What these numbers are — and what they are not
The nine probability inputs (P1 through P9) are structured legal judgements expressed as numbers. They are not statistical frequencies drawn from a large dataset of identical cases. There is no such dataset — there is only one Harrison v Aegon. What the numbers represent is this: if a competent legal analyst were to assess the probability that a court, having heard all the evidence, would find for the plaintiff on a specific discrete legal issue — that assessment, made explicit, documented, and calibrated against empirical anchor cases, is the input.
The justification for each number has four components, which are combined using a documented methodology:
Step 1 — Case-specific legal assessment (60% weight)
The facts, admissions, documents, and legal arguments specific to this proceeding.
Step 2 — Empirical calibration (40% weight)
14 verified HCA/NSWCA anchor cases scored for relevance.
Weighted mean for the relevant legal dimension anchors 40% of the input.
Step 3 — SAI and EFI uplift (applied after blend)
Structural Advantage Index (SAI = 91.25, compressed to 78.9) × 0.001 per input.
Environmental Factors Index (EFI) × 0.0005 per input (half weight).
Step 4 — Hard caps and stress reductions
P2 Deed capped at 0.92. P4 s.55 capped at 0.82. P3 EA capped at 0.88.
Eight deliberate score reductions applied to prevent artificial inflation.
What 18% residual probability against (e.g. on P3 at 82%) does not mean is that the plaintiff's lawyers are 18% likely to have the law wrong. It means: 18% allowance for judicial discretion, unexpected evidentiary rulings, cross-examination outcomes, arguments not yet anticipated, and the inherent unpredictability of any contested proceeding. The number is not a prediction. It is a calibrated, documented, stress-testable planning tool.
The consequence: any variation from a default value is a deliberate departure from the calibrated position. It requires a specific legal reason — new evidence, a counsel assessment, a material change in the defendant's position — not a preference for a better-looking output.
Navigate to each input
P1 — Korea rate reduction (90%)
P2 — Deed / summary judgment (89%)
P3 — EA breach methodology (82%)
P4 — s.55 postponement (80%)
P5 — Demand accrual / s.63 (85%)
P6 — CA stream (81%)
P7 — Component B (71%)
P8 — Route C / fresh fraud (78%)
P9 — Conduct / costs factor (78%)
Absolute worst case — joint failure analysis
P1 — 90%
Component 1 — The contractual position (binary)
The rate reduction required written amendment. None was executed. The contract is clear.
EA Clause 20 and GPR-TIMAP s.9.6 (the Double Modification Lock) both require written amendment signed by both parties for any change to the commission rate. The rate was reduced from 1.875% to 0.375% — an 80% cut — without any executed written amendment. There is no document that authorises the reduction on its face. This is the starting contractual position, and it is strong: the contractual argument in isolation would be assessed at approximately 85–90%.
EA Clause 20 — written amendment gate
GPR-TIMAP s.9.6 — Double Modification Lock
MOC 8 — original Korea rate 1.875%
Nine successive written agreements all at 1.875%
Component 2 — The five-position collapse (shifts probability upward)
Aegon advanced five successive justifications for the rate reduction across 19 years. Each was abandoned. Sequential collapse establishes consciousness of wrongdoing.
P1 (May 2006 Eubanks): Marketing Agreement authority — abandoned because the Marketing Agreement had expired June 2003, 2.5 years before this justification was advanced.
P1b (Jul 2006 / Oct 2008 Allens): A fictional LGI-TIMAP Agreement governs commissions — this agreement does not exist. Cited twice by Allens over 27 months. Never produced.
P2 (Jul 2021 inadvertent): Marketing Agreement was the actual basis — contradicted by the finance manager's own Excel formula disclosure (CC-5).
P2.5 (Aug 2024 Hallal — SMOKING GUN): "Marketing Agreement only relates to Australia — not Korea." This single admission destroys 19 years of justifications simultaneously. Every Korea calculation from 2005 was therefore unauthorised on Aegon's own concession.
P3 (Oct 2024 Hallal): Marketing Agreement irrelevant — MOC confers unilateral authority. This violates EA Clause 20 and was abandoned in 2003 when Harrison explicitly refused the rate reduction and Aegon accepted that refusal as decisive (CC-24).
P1b (Jul 2006 / Oct 2008 Allens): A fictional LGI-TIMAP Agreement governs commissions — this agreement does not exist. Cited twice by Allens over 27 months. Never produced.
P2 (Jul 2021 inadvertent): Marketing Agreement was the actual basis — contradicted by the finance manager's own Excel formula disclosure (CC-5).
P2.5 (Aug 2024 Hallal — SMOKING GUN): "Marketing Agreement only relates to Australia — not Korea." This single admission destroys 19 years of justifications simultaneously. Every Korea calculation from 2005 was therefore unauthorised on Aegon's own concession.
P3 (Oct 2024 Hallal): Marketing Agreement irrelevant — MOC confers unilateral authority. This violates EA Clause 20 and was abandoned in 2003 when Harrison explicitly refused the rate reduction and Aegon accepted that refusal as decisive (CC-24).
CC-74: Marketing Agreement abandoned Oct 2024
Hallal Aug 2024: Marketing Agreement = Australia only
CC-24: Harrison refused rate reduction Jul 2003 — Aegon accepted
CC-76A Apr 2025: Settlement offer restoring 1.875% — tacit admission
Component 3 — The nine-agreement pattern
Two structurally different calculation architectures across two countries and nine agreements all converge on 1.875%. This cannot be a coincidence of arithmetic.
Korea arrived at 1.875% through an indirect three-variable chain: 69.45% × 3.0% GPR-TIMAP revenue × 90% Reinsured Risk = 1.875%. Japan arrived at 1.875% through a direct single-step statement: 1.875% of gross premium received. The fact that two entirely different architectural routes reach identical results establishes that 1.875% was the agreed target, not an incidental arithmetic outcome. Nine consecutive written agreements across two countries, two insurers, and two calculation architectures all produce the same number.
MOC 8: Korea 1.875% (indirect architecture)
MOCs 10, 11, 14, 16: Japan 1.875% (direct statement)
Two-architecture significance analysis: Doc 2.6 §A.2
Component 4 — Stress-test and residual uncertainty (10%)
What the 10% residual allows for
The 10% is not a probability that the contract argument fails. It is an allowance for: (a) trial unpredictability — a judge's reading of credibility evidence may differ from the documentary record; (b) the possibility that Aegon produces a document or witness not yet anticipated; (c) the inherent uncertainty in any contested proceeding regardless of evidence strength; and (d) quantum disputes even if liability is established. The Korea claim is assessed as the strongest of the three streams on the merits alone.
△
Why 90% not higher: The documentary record is extraordinary but the claim goes to trial, not summary determination. Trial introduces judge-specific unpredictability, the risk of unexpected cross-examination outcomes, and the possibility of credibility findings that do not precisely track the documents. 90% is the appropriate calibrated ceiling for a matter this strong that nonetheless requires a contested trial. The Hallal August 2024 admission is the single most important piece of evidence — it alone brings the residual probability of Aegon successfully defending the Korea rate reduction close to zero on the merits, but judicial uncertainty prevents 100%.
P2 — 89%
Component 1 — The demand and default (near-certain facts)
The demand was made 25 November 2025. The compliance deadline was 2 December 2025. Default is now over 95 days confirmed. These are not contested.
The Deed of Guarantee and Undertaking (1998) imposes an unconditional obligation on the guarantor. Demand was served 25 November 2025 with a compliance deadline of 2 December 2025 under Clause 2.1. The defendant did not comply. The facts of demand, deadline, and non-compliance are documentary and uncontested. On an unconditional demand guarantee with proven non-compliance, the merits of the underlying claim are irrelevant to the guarantee liability — the guarantor is bound to pay on demand.
Deed demand served 25 Nov 2025 — Exhibit KAH-8
Compliance deadline 2 Dec 2025 — passed
Deed Clause 2.1: unconditional payment obligation
Sunbird Plaza v Maloney (1988) 166 CLR 245
Component 2 — The TLIC succession chain
JCPLIG → JCPLIC → TLIC. The successor guarantor is identified, Iowa-domiciled, and regulated.
The original guarantor JCPLIG (J.C. Penney Insurance Group, Inc.) passed the guarantee obligation through to JCPLIC (JCPenney Life Insurance Company) and then to TLIC (Transamerica Life Insurance Company, NAIC 69078, Iowa-domiciled). The TLIC assumption letter confirms the succession. The succession chain is established in the bundle (06-05, 06-14 v3). Corrs is verifying the precise corporate succession documentation (CP items) — this is the primary remaining uncertainty in P2 and the reason the hard cap is set at 0.92 rather than higher.
06-05: Guarantor Succession Chain
06-14 v3: Deed succession — "supported by indemnity" — CP pending Corrs verification
TLIC NAIC 69078 — Iowa Insurance Division regulated
Component 3 — Self-drafting and unified drafter estoppel
Allens drafted the Deed, the EA, and the Marketing Agreement as a single integrated suite. Contra proferentem applies at maximum force.
Issue A (self-drafting): the guarantor entities (TLIC/Aegon as 100% owner of TIMAP) directed Allens to draft all three instruments. A sophisticated institutional drafter cannot advance ambiguity arguments in enforcement proceedings. Issue B (unified drafter): Allens drafted the EA, all six amendments, the Marketing Agreement, and the Deed simultaneously as an integrated commercial suite. Issue C (physical annexation): the EA as amended is physically annexed to the Deed as Schedule 1 (Recital B). The EA's Clause 20 written-amendment requirement — which Aegon violated — was in the instrument TLIC guaranteed. TLIC cannot claim ignorance of a requirement that was in Schedule 1 of the instrument it executed.
Issues A, B, C — Addendum to Cover Memorandum Mar 2026
Johnston Cover Letter 1 Feb 2001 — exhaustive suite confirmed
Contra proferentem — maximum force
Component 4 — s.54 acknowledgment (resets limitation independently)
The February 2024 corrective payment of $3,319,108 is a s.54 Limitation Act acknowledgment with 99.9% certainty (D3.70).
D3.70 (2 August 2024) establishes the s.54 acknowledgment of the debt with 99.9% documentary certainty. The $3,319,108 payment was expressly described as acknowledgment of "material accounting irregularities" — it is an admission, not a settlement. Under Foakes v Beer, payment of a lesser sum is not satisfaction of a greater debt absent fresh consideration. The payment restarts the limitation clock for the full balance independently of all other limitation arguments.
D3.70 — 2 Aug 2024 — s.54 acknowledgment (99.9% certainty)
$3,319,108 corrective payment Feb 2024
s.54 Limitation Act 1969 (NSW)
Foakes v Beer [1884] — no satisfaction without consideration
△
Why 89% not higher: Hard-capped at 92% per Spencer v Commonwealth summary judgment caution — courts apply a higher threshold for summary determination on matters of this complexity and value. The Corrs-pending CP items on the succession chain verification introduce a small residual uncertainty. Once Corrs confirms the succession documentation, the base assessment for P2 should move to the cap of 0.92. The Deed ultimate probability (L2 OR L5) is approximately 98.4% at defaults — either route succeeding is sufficient.
P3 — 82%
Component 1 — The contractual obligation (plain meaning)
The contract says "actual reinsurance premiums received." Aegon used estimates. The construction argument is strong.
GPR-TIMAP Agreement Clause 2.1(a) specifies "Reinsurance Premium received by Global Premier." The EA Schedule requires that commission be calculated on the actual reinsurance premium base. The construction authorities direct courts to the ordinary commercial meaning of the words used: Mount Bruce Mining v Wright Prospecting [2015] HCA 37 and Electricity Generation Corporation v Woodside Energy [2014] HCA 7 both confirm that commercial contracts are read according to the meaning a reasonable business person would give the words. A reasonable business person reading "reinsurance premiums received" understands it to mean actual premiums received — not a formula-driven estimate. This contractual argument in isolation: approximately 85%.
GPR-TIMAP Clause 2.1(a): "Reinsurance Premium received"
Mount Bruce Mining v Wright Prospecting [2015] HCA 37
Electricity Generation Corp v Woodside Energy [2014] HCA 7
Component 2 — The admissions (binding)
CC-13 confirms estimation commenced July 2005. CC-8 and CC-9 confirm Aegon cannot retrieve the basis for its own estimates. The $3,319,108 payment is a binding admission the prior calculation was wrong.
CC-13 (Mar 2023): Controlling admission — permanent substitution of estimates commenced July 2005. This is the pivot date for the entire claim.
CC-6 (Oct 2021) and CC-7 (Dec 2021): Progressive retraction of the cessation date (late 2007 → October 2007 → July 2005), each retraction an implicit acknowledgment the prior stated date was false — direct evidence of concealment.
CC-8 (Feb 2022) and CC-9 (Mar 2022): Aegon admits it cannot retrieve the basis or documentation for its estimation rates. A party that cannot document the basis of its calculation methodology cannot defend that methodology as contractually authorised.
$3,319,108 corrective payment (Feb/Mar 2024): Paid expressly as acknowledgment of "material accounting irregularities." You cannot make a corrective payment without implicitly admitting the prior calculation was wrong. The payment is an admission that cannot be walked back.
CC-6 (Oct 2021) and CC-7 (Dec 2021): Progressive retraction of the cessation date (late 2007 → October 2007 → July 2005), each retraction an implicit acknowledgment the prior stated date was false — direct evidence of concealment.
CC-8 (Feb 2022) and CC-9 (Mar 2022): Aegon admits it cannot retrieve the basis or documentation for its estimation rates. A party that cannot document the basis of its calculation methodology cannot defend that methodology as contractually authorised.
$3,319,108 corrective payment (Feb/Mar 2024): Paid expressly as acknowledgment of "material accounting irregularities." You cannot make a corrective payment without implicitly admitting the prior calculation was wrong. The payment is an admission that cannot be walked back.
CC-13: July 2005 substitution date confirmed
CC-8/CC-9: Aegon cannot retrieve estimation basis
$3,319,108 corrective payment Feb 2024
D3.70: actuals available at reinsurer level, withheld from commission team
Component 3 — Two Sets of Books (TAB_38_3 — strategic reserve)
The correct commission ran in the Lumley Ingenium system for 18 years. The "Comm to KH" column proves Aegon always knew the correct figure.
The Two Sets of Books forensic finding (TAB_38_3, held in strategic reserve and not yet presented to Aegon) establishes that the correct commission calculation ran in the St George/Lumley Ingenium system continuously for 18 years while manual reports paid the lower amount. The "Comm to KH" column in the system proves the correct figure was always known and calculated correctly internally. This evidence, when deployed, reduces P3 further toward certainty. It is the single most powerful merits evidence in the entire bundle and is held for the maximum tactical moment.
TAB_38_3 — STRATEGIC HOLD — not yet presented to Aegon
"Comm to KH" column — 18-year dual record system
Six documented fraudulent statements: D.2.67, D.2.83b, D.2.57, D.2.76b, D.2.85
Component 4 — Empirical calibration (40% weight) and 60/40 blend
The empirical anchor cases produce a weighted mean of approximately 0.77–0.80 for this type of commercial calculation methodology dispute.
14 verified HCA/NSWCA anchor cases scored across the Contract/Construction dimension produce a weighted mean of approximately 0.78. Applying the 60/40 blend: (0.85 × 0.60) + (0.78 × 0.40) = 0.510 + 0.312 = 0.822. The SAI/EFI uplift adds approximately +0.003. The hard cap of 0.88 is not reached. Result: 0.82. The 18% residual is the honest allowance for the defendant's viable innocent explanation: that the estimation methodology was commercially reasonable, practically necessary, and implicitly accepted over many years. That argument is weak given CC-8 and CC-9 — but it will be made at trial.
△
When P3 would move: Upward to ~0.88 (the hard cap) if: (a) TAB_38_3 is deployed and the Two Sets of Books evidence is admitted; or (b) a court production order produces the actual premium bordereaux proving the quantum. Downward if Aegon produces evidence of a genuine contemporaneous agreement authorising estimates (no such document has been produced in 19 years of requests — the probability of its emergence is very low).
P4 — 80%
Component 1 — Five concealment pillars
s.55 requires that the plaintiff could not, with reasonable diligence, have discovered the right of action. Five independent pillars establish this.
Pillar 1 — Data monopoly and systematic refusal: Aegon held all actual premium bordereaux. Six documented refusals to produce actual data 2020–2024. Harrison could not calculate his correct entitlement without data only Aegon possessed.
Pillar 2 — Active misdirection through eight contradictory explanations (KNOW→UNKNOW→KNOW sequence): Aegon knew the correct calculation (Two Sets of Books), claimed not to know the basis (CC-8/CC-9), then made admissions confirming it had always known (D3.70). The KNOW→UNKNOW→KNOW pattern is direct evidence of conscious concealment.
Pillar 3 — NDA weaponisation: Audit NDAs imposed at each verification attempt containing (a) mandatory destruction clauses, (b) conditions requiring Harrison to surrender fraud claims, and (c) penalty provisions deterring challenge to Aegon's disclosed figures. Clause 6.7(f) of the EA contains no provision permitting NDA conditions as a precondition to verification — the NDA conditions were therefore themselves a breach of the verification obligation and a tool of concealment.
Pillar 4 — Document destruction: D3.05 admission of document destruction after request.
Pillar 5 — 24-hour premeditation: D3.70 (2 August 2024) proves the suppression was deliberate and planned — "actuals were available at Reinsurer level... not made available to commission team." Deliberate suppression by senior management.
Pillar 2 — Active misdirection through eight contradictory explanations (KNOW→UNKNOW→KNOW sequence): Aegon knew the correct calculation (Two Sets of Books), claimed not to know the basis (CC-8/CC-9), then made admissions confirming it had always known (D3.70). The KNOW→UNKNOW→KNOW pattern is direct evidence of conscious concealment.
Pillar 3 — NDA weaponisation: Audit NDAs imposed at each verification attempt containing (a) mandatory destruction clauses, (b) conditions requiring Harrison to surrender fraud claims, and (c) penalty provisions deterring challenge to Aegon's disclosed figures. Clause 6.7(f) of the EA contains no provision permitting NDA conditions as a precondition to verification — the NDA conditions were therefore themselves a breach of the verification obligation and a tool of concealment.
Pillar 4 — Document destruction: D3.05 admission of document destruction after request.
Pillar 5 — 24-hour premeditation: D3.70 (2 August 2024) proves the suppression was deliberate and planned — "actuals were available at Reinsurer level... not made available to commission team." Deliberate suppression by senior management.
D3.70: deliberate data suppression — 2 Aug 2024
CC-8/CC-9: cannot retrieve estimation basis
NDA conditions: three illegal prerequisites — Clause 6.7 has no such conditions
D3.05: document destruction after request
s.55 Limitation Act 1969 (NSW)
Component 2 — Progressive retraction as concealment indicator
CC-6 → CC-7 → CC-13: the cessation date was retracted three times. Each retraction is an implicit acknowledgment the prior stated date was false.
CC-6 (Oct 2021): estimates stopped "late 2007." CC-7 (Dec 2021): refined to "October 2007." CC-13 (Mar 2023): controlling admission — "July 2005." Each retraction is a direct acknowledgment that the earlier statement was false — and that Aegon knew the truth at the time it made the false statement. A party that knew the true cessation date was July 2005 but told Harrison it was late 2007 was actively concealing the full scope of the breach. Under s.55, the limitation clock does not start until the plaintiff could, with reasonable diligence, have discovered the concealed fact. The progressive retraction evidence establishes that discovery was structurally impossible without Aegon's disclosure.
Component 3 — Legal standard and hard cap
NSW courts apply s.55 cautiously. "Fraudulent" in s.55 may require more than conscious concealment in some readings. Hard cap at 82% reflects judicial caution.
The primary legal question is whether "fraudulently concealed" in s.55 requires dishonest intent or merely conscious concealment of a fact that the concealing party knows would give rise to a right of action. NSW cases support the broader construction (conscious concealment sufficient). However, some judicial readings require positive dishonest intent. The Briginshaw standard (civil fraud allegations must be proved to a high degree of satisfaction commensurate with the seriousness of the finding) applies to any concealment finding. The hard cap at 0.82 reflects this judicial caution. The base value was then reduced from 0.82 to 0.80 per the empirical dataset: the Limitation/Timing weighted mean across 14 anchor cases = 0.624, producing a 60/40 blend of (case-specific 0.87 × 0.60) + (empirical 0.624 × 0.40) = 0.522 + 0.250 = 0.772, lifted by SAI/EFI to approximately 0.80.
s.55 Limitation Act 1969 (NSW)
Briginshaw v Briginshaw (1938) 60 CLR 336
Great Northern Developments v Lane [2021] NSWCA 150
s.56 mistake — available as alternative if fraud not found
△
Note: Even if P4 fails entirely, the Deed stream (P2/P5) and Korea stream (P1) operate independently. P4 is not the only limitation route — Route A (12-year Deed specialty period to 2037), Route D (s.54 acknowledgment Feb 2024), and Route E (s.56 mistake) all operate independently of s.55. P4 is load-bearing for the full historical EA quantum but not for the overall recovery.
P5 — 85%
Component 1 — Demand-based accrual (established High Court authority)
Where a guarantee provides for payment "on demand," the cause of action accrues when demand is made. This is not a device. It is settled law.
Wardley Australia Ltd v Western Australia (1992) 175 CLR 514 establishes that a cause of action on an indemnity does not accrue until the plaintiff suffers loss. Sunbird Plaza Pty Ltd v Maloney (1988) 166 CLR 245 confirms that on a demand guarantee, the cause of action accrues when demand is made and not complied with. Bank of Western Australia v Abdul [2012] NSWCA 332 applies demand-based accrual directly to guarantee obligations. The demand was made 25 November 2025. The Deed cause of action accrued on that date. The 12-year specialty limitation period (Route A) runs to 25 November 2037. No limitation defence is available to Aegon on the Deed claim on any analysis.
Wardley Australia v Western Australia (1992) 175 CLR 514
Sunbird Plaza v Maloney (1988) 166 CLR 245
Bank of WA v Abdul [2012] NSWCA 332
Component 2 — Three-limb s.63 rebuttal
s.63 extinguishment requires that the principal obligation be time-barred. The Deed survives through three independent routes even if s.63 is pressed.
Limb 1 (primary indemnity): Deed Clause 2.1 second limb reads: "shall indemnify Harrison against any loss, expense or liability." This is primary indemnity language — not merely a guarantee. Ankar v National Westminster (1987) and Bank of WA v Abdul confirm that a primary indemnity operates independently of the principal obligation and is not subject to s.63 accessory extinguishment.
Limb 2 (demand accrual sequencing): The Deed cause of action did not accrue until the November 2025 demand. A right that had not yet accrued cannot be extinguished by s.63. Wardley Australia sequencing confirms this.
Limb 3 (s.54 acknowledgment): The February 2024 corrective payment constitutes a s.54 acknowledgment that independently resets the limitation clock for any remaining Deed obligation. Even on the most hostile limitation reading, s.54 protects the Deed claim.
Limb 2 (demand accrual sequencing): The Deed cause of action did not accrue until the November 2025 demand. A right that had not yet accrued cannot be extinguished by s.63. Wardley Australia sequencing confirms this.
Limb 3 (s.54 acknowledgment): The February 2024 corrective payment constitutes a s.54 acknowledgment that independently resets the limitation clock for any remaining Deed obligation. Even on the most hostile limitation reading, s.54 protects the Deed claim.
Ankar v National Westminster Finance (1987) 162 CLR 549
s.63 Limitation Act 1969 (NSW) — three-limb rebuttal
s.16 Limitation Act 1969 (NSW) — 12-year specialty period
s.54 acknowledgment: Feb 2024 payment resets clock
△
Why 85% not higher: The 15% residual reflects the possibility of an unexpected judicial treatment of the s.63 issue — specifically, a court that characterises the Deed as purely accessory and finds one of the three rebuttal limbs insufficient. This is considered unlikely given Ankar and Sunbird Plaza but the question has not been definitively resolved in NSW in these precise circumstances. Note that L2 (P2) and L5 (P5) are combined through OR-gate logic — Deed ultimate probability is approximately 98.4% at defaults.
P6 — 81%
Component 1 — Same methodology breach, different instrument
The CA applies the same estimation substitution that breached the EA. The contractual obligation is identical in structure. The same admissions bind both streams.
The Consulting Agreement Clause 4 imposes the same obligation to calculate commissions on actual reinsurance premiums received. The same estimation substitution methodology was applied from July 2005 (CC-13). The same corrective payment of $3,319,108 in February 2024 expressly covered CA underpayments (the Japan transposition error — breach 29 in the Comprehensive Breach Matrix — was admitted and paid). The Two Sets of Books evidence (TAB_38_3) applies equally to the CA stream. The contractual construction argument (Mount Bruce Mining; Electricity Generation Corp) applies with equal force.
Breach 29: Japan transposition — admitted, paid Feb 2024
Breach 30: Two Sets of Books — $2.7m admitted
Breach 31: Korea Short-Term rate — admitted Jun 2024
Breach 32: Australia estimate when actuals held — D2.67.1 admits actuals available
Component 2 — Why 81% not 82% (marginally lower than P3)
The CA stream has less CA-specific documentary corroboration than the EA stream. The reduction is small but honest.
The EA stream (P3 at 82%) is supported by the CC-6 through CC-13 correspondence chain, the D3.70 admission, and the Two Sets of Books evidence — all of which refer specifically to the EA calculation methodology. The CA stream relies more heavily on extrapolation from the EA evidence and on the admitted CA-specific breaches. The Allens counterclaim of approximately $2.4m (alleging CA overpayment) also introduces a live dispute over CA quantum that does not affect EA quantum — this adds a small increment of CA-specific risk not present in P3. The 1% reduction from 0.82 to 0.81 is conservative and honest.
Allens counterclaim ~$2.4m on CA: immaterial (3% of net claim) but live
Chaplin v Hicks [1911] — defendant cannot defeat claim by causing quantum uncertainty
△
Note on the Allens counterclaim: The ~$2.4m Allens counterclaim alleges CA overpayment. The Korea component of the counterclaim ($31k) constitutes Aegon's implicit concession that Korea commissions were underpaid — because the Australia overpayment arises from the same estimation methodology that produced the Korea underpayment. The counterclaim is self-defeating and is not a material risk to P6.
P7 — 71%
Component 1 — The inferential gap (why P7 is the lowest input)
P7 is the most honest expression of uncertainty in the model. It reflects what can be inferred from the established breach pattern without yet having the underlying data.
The methodology breach established for the EA and CA streams (P3 and P6) was systematic and applied across Korea, Japan, and Australia from July 2005. The logical inference is that the same estimation substitution applied uniformly across all programmes and cohorts — Aegon used one estimation system, and there is no evidence it applied that system selectively to some programmes but not others. However, Component B requires the court to accept this inference without programme-specific corrective payment evidence for every cohort. The Clause 6.7 demand for production of actual premium bordereaux across all territories (production order sought in the Notice of Motion) is the key to converting Component B from inference to direct proof. Until those records are produced, P7 reflects the honest inferential position.
Notice of Motion §5: production order for all actual premium bordereaux 2005-present
CC-9: Aegon admits it cannot retrieve basis for estimation rates
Jones v Dunkel (1959) — adverse inference from withholding evidence
Chaplin v Hicks [1911] — defendant-created quantum uncertainty
Component 2 — IDO (interrogatory / discovery order) scenario
Production of actual premium records would convert Component B from inference to direct proof. P7 would be expected to move to approximately 0.80 on IDO.
If the court grants the production order for actual premium bordereaux (Notice of Motion §5), the actual records will either confirm the Component B generalisation (expected, given the uniform estimation system) or reveal programme-specific variations. On the most probable scenario — uniform application confirmed — P7 would move from 0.71 to approximately 0.80 (the hard cap of 0.88 is not reached because some programme-specific variation is always possible even within a uniform system). The Jones v Dunkel adverse inference from Aegon's continued refusal to produce the records is already partially factored into P7 at 0.71 — the inference is that the records, if produced, would support the Component B claim.
△
Why 71% is correct and should not be inflated: The temptation is to set P7 higher because the logical inference from the established breach is strong. The reason to resist that temptation: P7 is the input that requires the most judicial discretion — a court may decline to extend the established breach to unspecified cohorts without specific evidence for each programme. Setting P7 at 0.71 is the honest expression of that residual. It is the model's most conservative input for a reason.
P8 — 78%
Component 1 — Route C is entirely independent of all other routes
Route C is an independent cause of action within the 6-year limitation window regardless of all other limitation arguments. Accrual: October 2025. Expiry: October 2031.
Route C does not depend on P4 (s.55) or P5 (demand accrual) succeeding. It is an entirely new and independent limitation mechanism. The fresh fraud basis: Aegon's conduct in the period 2021–2025 — renewed written denials of liability, repeated refusals to produce actual premium records, the illegal NDA conditions attaching fraud-immunity requirements, and continued payment at the wrong rate after written demand — constitutes fresh fraudulent conduct within the standard 6-year limitation period. The Allens ~$2.4m false debt counterclaim (October 2025) is itself a fresh fraudulent act within time on any analysis.
s.249K Crimes Act 1900 (NSW) — fresh fraud limitation extension
Allens ~$2.4m counterclaim Oct 2025 — fresh fraud within time
2021–2025: renewed denials after written requests
Component 2 — Six documented fraudulent statements
D.2.67, D.2.83b (false "actual data" claims), D.2.57, D.2.76b, D.2.85 (false denials). These are the evidentiary foundation for Route C.
The Two Sets of Books evidence (TAB_38_3, strategic reserve) provides six specific documented fraudulent statements: D.2.67 and D.2.83b are false claims that actual data was being used when estimates were in fact being used; D.2.57, D.2.76b, and D.2.85 are false denials of the methodology breach. These statements were made after written requests for the actual data — they are not historical misstatements but current representations made to defeat a live demand. Route C exists because these 2021–2025 statements are independently actionable within the ordinary limitation period regardless of the historical claim period.
D.2.67, D.2.83b: false "actual data" claims — TAB_38_3 strategic reserve
D.2.57, D.2.76b, D.2.85: false denials — TAB_38_3 strategic reserve
Component 3 — v9.1 correction: L8 is now correctly wired
In model v9, L8 was scored in the inputs but absent from the Boolean win function gates. The Shapley sheet showed zero contribution — a formula defect, not a legal conclusion.
The v9 Shapley decomposition showed L8 contributing zero to overall win probability. This was not a reflection of L8's legal weakness — it was a formula defect. L8 was absent from the Boolean win function's Contract stream limitation gate despite being included in the probability inputs. v9.1 corrects this: L8 is now correctly wired as an independent OR gate in the Contract stream limitation layer alongside L4 (s.55) and L5 (demand accrual). L8 now registers its true marginal contribution. The Boolean logic is: Contract stream wins if merits succeed (L3 OR L6 OR L7) AND limitation clears (L4 OR L5 OR L8). L8 is an independent third route for the limitation layer.
△
Why 78%: Route C is strong as a legal mechanism but depends on courts accepting the specific 2021–2025 conduct as constituting fresh fraud under s.249K. The primary legal question is whether conduct that is also relevant as concealment evidence for s.55 purposes can independently constitute a fresh cause of action under s.249K, or whether the doctrine of merger applies. NSW authority is not settled on this precise question. 78% reflects Route C's strength as a legal argument with an honest allowance for the uncertainty.
P9 — 78%
Component 1 — Isolation from merits (critical architectural point)
P9 is architecturally isolated from P1–P8. It affects the costs-adjusted exposure figure only. It cannot be used to argue that misconduct increases win probability.
P9 operates exclusively on the Costs-Adjusted Total Exposure KPI (EV + P9 × costs estimate). It does not affect: overall win probability, expected recovery (EV), settlement floor (P10), branch-weighted PV, or DFCR-adjusted EV. This architectural isolation is deliberate and prevents double-counting — the same conduct facts that support P9 are already embedded in the SAI score (which generates uplifts on P1–P8) and in the EFI score. Counting them again through P9 on the merits would be double-counting. P9 is the one and only mechanism through which conduct is reflected in the costs calculation.
Component 2 — The conduct record supporting an indemnity costs order
Five specific conduct categories establish the basis for indemnity costs. The research range for comparable conduct is 65–80%. P9 is set at 78%.
Systematic document withholding: Six formal data requests refused 2020–2024. Clause 6.7(f) verification rights denied. Jones v Dunkel adverse inference available.
Illegal NDA conditions: Three conditions attached to audit access that had no contractual basis and amounted to a demand that Harrison surrender fraud claims and consent to evidence destruction as a precondition to seeing his own commission data.
Five irreconcilable positions over 19 years: The sequential advancement and abandonment of five mutually inconsistent justifications for the Korea rate reduction, each requiring the court to conclude that the previous position was wrong, is the kind of conduct that attracts indemnity costs awards in NSW courts.
Continued denial after the Hallal admission: The August 2024 Hallal admission that the Marketing Agreement did not apply to Korea — which destroyed 19 years of prior justifications — did not produce any concession or settlement offer until April 2025, and then only a partial concession. Continued denial after a conclusive admission warrants indemnity costs.
Corrective payment without concession: The $3,319,108 corrective payment was made without any accompanying admission of liability — a tactical manoeuvre that required Harrison to continue litigation to establish what Aegon had already impliedly admitted by paying.
Illegal NDA conditions: Three conditions attached to audit access that had no contractual basis and amounted to a demand that Harrison surrender fraud claims and consent to evidence destruction as a precondition to seeing his own commission data.
Five irreconcilable positions over 19 years: The sequential advancement and abandonment of five mutually inconsistent justifications for the Korea rate reduction, each requiring the court to conclude that the previous position was wrong, is the kind of conduct that attracts indemnity costs awards in NSW courts.
Continued denial after the Hallal admission: The August 2024 Hallal admission that the Marketing Agreement did not apply to Korea — which destroyed 19 years of prior justifications — did not produce any concession or settlement offer until April 2025, and then only a partial concession. Continued denial after a conclusive admission warrants indemnity costs.
Corrective payment without concession: The $3,319,108 corrective payment was made without any accompanying admission of liability — a tactical manoeuvre that required Harrison to continue litigation to establish what Aegon had already impliedly admitted by paying.
Six formal data refusals 2020–2024
Three illegal NDA conditions — no contractual basis
Five irreconcilable positions — 19 years
Continued denial post-Hallal admission
Corrective payment without liability concession
△
Why 78%: Research range for indemnity costs on comparable conduct is 65–80%. The conduct record here is at the upper end of that range — systematic, documented, and spanning 19 years. 78% reflects that the conduct is strong but that courts retain full discretion on costs and that some elements of the record (particularly the NDA argument) have not yet been tested in NSW proceedings. The 22% residual is the honest allowance for judicial costs discretion and the risk that the court awards ordinary costs rather than indemnity costs even on a finding of misconduct.
The discipline rule that applies to all nine inputs: These rationales document the best available calibration at March 2026. Every default was derived through a documented process. Every default can be challenged — by new evidence, by counsel assessment, by changes in the defendant's position. What none of them should ever be challenged by is a preference for a more favourable output. The integrity of the model is entirely dependent on the integrity of its inputs. Treat these rationales as the reference point for every slider movement.