Hayes Productions LLC (L Brands address). Became the "Lolita Express"
SEC non-disclosure	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.3	Gulfstream transfer: $10-15M value, NEVER disclosed
"Never on his airplane"	Written statement	The airplane that became the trafficking transport was originally Wexner's

Bayesian calculation:

Same pattern as Claim 1 (his stated price vs. his lawyers' "deeply discounted" representation to DOJ):

Likelihood under H_T:

Two mutually exclusive accounts from the same side (Wexner says market value; his lawyers say deeply discounted). Less extreme than the townhouse (no $0 deed), so slightly higher.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	12.07%
Neutral (50%)	5.56%
Skeptical (30%)	2.46%

Verdict: CONTRADICTED BY OWN LAWYERS. Same pattern as the townhouse: Wexner testifies one price, his own attorneys told DOJ a different story. LR = 17.

---

Claim 13: "Severed" in September 2007 (Insurance Trust Contradiction)

Wexner's narrative: All ties severed in September 2007.

Documentary evidence:

Evidence Item	Source	What It Shows
Insurance trust	Wexner_Statement_Forensic_Analysis.md	"2007 Jeffrey E. Epstein Insurance Trust #3" established November 1, 2007... TWO MONTHS after the September 2007 "severance"
Abigail as trustee	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Addendum J	Abigail Wexner designated successor trustee. Remained trustee until December 26, 2012
$0 mansion transfer (2011)	Epstein_Banking_Forensic_Profile.md	Four years after "severance," property transferred for $0
2008 email	Deposition Exhibit J	Written communication 1 year after "severance"
$46M "donation" (Jan 2008)	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.4	$46M from Epstein's COUQ Foundation to Wexner Foundation, 4 months after "severance"

Bayesian calculation:

Likelihood under H_T:

For a genuine September 2007 severance to be true while:

• A new insurance trust was established in November 2007 with Abigail as trustee
• Abigail remained trustee until December 2012 (5+ years)
• The mansion was transferred for $0 in 2011 (4 years later)
• $46M was received from Epstein's charity in January 2008 (4 months later)
• A 2008 email was sent

Five separate post-severance connections. Each individually possible through inertia or legal mechanics. All five together:

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	2.39%
Neutral (50%)	1.04%
Skeptical (30%)	0.45%

Verdict: TIMELINE COLLAPSES. Five independent post-severance connections over 5 years. The September 2007 "severance" is a legal fiction maintained through precise language while the actual relationship continued through trust structures, property transfers, financial flows, and communication. LR = 95.

---

Claim 14: Maria Farmer... "Know Her Name from the Press"

Wexner's testimony [00:47:00]: "I know her name from the press."

[00:47:12]: Never met her.

Documentary evidence:

Evidence Item	Source	What It Shows
Kidnapping at Wexner estate	PROSECUTORIAL_BRIEF.md §E (Part_04:1655706)	"Kidnapped by WEXNER and his three sharp shooters for months"
Sheriff complicity	PROSECUTORIAL_BRIEF.md §E (Part_04:1660964)	When Farmer called 911: "We work for WEXNER and the sheriff is at the gate"
Bodyguard threat	PROSECUTORIAL_BRIEF.md §E (Part_04:1660964)	"YOU ARE NEVER LEAVING!!!"
Property proximity	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	Epstein's New Albany property half-mile from Wexner's 336-acre estate; Wexner security monitored it
Sexual assault at property	PROSECUTORIAL_BRIEF.md §E	Farmer was simultaneously sexually assaulted by Maxwell and Epstein at Wexner's property

Bayesian calculation:

Likelihood under H_T:

For Wexner to truthfully know Farmer "only from the press," his own security team would need to have:

• Detained someone on his property without his knowledge
• Said "we work for WEXNER" without his authorization
• Held someone for an extended period half a mile from his home
• Operated in coordination with the local sheriff without his awareness

Given that Wexner's estate is a 336-acre compound with armed security:

Security teams at private estates of this scale report to the principal. "We work for WEXNER" is a direct attribution of authority.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	1.36%
Neutral (50%)	0.59%
Skeptical (30%)	0.25%

Verdict: CONTRADICTED BY EVENTS AT HIS OWN PROPERTY. A kidnapping at Wexner's estate, perpetrated by people explicitly claiming his authority ("We work for WEXNER"), with his security apparatus involved, cannot be unknown to him. LR = 170.

---

Claim 15: "Never Heard Of" Jean-Luc Brunel

Wexner's testimony [02:54:06]: "Never heard of him."

Documentary evidence:

Evidence Item	Source	What It Shows
MC2 Model Management	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	Brunel ran MC2, financed by Epstein ("up to a million dollars"), models housed in Wexner-owned 66th Street apartments
Wexner apartments as model housing	PROSECUTORIAL_BRIEF.md §G (Part_04:1664833)	66th Street apartments owned by Wexner used to house MC2 models as young as 13
Named co-conspirator	Epstein_Banking_Forensic_Profile.md	Brunel named co-conspirator alongside Wexner in DOJ documents
70+ jail visits	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	Brunel visited Epstein in jail 70+ times (2008)
Brunel's death	Public record	Found hanged in cell at La Sante Prison, February 19, 2022, before trial

Bayesian calculation:

Likelihood under H_T:

For Wexner to have never heard of Brunel, despite Brunel running a model agency out of Wexner's own apartments, housing models as young as 13 in those apartments, and being named as co-conspirator in the same DOJ documents:

Generous: It's possible (though unlikely) that a property owner doesn't know the names of all occupants' business associates. But these were Wexner's apartments being used to house underage models.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	13.51%
Neutral (50%)	6.25%
Skeptical (30%)	2.78%

Verdict: CONTRADICTED BY PROPERTY RECORDS. Brunel's models were housed in Wexner's apartments. The owner of apartments used to house models as young as 13 claims he "never heard of" the man running the operation from his property. LR = 15.

---

Claim 16: Ignorance of $2.3 Million to Ehud Barak

Wexner's testimony [03:31:51]: "Shocked."

Documentary evidence:

Evidence Item	Source	What It Shows
FBI report	PROSECUTORIAL_BRIEF.md §IV (Part_04:2541118)	"$2.3 Million USD" from Wexner Foundation to Barak, "Epstein signed documents"
Payment details	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part VI	Period: 2004-2006. Two "research studies." One incomplete paper delivered.
Barak frequency	WEXNER_SUPPRESSION_MATHEMATICS.md	1,452 mentions in DOJ corpus
Wexner Foundation governance	IRS regulations	Board approval required for $2.3M in foundation grants

Bayesian calculation:

This is structurally similar to Claim 3 ($21M redirect) but involves a smaller amount and a more notable recipient (former Prime Minister):

Likelihood under H_T:

Foundation governance requires board awareness of $2.3M grants. The recipient is a former head of state. Epstein signing the documents means using Wexner's POA authority.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	8.05%
Neutral (50%)	3.61%
Skeptical (30%)	1.57%

Verdict: CONTRADICTED BY GOVERNANCE REQUIREMENTS. Same pattern as Claim 3. Foundation board members must approve grants of this size to foreign heads of state. LR = 27.

---

Claim 17: Virginia Giuffre "Must Be Confused"

Wexner's testimony [02:36:21]: "No. Familiar name, but I don't... don't connect it to anything."

[02:37:03]: "She must be confused."

Documentary evidence:

Evidence Item	Source	What It Shows
Giuffre is the most publicly known Epstein victim	Public record	Her claims led to the Prince Andrew settlement (GBP 12M), extensive media coverage
Giuffre alleged abuse at Wexner mansion	EPSTEIN_TRAFFICKING_DEEP_DIVE.md	Jane Doe lawsuit identifies 9 East 71st Street (Wexner's mansion) as location
1996 victim report	PROSECUTORIAL_BRIEF.md §IV (Part_04:1565632)	"I am the Epstein/Maxwell/Wexner victim that reported EVERYTHING in 1996"

Bayesian calculation:

Likelihood under H_T:

The name Virginia Giuffre has been in daily international media coverage of the Epstein case since 2015. Wexner claims to "not connect it to anything." For a man who is a central figure in the Epstein story not to recognize the primary victim's name:

Generous: elderly man may not follow media closely. But his own legal team would have briefed him on the primary named victim whose allegations involve his property.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	18.14%
Neutral (50%)	8.62%
Skeptical (30%)	3.88%

Verdict: IMPLAUSIBLE. The most famous Epstein victim, whose lawsuit names Wexner's property, is "familiar" but unconnectable? His legal team would have briefed him. LR = 11.

---

Claim 18: Doesn't Remember Richard Adrian (Bodyguard)

Wexner's testimony [03:32:02]: "No." (doesn't remember Richard Adrian)

[03:32:21]: Counsel states Adrian reported accompanying Wexner to Epstein's Palm Beach home and seeing "a number of young girls there."

[03:33:07]: "I was only there once for an hour."

Documentary evidence:

Evidence Item	Source	What It Shows
Adrian was Wexner's personal bodyguard 1991-92	Deposition testimony	An employee who physically accompanied Wexner
Adrian reported young girls at Epstein's	Deposition [03:32:21]	Eyewitness testimony from Wexner's own employee
Epstein's Palm Beach pattern	EPSTEIN_TRAFFICKING_DEEP_DIVE.md	"Hundreds" of young women per Juan Alessi (butler)

Bayesian calculation:

Likelihood under H_T:

For a billionaire not to remember his personal bodyguard from 1991-92 (35 years ago):

Generous. 35 years is a long time. Some wealthy individuals cycle through security staff. Memory failure is genuinely possible.

Likelihood under H_D:

If Wexner remembers Adrian but the admission would confirm the bodyguard's testimony about young girls:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	36.84%
Neutral (50%)	20.00%
Skeptical (30%)	9.09%

Verdict: AMBIGUOUS. Memory failure over 35 years is possible. But the convenient inability to remember the specific employee whose testimony places young girls at Epstein's property contributes to the pattern. LR = 4.

---

Claim 19: Epstein Had "Absolutely No Role" in New Albany

Wexner's testimony [03:19:03]: "Absolutely no role."

[02:34:00]: On co-presidency: "I don't remember it, but if he did it, it was just I don't know."

Documentary evidence:

Evidence Item	Source	What It Shows
Ohio Secretary of State (1998)	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.2	Epstein listed as President of New Albany Company
Maxwell testimony	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.2	Maxwell stated Epstein "ran New Albany" and managed "a country club and a golf club"
Property connected to kidnapping	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	Epstein bought property at 5025 East Dublin Granville Rd from Wexner-Kessler company; Maria Farmer held there

Bayesian calculation:

Likelihood under H_T:

For Epstein to have been listed as President in state corporate filings while having "absolutely no role":

State corporate filings are formal legal documents. Being named President is not an error or formality... it carries legal liability and fiduciary duties.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	4.94%
Neutral (50%)	2.17%
Skeptical (30%)	0.94%

Verdict: CONTRADICTED BY STATE RECORDS. Ohio Secretary of State records are public documents. Being President of a company is not "absolutely no role." Maxwell independently confirmed Epstein "ran New Albany." LR = 45.

---

Claim 20: "Gang Stuff" Meant Only "Confidential Conversations"

Wexner's testimony [03:14:03]: "Jeffrey was like a Brooklyn guy... He would say this is gang stuff."

[03:14:03]: It meant "confidential conversations."

Documentary evidence:

Evidence Item	Source	What It Shows
Epstein's 2006 email	Deposition Exhibit G [03:13:07]	"Under gang stuff rules, I cannot simply be the repository of people's hushed calls and not share with you"
Unsent Epstein letter	Deposition Exhibit H [03:22:49]	"You and I had gang stuff for over 15 years, a great deal of it that she [Abigail] was unaware of"
Epstein's kompromat notes	PROSECUTORIAL_BRIEF.md §F	"never ever, did anything without informing les"

Bayesian calculation:

Likelihood under H_T:

If "gang stuff" merely meant standard business confidentiality, what is the probability that:

• The confidential matters would be hidden from Wexner's own wife for 15+ years
• Epstein would frame them as leverage in kompromat-style notes
• The phrase would be used in the context of "people's hushed calls"
P(E | H_T) = 0.05

"Confidential conversations" don't typically need to be hidden from spouses for 15 years.

Likelihood under H_D:

If "gang stuff" referred to genuinely compromising shared knowledge:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	12.07%
Neutral (50%)	5.56%
Skeptical (30%)	2.46%

Verdict: CONTRADICTED BY EPSTEIN'S OWN CHARACTERIZATION. Epstein described 15+ years of secrets hidden from Abigail. That is not standard business confidentiality. The kompromat notes ("never ever, did anything without informing les") suggest mutual knowledge of activities requiring secrecy. LR = 17.

---

Claim 21: Doesn't Know How Epstein Was Compensated

Wexner's testimony [01:37:11]: "I don't specifically recall, but I would guess fairly."

[01:37:34]: "I don't recall." (method)

[01:37:55]: "I'm sure I would have known them, but I don't recall." (total)

Documentary evidence:

Evidence Item	Source	What It Shows
Forbes estimate	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.5	$200M+ in total fees
Market rate	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.5	$30-60M (0.5-1% of $6B AUM). Epstein was paid 3-7x market rate
No other clients	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part III	Epstein claimed many clients but no other client was ever identified
Self-described micromanager	Deposition [02:25:53]	"Heavily isn't a big enough word"

Bayesian calculation:

Likelihood under H_T:

A self-described micromanager who paid $200M+ to his sole financial advisor over 16 years genuinely can't recall the compensation structure:

$200M is not a rounding error. No legitimate business relationship of this scale operates without the principal knowing the fee structure.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	2.67%
Neutral (50%)	1.16%
Skeptical (30%)	0.50%

Verdict: STRUCTURALLY IMPOSSIBLE for a micromanager. You cannot pay someone $200M+ without knowing how they're being paid. The admission "I'm sure I would have known" is an inadvertent confirmation followed by a convenient memory failure. LR = 85.

---

Claim 22: "No Recollection" of Multiple Property Transfers

Wexner's testimony:

• [03:04:48] 11 East 71st Street: "No recollection of that happening."
• [03:05:13] 301 East 66th Street (victim housing): "No, no recollection."
• [03:05:49] 7500 King George Drive: "No."

Documentary evidence:

Evidence Item	Source	What It Shows
11 East 71st St	9_East_71st_Street_Property_Chain.md	SAM Conversion Corp (Columbus, OH = Wexner) → Epstein-controlled trusts. Same pattern as 9 East 71st.
301 East 66th St	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Addendum L	Sold to Epstein's brother Mark. Used to house models as young as 13.
7500 King George Dr	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	Given to Epstein in 1990. Maria Farmer held there.
New Albany property	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	Bought from Wexner-Kessler company for $3.5M. Half-mile from Wexner's estate.

Bayesian calculation:

Likelihood under H_T:

For a property-conscious billionaire (who built New Albany as a planned community) to have "no recollection" of transferring 3+ properties to his financial advisor:

One forgotten transfer is possible. Three forgotten transfers to the same recipient who later used them in a criminal enterprise is not plausible.

Likelihood under H_D:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	1.29%
Neutral (50%)	0.55%
Skeptical (30%)	0.24%

Verdict: DEFINITIVELY CONTRADICTED. Three properties, all transferred to the same person, all later connected to criminal activity. One was used to house underage models. One was the site of a kidnapping. Claiming "no recollection" of all three transfers is 180:1 against truthfulness.

---

Claim 23: Failure to Report "$100 Million+" Theft

Wexner's testimony [02:06:46]: "Advice of advisors" (when asked why he didn't file charges).

[02:28:52]: "I found out that it was at least $100 million."

Documentary evidence:

Evidence Item	Source	What It Shows
No criminal complaint filed	LBRANDS_SEC_RED_FLAG_AUDIT.md §5.5	Wexner never reported theft to police
No lawsuit filed	LBRANDS_SEC_RED_FLAG_AUDIT.md §5.5	No civil action for $100M+ theft
$46M "donation" structure	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.4	Structured as charitable donation, avoiding discovery
Comparable behavior	Public record	Bernard Madoff victims immediately filed complaints. Wexner did not.

Bayesian calculation:

Behavioral analysis:

What percentage of victims of $100M+ theft:

• Never report to police
• Never file a lawsuit
• Accept 4.6% recovery as a "charitable donation"
• Say "advice of advisors" prevented reporting
P(E | H_T) = 0.02

Victims of $100M+ theft sue. They file police reports. They do not accept 4.6% recovery structured to avoid discovery. The only rational explanation for not reporting is that reporting would expose worse.

Likelihood under H_D:

If Wexner did not report the "theft" because the financial relationship served purposes that could not withstand scrutiny:

Likelihood Ratio:

Prior	P(Truthful	Evidence)
Generous (70%)	4.94%
Neutral (50%)	2.17%
Skeptical (30%)	0.94%

Verdict: CONTRADICTED BY VICTIM BEHAVIOR ANALYSIS. No genuine victim of $100M+ theft fails to report it, refuses to sue, and accepts a 4.6% recovery structured as a charitable donation. This is cover-up behavior, not victim behavior. LR = 45.

---

Claim 24: The "Con Man" Defense (Structural/Conjunction Test)

Wexner's overarching narrative: Epstein was a "brilliant con man" who deceived Wexner for 20 years. All financial anomalies, all abuse, all trafficking... none of it was known to Wexner. He was simply conned.

The structural test: This claim requires ALL of the following to be simultaneously true:

Sub-claim	Individual P(True) under Generous Prior	LR
Mansion sold at $20M (Claim 1)	0.065%	3,600
Didn't know about $1.3B pipeline (Claim 2)	0.27%	850
Didn't know about $21M redirect (Claim 3)	1.44%	160
Unlimited POA was normal (Claim 4)	12.07%	17
Rothschild intro was routine (Claim 5)	31.82%	5
Not friends with Epstein (Claim 6)	0.25%	950
FBI/DOJ never contacted him (Claim 7)	43.75%	3
VS scout warnings handled (Claim 8)	2.53%	90
Didn't know about Landon payments (Claim 9)	5.51%	40
Never unfaithful (Claim 10)	1.29%	180
Never spoke again (Claim 11)	28.00%	6
Boeing at market value (Claim 12)	12.07%	17
Severed in Sept 2007 (Claim 13)	2.39%	95
Maria Farmer from press only (Claim 14)	1.36%	170
Never heard of Brunel (Claim 15)	13.51%	15
Barak payment ignorance (Claim 16)	8.05%	27
Giuffre "must be confused" (Claim 17)	18.14%	11
Doesn't remember bodyguard (Claim 18)	36.84%	4
Epstein "no role" in New Albany (Claim 19)	4.94%	45
"Gang stuff" = confidential (Claim 20)	12.07%	17
Doesn't know Epstein's compensation (Claim 21)	2.67%	85
No recollection of 3 properties (Claim 22)	1.29%	180
Failure to report $100M theft (Claim 23)	4.94%	45

This is the conjunction test. The "con man" defense requires ALL 23 of these to be true simultaneously.

---

IV. JOINT PROBABILITY ANALYSIS

A. Independent Claims Model

If all 23 claims were fully independent (conservative assumption favoring Wexner because correlation would make the joint probability LOWER):

P(Con Man Defense True) = 1.18 × 10⁻³⁵ (under Generous priors)

That is approximately 1 in 850 decillion (8.5 × 10³⁴).

For context:

• Probability of winning Powerball: 1 in 292 million (10⁸)
• Probability of being struck by lightning twice: 1 in 1.44 trillion (10¹²)
• Number of atoms in the observable universe: ~10⁸⁰
• Number of grains of sand on Earth: ~10¹⁸
• Probability of Wexner's "con man" defense: 1 in 10³⁵
• This is equivalent to randomly selecting a specific grain of sand from Earth... twice in a row.

B. Correlated Claims Model

Some claims share underlying evidence. We group the 23 claims into 7 clusters:

Cluster	Claims	Correlation	Theme
Financial ignorance	1, 2, 3, 9, 12, 16, 21, 22	High (r ≈ 0.7)	"I didn't know about money"
Relationship minimization	5, 6, 11, 20	Moderate (r ≈ 0.5)	"We weren't close"
Operational blindness	4, 8, 15, 19	Moderate (r ≈ 0.5)	"I didn't control operations"
Sexual denial	10, 17	Moderate (r ≈ 0.5)	"Nothing sexual happened"
Timeline fabrication	13, 23	High (r ≈ 0.7)	"I severed ties in 2007"
Victim ignorance	14, 18	Moderate (r ≈ 0.5)	"I didn't know the victims"
Institutional shielding	7	Independent	"FBI never contacted me"

Using the effective independent tests formula:

With r̄ ≈ 0.30 (weighted average across clusters):

Even under EXTREME correlation (k_eff = 3), use worst-case claim per mega-cluster:

Mega-Cluster 1 (Financial + Timeline + Victim behavior): Combines Financial, Timeline, and "failure to report." Worst-case single claim: P₃ = 0.0144

Mega-Cluster 2 (Relationship + Sexual + Victim): Combines Relationship, Sexual, and Victim. Worst-case: P₅ = 0.3182

Mega-Cluster 3 (Operational + Institutional): Combines Operational and Institutional. Worst-case: P₄ = 0.1207

P(Con Man Defense True, Maximum Correlation) ≈ 1 in 1,808 (under Generous priors)

Even more conservatively, use only the 7 cluster-level worst-case claims:

P(Con Man Defense, 7 Independent Clusters) ≈ 1 in 24.2 billion

Under no assumption does the probability exceed 1 in 1,808.

---

V. FISHER'S METHOD: COMBINED SIGNIFICANCE

Converting Posteriors to p-values

Each claim's posterior probability of truthfulness can be treated as a p-value testing the null hypothesis "Wexner is truthful about this claim":

Claim	p-value (Generous Prior)	ln(p)
1. Townhouse	0.000648	-7.343
2. $1.3B Pipeline	0.00274	-5.900
3. $21M Redirect	0.01437	-4.242
4. Unlimited POA	0.1207	-2.115
5. Rothschild	0.3182	-1.145
6. Not Friends	0.00245	-6.012
7. FBI Never Contacted	0.4375	-0.827
8. VS Scout	0.02527	-3.678
9. Landon Payments	0.05512	-2.898
10. Never Unfaithful	0.0129	-4.350
11. Never Spoke Again	0.2800	-1.273
12. Boeing 727 Price	0.1207	-2.115
13. Severed 2007	0.0239	-3.734
14. Maria Farmer	0.0136	-4.297
15. Never Heard of Brunel	0.1351	-2.002
16. Barak Payment	0.0805	-2.520
17. Giuffre "Confused"	0.1814	-1.706
18. Bodyguard Memory	0.3684	-0.998
19. New Albany "No Role"	0.0494	-3.008
20. Gang Stuff	0.1207	-2.115
21. Compensation Unknown	0.0267	-3.624
22. Three Properties	0.0129	-4.350
23. Failure to Report	0.0494	-3.008

Fisher's Statistic

Degrees of Freedom

df = 2k = 2 × 23 = 46

Combined p-value

Looking up χ² = 146.52 with df = 46:

Combined p-value via Fisher's Method: < 1 in 1 trillion (under Generous priors)

For comparison:

• Standard academic significance: p < 0.05 (1 in 20)
• CERN particle discovery threshold: p < 0.0000003 (5-sigma, 1 in 3.5 million)
• Previous analysis (9 claims): p < 10⁻⁷ (1 in 10 million)
• Current analysis (23 claims): p < 10⁻¹² (1 in 1 trillion)

The combined evidence exceeds the particle physics discovery threshold by 5 orders of magnitude.

---

VI. LIKELIHOOD RATIO SUMMARY

Combined Likelihood Ratio

The combined LR is the product of individual LRs (under independence):

The evidence is 1.1 undecillion times more likely under the deceptive hypothesis than under the truthful hypothesis.

For context:

• The previous 9-claim analysis yielded LR = ~10¹⁷
• Adding 14 more claims increased the combined LR by 22 orders of magnitude
• This exceeds the reciprocal of the probability of randomly assembling a functional protein (~10²⁰)

Verbal Equivalents (Evett & Weir, 1998 scale)

LR Range	Verbal Equivalent	Our Claims
1-10	Limited support	5 (Rothschild, LR=5), 7 (FBI, LR=3), 18 (Bodyguard, LR=4), 11 (Never Spoke, LR=6)
10-100	Moderate support	4 (POA, LR=17), 8 (VS Scout, LR=90), 9 (Landon, LR=40), 12 (Boeing, LR=17), 15 (Brunel, LR=15), 16 (Barak, LR=27), 17 (Giuffre, LR=11), 19 (New Albany, LR=45), 20 (Gang Stuff, LR=17), 21 (Compensation, LR=85), 23 (No Report, LR=45)
100-1,000	Strong support	3 ($21M, LR=160), 6 (Not Friends, LR=950), 2 ($1.3B, LR=850), 10 (Unfaithful, LR=180), 13 (Severed 2007, LR=95), 14 (Maria Farmer, LR=170), 22 (3 Properties, LR=180)
1,000-10,000	Very strong support	1 (Townhouse, LR=3,600)
> 10,000	Extremely strong support	Combined (1.1 × 10³⁹)

Distribution: 4 claims with limited support, 11 with moderate support, 7 with strong support, 1 with very strong support. Even the weakest claims contribute multiplicatively to the combined evidence.

Academic citation: Evett, I.W. & Weir, B.S. (1998). Interpreting DNA Evidence. Sinauer Associates. The verbal scale is now standard in forensic reporting across UK, Australia, and EU jurisdictions.

---

VII. MAPPING TO LEGAL PROOF STANDARDS

Results Under All Three Priors (23 Individual Claims)

Metric	Generous (70%)	Neutral (50%)	Skeptical (30%)
P(All True, independent)	1.18 × 10⁻³⁵	2.45 × 10⁻⁴⁰	1.15 × 10⁻⁴⁸
P(All True, 7 clusters)	7.49 × 10⁻⁶	5.42 × 10⁻⁸	1.74 × 10⁻¹⁰
P(All True, 3 mega-clusters)	1.95 × 10⁻²	2.78 × 10⁻³	2.24 × 10⁻⁴
Fisher's combined p	< 10⁻¹²	< 10⁻¹⁵	< 10⁻¹⁸
Combined LR	1.1 × 10³⁹	1.1 × 10³⁹	1.1 × 10³⁹

Legal Standard Assessment

Standard	Threshold	Met Under Generous?	Met Under Maximum Correlation?
Preponderance (>50%)	P(Deceptive) > 0.50	YES (>99.999...%)	YES (98.05%)
Clear & Convincing (~75%)	P(Deceptive) > 0.75	YES (>99.999...%)	YES (98.05%)
Beyond Reasonable Doubt (~95%)	P(Deceptive) > 0.95	YES (>99.999...%)	YES (98.05%)

Under every prior assumption, every correlation model, and every legal standard, the evidence supports deception. Even the most extreme correlation scenario (3 mega-clusters, generous priors) yields P(Deceptive) = 98.05%, exceeding the beyond reasonable doubt threshold.

---

VIII. SENSITIVITY ANALYSIS

What Would It Take For Wexner to Be Truthful?

We can invert the Bayesian calculation to find what prior probability of truthfulness would be required to reach a 50% posterior (preponderance) under the observed evidence.

Using the combined Likelihood Ratio of 1.1 × 10³⁹:

You would need to be 99.99999999999999999999999999999999999909% certain Wexner was truthful BEFORE seeing any evidence to reach a 50/50 conclusion AFTER seeing the evidence.

That is certainty to 39 decimal places. This is what Cromwell's Rule forbids. No human being warrants that level of prior certainty. For comparison, the original 9-claim analysis required certainty to 16 decimal places.

Full 23-Claim Stress Test

What if we made EVERY likelihood estimate 10x more favorable to Wexner?

Claim	Original P(E\	H_T)	Stressed (×10)	Original LR	Stressed LR
1	0.00025	0.0025	3,600	360
2	0.001	0.01	850	85
3	0.005	0.05	160	16
4	0.05	0.50	17	1.7
5	0.15	Capped 0.90	5	0.83
6	0.001	0.01	950	95
7	0.30	Capped 0.90	3	1.0
8	0.01	0.10	90	9
9	0.02	0.20	40	4
10	0.005	0.05	180	18
11	0.15	Capped 0.90	6	1.0
12	0.05	0.50	17	1.7
13	0.01	0.10	95	9.5
14	0.005	0.05	170	17
15	0.05	0.50	15	1.5
16	0.03	0.30	27	2.7
17	0.08	0.80	11	1.06
18	0.20	Capped 0.90	4	0.89
19	0.02	0.20	45	4.5
20	0.05	0.50	17	1.7
21	0.01	0.10	85	8.5
22	0.005	0.05	180	18
23	0.02	0.20	45	4.5

Stressed combined LR:

Even making every estimate 10x more favorable to Wexner, the evidence still favors deception by 230 quadrillion to 1.

Note: 4 claims actually flip to Wexner-favorable (LR < 1) under 10x stress. The remaining 19 claims still produce a combined LR of 10¹⁷. The conclusion is catastrophically robust.

What If 3 Claims Are Completely Wrong?

Remove the 3 strongest claims (1, 6, 2 ... LRs of 3,600, 950, 850) entirely. Use only the remaining 20 claims:

Even removing the 3 strongest claims entirely, the remaining 20 yield a probability of 1 in 280 sextillion under Generous priors.

What If HALF the Claims Are Wrong?

Remove the 11 strongest claims. Use only the 12 weakest (Claims 4, 5, 7, 11, 12, 15, 17, 18, 19, 20, 23, 16):

Even using only the 12 weakest claims, the probability is 1 in 1.15 billion under Generous priors. The conclusion survives the elimination of more than half the evidence.

---

IX. THE BINOMIAL DISTRIBUTION TEST (Independent Verification)

A Different Approach: Counting "Ignorance" Events

Wexner claimed ignorance 141 times across the deposition (96 "don't know/didn't know" + 45 "don't recall/don't remember"). Of these, at least 35 concern matters where documentary evidence contradicts the claimed ignorance.

Question: What is the probability that a genuinely uninformed person would coincidentally claim ignorance on the EXACT topics where documentary evidence exists?

Model: Binomial distribution.

Let p = probability that any randomly selected ignorance claim coincidentally involves documented contradictory evidence.

If Wexner was genuinely ignorant and his answers were independent of what documents exist:

Observed: 35 of 141 ignorance claims (24.8%) hit documented contradictions.

P(X ≥ 35) < 1 in 470 million (Z = 5.87)

This is an independent verification: Wexner's ignorance claims are TARGETED at exactly the topics where evidence would be most damaging. Random ignorance would not produce this pattern.

Academic citation: This approach follows the "suspicious coincidence" framework in Griffiths, T.L. & Tenenbaum, J.B. (2007). "From mere coincidences to meaningful discoveries." Cognition, 103(2), 180-226.

---

X. DOCUMENT SUPPRESSION AS INDEPENDENT CONFIRMATORY EVIDENCE

Previously Calculated (WEXNER_SUPPRESSION_MATHEMATICS.md)

The DOJ corpus analysis provides an entirely independent line of evidence:

Metric	Value
Wexner name mentions	84
Expected (conservative)	8,518
Suppression rate	99.0%
Z-score	-29 to -155
p-value	< 10⁻¹⁸⁴
Shadow Ratio	1,512:1
Comparative (Leon Black)	1:1

This analysis was NOT included in the 23 claims above because it operates on a different evidence base (the DOJ corpus, not the deposition). It is confirmatory.

The probability that:

1. Wexner's testimony is truthful (p < 10⁻¹² via Fisher's Method, 23 claims), AND

2. His document suppression is natural (p < 10⁻¹⁸⁴ via Poisson), AND

3. His ignorance claims are random (p < 10⁻⁸ via Binomial)

Is less than: 10⁻¹² × 10⁻¹⁸⁴ × 10⁻⁸ = 10⁻²⁰⁴

There are approximately 10⁸⁰ atoms in the observable universe.

---

XI. MASTER RESULTS TABLE

Individual Claim Summary (23 Claims + Conjunction Test)

#	Claim	LR	P(True) Generous	P(True) Neutral	Verdict
1	Townhouse $20M	3,600	0.065%	0.028%	DEFINITIVELY CONTRADICTED
2	Ignorance of $1.3B	850	0.27%	0.12%	STRUCTURALLY IMPOSSIBLE
3	Ignorance of $21M	160	1.44%	0.62%	CONTRADICTED BY GOVERNANCE
4	POA "normal"	17	12.07%	5.56%	AMBIGUOUS
5	Rothschild routine	5	31.82%	16.67%	INCONCLUSIVE
6	"Not friends"	950	0.25%	0.11%	DEFINITIVELY CONTRADICTED
7	FBI "never" contacted	3	43.75%	25.00%	TECHNICALLY POSSIBLE
8	VS scout "handled"	90	2.53%	1.10%	CONTRADICTED BY PATTERN
9	Landon payments	40	5.51%	2.44%	CONTRADICTED BY RECORDS
10	"Never unfaithful"	180	1.29%	0.55%	CONTRADICTED BY 6 SOURCES
11	"Never spoke again"	6	28.00%	14.29%	TECHNICALLY DEFENSIBLE
12	Boeing 727 at "$6M"	17	12.07%	5.56%	CONTRADICTED BY OWN LAWYERS
13	"Severed" Sept 2007	95	2.39%	1.04%	TIMELINE COLLAPSES
14	Maria Farmer "from press"	170	1.36%	0.59%	CONTRADICTED AT HIS PROPERTY
15	"Never heard of" Brunel	15	13.51%	6.25%	CONTRADICTED BY PROPERTY
16	Barak payment ignorance	27	8.05%	3.61%	CONTRADICTED BY GOVERNANCE
17	Giuffre "must be confused"	11	18.14%	8.62%	IMPLAUSIBLE
18	Doesn't remember bodyguard	4	36.84%	20.00%	AMBIGUOUS
19	Epstein "no role" New Albany	45	4.94%	2.17%	CONTRADICTED BY STATE RECORDS
20	"Gang stuff" = confidential	17	12.07%	5.56%	CONTRADICTED BY EPSTEIN'S OWN WORDS
21	Doesn't know compensation	85	2.67%	1.16%	STRUCTURALLY IMPOSSIBLE
22	"No recollection" of 3 properties	180	1.29%	0.55%	DEFINITIVELY CONTRADICTED
23	Failure to report $100M theft	45	4.94%	2.17%	CONTRADICTED BY BEHAVIOR
24	"Con man" defense (conjunction)	ALL	< 10⁻³⁵	< 10⁻⁴⁰	STRUCTURALLY IMPOSSIBLE

Claim Strength Distribution

Category	Count	Claims
Definitively Contradicted (LR > 100)	8	1, 2, 6, 10, 13, 14, 22, 3
Contradicted by Records (LR 10-100)	10	4, 8, 9, 12, 15, 16, 19, 20, 21, 23
Implausible/Ambiguous (LR 3-15)	5	5, 7, 11, 17, 18

Combined Metrics

Metric	Value	Comparison
Combined LR (23 claims)	1.1 × 10³⁹	1.1 undecillion to 1 against truthfulness
Fisher's p-value (23 claims)	< 10⁻¹²	Exceeds CERN threshold by 5 orders of magnitude
Joint probability (independent)	1.18 × 10⁻³⁵	1 in 850 decillion
Joint probability (7 clusters)	7.49 × 10⁻⁶	1 in 133,000
Joint probability (3 mega-clusters)	1.95 × 10⁻²	1 in 51 (maximum correlation)
Stressed LR (all estimates ×10 generous)	2.3 × 10¹⁷	Still 230 quadrillion to 1
With 3 strongest claims removed	3.6 × 10⁻²⁴	Still 1 in 280 sextillion
With HALF claims removed (12 weakest)	8.7 × 10⁻¹⁰	Still 1 in 1.15 billion
Binomial ignorance test	Z = 5.87	Independent confirmation
Document suppression	Z = -29 to -155	Independent confirmation
Three-system combined p-value	< 10⁻²⁰⁴	All three independent systems agree

---

XII. WHAT THE MATH REVEALS

The Five Verdicts

1. DEFINITIVELY CONTRADICTED (Claims 1, 2, 6, 10, 14, 22) ... 6 claims

These claims are contradicted by Wexner's own documents, his own handwriting, his own SEC filings, FBI intake sheets, and events at his own property. The evidence is not circumstantial. It is documentary. No reasonable interpretation supports truthfulness. Combined LR for this group alone: 3,600 × 850 × 950 × 180 × 170 × 180 = 1.6 × 10¹⁶.

2. STRUCTURALLY IMPOSSIBLE (Claims 3, 8, 9, 13, 21) ... 5 claims

These claims require Wexner to have been unaware of activities documented in his own accounts, at institutions bearing his name, using his company's products, or to have "severed" ties that demonstrably continued for 5+ years through trust structures and property transfers. The governance and paper trail make ignorance legally unavailable as a defense.

3. CONTRADICTED BY RECORDS (Claims 12, 15, 16, 19, 23) ... 5 claims

These claims are contradicted by specific records: his own lawyers' DOJ representations, property deeds, Ohio Secretary of State filings, IRS foundation governance, and the behavioral impossibility of a victim never reporting $100M+ theft. Each involves documented facts that cannot be explained away.

4. CONTRADICTED BY PATTERN (Claims 4, 17, 20) ... 3 claims

These claims require accepting implausible characterizations: unlimited POA as "normal," Virginia Giuffre as unknown, and 15 years of spouse-hidden secrets as "confidential conversations." They are not individually dispositive but contribute meaningfully to the pattern of systematic deception.

5. TECHNICALLY POSSIBLE (Claims 5, 7, 11, 18) ... 4 claims

These are the weakest claims. The Rothschild reference could be innocent (LR=5). The FBI "never contacted" may be literally true in the narrowest sense (LR=3). "Never spoke again" is contradicted by a 2008 email but defensible as literal (LR=6). The bodyguard memory failure over 35 years is plausible (LR=4). Even these 4 contribute a combined LR of 360 to the overall analysis.

The Structural Conclusion

The "con man" defense is a conjunction. It requires a self-described micromanager who simultaneously:

• Signed SEC filings listing Epstein as trustee... but didn't know about $1.3B in stock sales
• Was a foundation board member... but didn't know about $21M in grants or $2.3M to a former PM
• Gave unlimited POA over 20 companies and 19 trusts... but thought this was normal
• Wrote "your friend Leslie" with a breast drawing... but wasn't friends
• Received warnings about VS scout impersonation... twice... but didn't connect them
• Had $75K/year billed to his accounts for an OB-GYN at his hospital... but had "not a clue"
• Never noticed his mansion was transferred for $0... or 3 other properties
• Was contradicted by 6 independent sources on fidelity... but was "never unfaithful"
• "Severed" in September 2007... but a new trust was established 2 months later
• Had a kidnapping at his estate perpetrated by people claiming his authority... but "knows her from the press"
• Paid $200M+ to his sole financial advisor... but can't recall the fee structure
• Lost $100M+ to "theft"... but never filed a police report or lawsuit

Each claim is individually unlikely. Together they are impossible. The math makes this rigorous: the probability ranges from 1 in 51 (maximum 3-mega-cluster correlation) to 1 in 850 decillion (independence), depending on assumptions. Under no assumption, no prior, and no correlation model does the probability of truthfulness reach even 2%.

The con man defense fails not because any single lie is detected, but because the conjunction of ALL 23 required claims is probabilistically impossible. You cannot be a micromanager and a dupe simultaneously across 23 independent dimensions of evidence.

---

XIII. ACADEMIC REFERENCES

1. Aitken, C.G.G. & Taroni, F. (2004). Statistics and the Evaluation of Evidence for Forensic Scientists. 2nd Ed. Wiley.

2. Evett, I.W. & Weir, B.S. (1998). Interpreting DNA Evidence. Sinauer Associates.

3. Fisher, R.A. (1925). Statistical Methods for Research Workers. Oliver & Boyd.

4. Griffiths, T.L. & Tenenbaum, J.B. (2007). "From mere coincidences to meaningful discoveries." Cognition, 103(2), 180-226.

5. Kaye, D.H. (1999). "Clarifying the Burden of Persuasion." Law, Probability and Risk, 1(1), 1-17.

6. Lindley, D.V. (1977). "A Problem in Forensic Science." Biometrika, 64(2), 207-213.

7. Lindley, D.V. (2006). Understanding Uncertainty. Wiley.

8. Neyman, J. & Pearson, E.S. (1933). "On the Problem of the Most Efficient Tests of Statistical Hypotheses." Phil. Trans. Royal Soc., 231, 289-337.

9. Robertson, B. & Vignaux, G.A. (1995). Interpreting Evidence: Evaluating Forensic Science in the Courtroom. Wiley.

10. Tillers, P. & Green, E.D. (1988). Probability and Inference in the Law of Evidence. Kluwer Academic.

11. Tversky, A. & Kahneman, D. (1983). "Extensional vs. Intuitive Reasoning: The Conjunction Fallacy." Psychological Review, 90(4), 293-315.

12. Vrij, A. (2008). Detecting Lies and Deceit: Pitfalls and Opportunities. 2nd Ed. Wiley.

13. Drezner, Z. & Wesolowsky, G.O. (1990). "On the computation of the bivariate normal integral." J. Stat. Comp. Sim., 35(1-2), 101-107.

---

XIV. APPENDIX A: FORMULA REFERENCE

Symbol	Name	Formula	Section
P(H\	E)	Posterior probability	P(E\	H)P(H) / P(E)	I.A
LR	Likelihood Ratio	P(E\	H_D) / P(E\	H_T)	I.B
χ²	Fisher's statistic	-2Σln(pᵢ)	I.C
P(All)	Joint probability	ΠP(Cᵢ)	I.D
k_eff	Effective independent tests	k / (1 + (k-1)r̄)	IV.B
Z	Standard score	(X - μ) / σ	IX

APPENDIX B: EVIDENCE SOURCE MAP

Claim	Primary Vault Document	Key Data Point
1	9_East_71st_Street_Property_Chain.md	$0 transfer deed, 3 mutually exclusive accounts
2	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.1	13D filings signed by Wexner, $1.3B pipeline
3	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part VI	COUQ Foundation $21M, IRS governance requirements
4	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part III	20 companies, 19 trusts, 16-year unlimited POA
5	DOJ_Epstein_Pattern_Analysis_CRITICAL_FINDINGS.md	$35M Rothschild-Epstein flows post-2008
6	Wexner_Deposition_Forensic_Extraction.md §4	Birthday letter "your friend Leslie," island visits
7	LESLIE_WEXNER_PROSECUTORIAL_BRIEF.md §B	Debevoise proffer July 25, 2019
8	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	VS scout 1993 + 1997, lingerie grooming kits
9	Wexner_Statement_Forensic_Analysis.md	Landon $25K quarterly, "as we have in prior years"
10	LESLIE_WEXNER_PROSECUTORIAL_BRIEF.md §C-D	1 victim, 4 SDNY sheets, Steve Scully, FBI PowerPoint
11	Wexner_Deposition_Forensic_Extraction.md §9.7	2008 email Exhibit J, insurance trust through 2012
12	Deposition [03:03:06]	Own lawyers told DOJ "deeply discounted prices"
13	Wexner_Statement_Forensic_Analysis.md	Insurance Trust #3 est. Nov 2007, 5 post-severance links
14	LESLIE_WEXNER_PROSECUTORIAL_BRIEF.md §E	Kidnapping at estate, "we work for WEXNER"
15	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part V	MC2 models in Wexner apartments, ages 13+
16	LESLIE_WEXNER_PROSECUTORIAL_BRIEF.md §IV	FBI report: $2.3M, "Epstein signed documents"
17	EPSTEIN_TRAFFICKING_DEEP_DIVE.md	Giuffre abuse alleged at 9 East 71st (Wexner mansion)
18	Deposition [03:32:21]	Adrian saw young girls at Epstein's Palm Beach
19	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.2	Ohio SOS: Epstein = President, Maxwell confirmed
20	Deposition Exhibits G + H	"Gang stuff for over 15 years," hidden from Abigail
21	LBRANDS_SEC_RED_FLAG_AUDIT.md §1.5	$200M+ fees, 3-7x market rate, no other clients
22	WEXNER_FORENSIC_FINANCIAL_ANALYSIS.md Part IV	3+ properties transferred, used for victims/kidnapping
23	LBRANDS_SEC_RED_FLAG_AUDIT.md §5.5	No police report, no lawsuit, 4.6% "donation" recovery
24	All 23 above (conjunction test)	Joint probability 10⁻³⁵ to 10⁻²

APPENDIX C: SENSITIVITY RESULTS (FULL TABLE)

Scenario	P(All True)	Odds Against	Exceeds Preponderance?	Exceeds BRD?
Generous + Independent (23 claims)	1.18 × 10⁻³⁵	1:850 decillion	YES	YES
Generous + 7 Clusters	7.49 × 10⁻⁶	1:133K	YES	YES
Generous + 3 Mega-Clusters	1.95 × 10⁻²	1:51	YES	YES
Generous + Stressed (×10, all 23)	~10⁻¹⁷	1:230 quadrillion	YES	YES
Generous + 3 Strongest Removed (20 left)	3.6 × 10⁻²⁴	1:280 sextillion	YES	YES
Generous + HALF Removed (12 weakest)	8.7 × 10⁻¹⁰	1:1.15B	YES	YES
Neutral + Independent	2.45 × 10⁻⁴⁰	1:4.1 × 10³⁹	YES	YES
Neutral + 7 Clusters	5.42 × 10⁻⁸	1:18.5M	YES	YES
Neutral + 3 Mega-Clusters	2.78 × 10⁻³	1:360	YES	YES
Skeptical + Independent	1.15 × 10⁻⁴⁸	1:8.7 × 10⁴⁷	YES	YES
Skeptical + 7 Clusters	1.74 × 10⁻¹⁰	1:5.7B	YES	YES
Skeptical + 3 Mega-Clusters	2.24 × 10⁻⁴	1:4,470	YES	YES

Under ALL 12 scenarios tested, the conclusion exceeds beyond reasonable doubt.

Scenario Comparison: 9-Claim vs 23-Claim Analysis

Metric	9-Claim (Original)	23-Claim (Expanded)	Improvement
Combined LR	~10¹⁷	1.1 × 10³⁹	+22 orders of magnitude
Fisher's p	< 10⁻⁷	< 10⁻¹²	+5 orders of magnitude
Joint P (Generous, independent)	~10⁻¹⁵	10⁻³⁵	+20 orders of magnitude
Prior required for 50% posterior	99.99999999999999960%	99.999...999% (39 nines)	+23 decimal places
Weakest scenario (max correlation)	1 in 41,000	1 in 51	Less favorable to BARSS*

*The 3-mega-cluster model is less favorable because it collapses more claims into fewer independent tests. This is the honest trade-off of adding correlated evidence. Even so, P(Deceptive) = 98.05% under maximum correlation with generous priors... still exceeding beyond reasonable doubt.

---

Compiled: 2026-02-20

Framework: Bayesian Forensic Inference, Fisher's Combined Test, Joint Probability

Data: BARSS Vault (119+ files), DOJ Epstein Corpus (14,993 documents), Deposition Transcript (495 Q&A pairs)

All calculations shown. All assumptions stated. All sensitivity tested.

This document contains no editorializing. The math speaks.