Private-Bank Client Defections During Regime Shifts Form a POT Process; Retention Exceedances Converge to GPD_{xi,beta} — Advisor Churn-Resistance is a Measurable xi-Attenuation Coefficient
A math tool for predicting financial disasters could reveal which wealth advisors actually stop rich clients from leaving.
Pickands-Balkema-de Haan theorem maps client-defection AUM exceedances above an advisor-book threshold u_a to a GPD with advisor-specific tail index xi_a; advisor value is formalized as Delta xi_a = xi^{baseline} - xi^{post-intervention}, quantifying measurable conversion of a heavy-tailed (Frechet) defection regime toward Gumbel.
7 bridge concepts›
How this score is calculated ›How this score is calculated ▾
6-Dimension Weighted Scoring
Each hypothesis is scored across 6 dimensions by the Ranker agent, then verified by a 10-point Quality Gate rubric. A +0.5 bonus applies for hypotheses crossing 2+ disciplinary boundaries.
Is the connection unexplored in existing literature?
How concrete and detailed is the proposed mechanism?
How far apart are the connected disciplines?
Can this be verified with existing methods and data?
If true, how much would this change our understanding?
Are claims supported by retrievable published evidence?
Composite = weighted average of all 6 dimensions. Confidence and Groundedness are assessed independently by the Quality Gate agent (35 reasoning turns of Opus-level analysis).
RQuality Gate Rubric
3/10 PASS · 7 CONDITIONAL
| Criterion | Result |
|---|---|
| Test Protocol | 7 |
| Novelty | 9 |
| Mechanism | 8 |
| Regulatory Accuracy | 10 |
| Confidence | 7 |
| Translational Utility | 7 |
| Falsifiable | 8 |
| Groundedness Per Claim | 7 |
| Mathematical Correctness | 9 |
| Counter Evidence Considered | 8 |
Claim Verification
Empirical Evidence
How EES is calculated ›How EES is calculated ▾
The Empirical Evidence Score measures independent real-world signals that converge with a hypothesis — not cited by the pipeline, but discovered through separate search.
Convergence (45% weight): Clinical trials, grants, and patents found by independent search that align with the hypothesis mechanism. Strong = direct mechanism match.
Dataset Evidence (55% weight): Molecular claims verified against public databases (Human Protein Atlas, GWAS Catalog, ChEMBL, UniProt, PDB). Confirmed = data matches the claim.
Extreme Value Theory (EVT) is the branch of statistics that deals with rare, catastrophic events — think: 100-year floods, market crashes, insurance mega-claims. It has a powerful theorem (called the Pickands-Balkema-de Haan theorem) that says: if you look only at the *worst* outcomes beyond some high threshold, their distribution reliably follows a specific mathematical shape called the Generalized Pareto Distribution. The key parameter of that shape, called 'xi' (pronounced 'ksee'), tells you how heavy the tail is — in plain terms, how likely truly extreme losses are compared to merely bad ones. This hypothesis proposes applying that framework to something surprisingly human: wealthy clients fleeing their financial advisors during turbulent times — market meltdowns, political upheaval, bank scandals. The idea is that the 'assets under management' lost when clients defect above some alarm threshold follows exactly that kind of extreme-value distribution, and crucially, the tail-heaviness (xi) varies *by advisor*. A skilled advisor, through trust-building and crisis communication, doesn't just reduce average defections — they fundamentally reshape the statistical tail, pulling it from a 'heavy' Fréchet regime (where catastrophic losses are alarmingly probable) toward a lighter Gumbel regime (where extreme losses become genuinely rare). The hypothesis then defines a concrete, single-number measure of advisor quality: how much they shrink xi. Why does this matter? Because right now, 'advisor value' is mostly assessed through fuzzy metrics — client satisfaction scores, retention rates, gut feel. This framework would give banks a mathematically grounded way to quantify which advisors are genuinely protecting the firm from tail-risk client exodus events, not just performing well on average. It also reframes a well-known industry observation — that a small fraction of relationship managers account for most client losses — not as a quirky rule-of-thumb, but as a signature of a heavy-tailed distribution across advisors themselves.
This is an AI-generated summary. Read the full mechanism below for technical detail.
Why This Matters
If confirmed, this hypothesis could give private banks and wealth management firms a rigorous, actuarial-style tool to identify and retain their most valuable advisors — not by gut instinct, but by a measurable tail-risk coefficient derived from real client-departure data. It could reshape how bonuses, promotions, and succession planning are structured, prioritizing advisors who demonstrably reduce catastrophic AUM-flight risk rather than just growing book size. The framework could also improve stress-testing models: regulators and risk managers could incorporate advisor-quality distributions into scenario analyses for systemic wealth outflows during crises. It's worth testing because even a partial validation — showing that xi estimates cluster differently across advisor cohorts — would mark the first time extreme value statistics have been formally applied to human relationship quality in finance.
Grounded claims cite published evidence. Parametric claims draw on general model knowledge. claims are explicitly flagged hypothetical leaps.
Mechanism
Let {D_k} denote client-defection events within an advisor book over [0,T] with associated AUM-at-risk S_k. Define advisor-specific threshold u_a (AUM-loss level triggering defection review). Model exceedances {S_k - u_a : S_k > u_a} as conditional excesses above u_a. By Pickands-Balkema-de Haan (Balkema-de Haan 1974 GROUNDED; Pickands 1975 GROUNDED), under standard maximum-domain-of-attraction conditions, P(S - u_a > y | S > u_a) → G_{xi_a, beta_a}(y) = (1 + xi_a y/beta_a)^{-1/xi_a} as u_a approaches the right endpoint of F_a. The advisor-specific xi_a is the formal object of advisory value. Define the churn-resistance coefficient Delta xi_a ≡ xi_a^{baseline} − xi_a^{post-intervention}, where baseline is estimated via contemporaneous benchmark advisors, the advisor's own pre-hire history, or cross-institutional ORX-style pooled baselines. An advisor with Delta xi_a > 0 measurably converts a heavy-tailed defection process (Frechet regime) into a less heavy one approaching Gumbel. By the ES-xi relationship ES_q/VaR_q → 1/(1-xi) as q → 1 (McNeil-Frey-Embrechts 2015 p.277 GROUNDED; Acerbi-Tasche 2002 GROUNDED), reducing xi from 0.30 to 0.15 produces ES reduction factor 0.8235 (17.6%), verified arithmetically by the Computational Validator. The PriceMetrix 2014 stylized fact — '50% of RMs account for 80% of lost clients' GROUNDED — is a symptomatic signature of a heavy-tailed distribution across advisors in xi_a-space, not previously estimated as such. This hypothesis predicts the cross-advisor distribution of xi_a has heavy tails itself.
Supporting Evidence
Pickands 1975 + Balkema-de Haan 1974 (POT/GPD convergence theorem); Hill 1975 (estimator); McNeil-Frey-Embrechts 2015 §5.2.4 (practical Hill guidance); Acerbi-Tasche 2002 (coherent ES); McKinsey-PriceMetrix 2014 (heavy-tail advisor attrition empirical); Computational Validator CV Check 5 (17.6% ES reduction arithmetically verified); de Fontnouvelle 2006 operational-risk LDA analog (CV Check 6).
How to Test
Internal CRM + AUM movement logs for N ≥ 100 advisors over T ≥ 5 years, covering at least two declared regime-shift windows (2020, 2022, 2023 for Italian context). Annual block structure (k=5 blocks for 2020-2024). Threshold u_a selected via Mean Excess Plot diagnostic at the linearity point (typical: 90th percentile of AUM outflow per advisor). Hill k-selection via stability plateau in k ∈ [5, floor(0.10·n_a)]. Bootstrap B=1000 with stratification by year, 95% BCa CI. Pool at book-cluster level (geography, team, AUM-decile) when individual-advisor n < 500. Primary acceptance: Spearman rank-correlation of xi_a between disjoint 2.5-year halves ≥ 0.4 (p<0.01); top-quartile vs bottom-quartile ES_{0.975} differential ≥ 15% during 2022-2024 window. Falsification: rank-correlation < 0.2 rejects; differential < 10% weakens. Identification strategy: DiD around regime-shift dates with matched advisor cohorts, leveraging Italian M&A-triggered reassignments as natural experiment.
Other hypotheses in this cluster
Basel III FRTB Standardized Approach Calibrated on Normal-Regime Windows Behaves Functionally as xi ≈ 0 Until Forced Recalibration: A Regime-Aware ES Correction Using Dynamic Hill Estimation Recovers Capital Underestimation
Bank risk models may underestimate crisis losses by 35%+ because they're blind to how extreme tail risk shifts during market turmoil.
Advisor Successions Are xi-Stable iff Post-Transition xi_c ≤ max(xi_{pre}, xi_{successor-baseline}) + ε: A Formal Criterion for Protocol-Quality in Private-Bank Advisor Turnover
A math formula could tell private banks whether an advisor handoff will cause clients to suffer outsized financial losses.
The Advisor xi-Ledger: Expected ES-Reduction Per Client-Year Achieved via xi-Attenuation — Integrating H1-H4 Into Private-Bank P&L Under FTG-Universality Accounting
A new accounting framework would measure wealth advisors' value by how much they reduce clients' worst-case financial losses.
Client Trust in Advisor = 1/xi_c: Trust as a Tail-Sensitivity Asset Priceable via EVT Expected Shortfall, Elicited via Percentile-Scale Subjective-Loss Questionnaires
A math formula from insurance risk modeling could turn client trust into a measurable, priceable financial asset.
Can you test this?
This hypothesis needs real scientists to validate or invalidate it. Both outcomes advance science.