Thesis Engine

Engine progress 0.5-beta.20

Reliability, research coverage and forecast skill are tracked separately.

System reliability99.3%

Research coverage40.0%

Forecast skill—

Source success100.0%

Evidence items19

Atomic active events23

Analysis events13

New events0

Seeded events0

Routine events pruned0

Quality flags1

Run delta: source success +0.0% · research coverage +0.0% · evidence 0.0 · atomic events 0.0 · analysis events 0.0 · true new events 0.0 · quality flags 0.0

batch_5_v0.5_beta20_cwen_dated_borrower_maturities — PASS (7/7)

✓ maturity review attached without forced selection
✓ maturity review matches selected annual and later primary
✓ dated calendar principal and scope preserved
✓ no synthetic current payment schedule
✓ no additional cafd deduction or interest schedule
✓ dated maturities do not approve common cash or capital
✓ specific maturity regression guards registered

batch_1_v0.2_alpha — PASS (6/6)

✓ no tracked python caches
✓ historical trend engine
✓ cross source crypto valuation
✓ richer stock fundamentals
✓ dimension influence cap
✓ performance scorecard

batch_2_v0.2_beta — PASS (7/7)

✓ crypto economic activity
✓ stock free cash flow
✓ stock market valuation
✓ stock balance sheet freshness guard
✓ peer relative context
✓ separated performance scores
✓ source health at least 90pct

batch_3_v0.3_event_intelligence — PASS (15/15)

✓ primary source event ingestion
✓ event provenance complete
✓ event deduplication enabled
✓ deterministic bull bear forensic roles
✓ neutral evidence confidence guard
✓ material event alert path
✓ bootstrap events separated
✓ event freshness watermark guard
✓ historical backfills not new
✓ routine github noise filtered
✓ analysis event clustering
✓ event relevance ranking
✓ sec primary document enrichment
✓ sec context parser v2
✓ source health at least 90pct

batch_4_v0.4_valuation_portfolio — PASS (15/15)

✓ monte carlo distribution all assets
✓ p1 success separate from p2 return
✓ uncertainty aware fractional kelly
✓ uncalibrated sizing haircut
✓ individual position caps enforced
✓ crypto concentration cap enforced
✓ tracked sleeve cap enforced
✓ correlation guard enabled
✓ missing correlation no diversification credit
✓ portfolio artifact generated
✓ distribution aware status policy
✓ mean only promotion guard
✓ model migration alert guard
✓ policy migration not market alert
✓ source health at least 90pct

batch_5_v0.5_relationship_calibration_foundation — PASS (21/21)

✓ dated forecast ledger
✓ forecast history not silently resolved
✓ relationship graph generated
✓ primary source relationship provenance
✓ public company registry enabled
✓ blind spot candidates research only
✓ relationship zero thesis weight
✓ multi hop depth capped
✓ research queue generated
✓ relationship source health at least 90pct
✓ semantic context parser v7
✓ relationship local exposure only
✓ common sector factors separated
✓ corroboration requires independence or reciprocity
✓ economic materiality requires verified exposure
✓ relationship edge cap not saturated
✓ semantic rejections audited
✓ relationship history state tracked
✓ negative evidence policy enabled
✓ mitigating context not in risk queue
✓ forecast skill still unclaimed

batch_5_v0.5_autonomous_discovery_learning — PASS (19/19)

✓ autonomous internet discovery enabled
✓ discovery source health at least 80pct
✓ stock universe source operational
✓ stock sec deep scan operational
✓ crypto universe sources operational
✓ crypto universe broad enough
✓ underdog research queue generated
✓ discovery candidates research only
✓ risk adjusted discovery scores
✓ queue scores current model only
✓ discovery ledger migration policy
✓ extreme fdv guard
✓ pathological stock ratio guard
✓ regression guard registry
✓ persistent discovery ledger
✓ recursive learning loop safe
✓ learning lessons have recurrence
✓ challenger changes proposal only
✓ forecast skill still unclaimed

batch_5_v0.5_beta_dossiers_short_forecasts — PASS (15/15)

✓ autonomous research dossiers generated
✓ dossier promotions bounded
✓ dossiers research only zero weight
✓ dossiers current scoring version only
✓ source independence semantics scoped
✓ independent source present each dossier
✓ issuer filings not independent fundamental confirmation
✓ short horizon forecast ledger generated
✓ short forecast horizons present
✓ short forecasts zero thesis weight
✓ short forecasts uncalibrated until outcomes
✓ forecast resolution lookahead guard
✓ brier only from resolved forecasts
✓ short forecast feedback integrated
✓ forecast skill still not claimed

batch_5_v0.5_beta1_variant_perception_edge — PASS (14/14)

✓ edge intelligence generated
✓ maximum curiosity policy explicit
✓ evidence and edge axes separated
✓ unknown is uncertainty not negative
✓ drawdown alone not edge
✓ consensus claim requires direct data
✓ edge queue zero thesis weight
✓ blind spot angles generated
✓ exploration lane bounded
✓ exploration lane zero weight and no severe risk
✓ limited evidence edge state supported
✓ relationship second order hypotheses zero weight
✓ edge score not in portfolio action path
✓ edge regression guards active

batch_5_v0.5_beta2_exploration_confidence — PASS (7/7)

✓ edge model v2 semantics
✓ evidence confidence is non probability index
✓ evidence confidence separate from edge
✓ exploration reserved outside deep research
✓ exploration relative asset class breadth
✓ exploration outcome cohort recorded
✓ no new outcome engine or fake skill claim

batch_5_v0.5_beta3_independent_context_causal_validation — PASS (8/8)

✓ candidate research context generated
✓ candidate context bounded
✓ independent industry context not fundamental
✓ generic bottleneck prompts not evidence
✓ crypto operational context scoped
✓ candidate context zero thesis weight
✓ context source health acceptable
✓ no new ranking or probability layer

batch_5_v0.5_beta31_entity_clause_hardening — PASS (7/7)

✓ relationship parser current live context
✓ ambiguous company name guard enabled
✓ relation company binding is clause local
✓ acquisition and competition excluded from causal angles
✓ bounded context cap is post quality ranking
✓ beta31 regression guards active
✓ no new scoring or portfolio layer

batch_5_v0.5_beta32_counterparty_role_semantics — PASS (6/6)

✓ relationship parser v7 role semantics
✓ cloud provider demand is not technology dependency
✓ end market dependency supported
✓ transaction and competition roles remain contextual
✓ beta32 regression guards active
✓ no new source score or portfolio layer

batch_5_v0.5_beta33_scoped_ticker_ci_enforcement — PASS (7/7)

✓ relationship parser v7 scoped tickers
✓ global ticker alias resolution disabled
✓ role scoped tickers only valid counterparty roles
✓ cloud end market reclassification consistent
✓ all relationship context stays zero weight
✓ beta33 regression guards active
✓ no new source score or portfolio layer

batch_5_v0.5_beta4_external_crosscheck_consensus — PASS (9/9)

✓ external stock context live
✓ independent publisher scope honest
✓ publisher fundamental crosschecks audited
✓ fundamental independent confirmation not inflated
✓ direct sell side consensus observed and zero weight
✓ ownership short interest context scoped
✓ external context does not change thesis weight
✓ beta4 regression guards active
✓ no new score probability or portfolio action

batch_5_v0.5_beta5_autonomous_valuation — PASS (11/11)

✓ autonomous valuation generated
✓ selected research candidates receive valuation
✓ stock scenario ranges complete
✓ analyst targets comparison only
✓ scenario probabilities not fabricated
✓ breakthrough upside scenario preserved
✓ crypto proxy not token fair value
✓ input conflict blocks valuation readiness
✓ valuation zero thesis and portfolio weight
✓ valuation regression guards active
✓ no new probability or automatic trade path

batch_5_v0.5_beta6_institutional_research_intelligence — PASS (12/12)

✓ institutional research model present
✓ aum 500b first party whitelist enforced
✓ deep analysis only after potential gate
✓ anti conservative structural or flagship override
✓ high potential papers atomic and zero weight
✓ same publisher repetition not confirmation
✓ candidate links are research hints not exposure
✓ publisher prestige not probability
✓ no peer outperformance probability fabricated
✓ institutional layer no automatic portfolio action
✓ institutional regression guards active
✓ institutional source health not catastrophic

batch_5_v0.5_beta61_institutional_theme_mapping_hardening — PASS (5/5)

✓ institutional mapping policy v2
✓ crypto not mapped to physical infrastructure themes
✓ energy theme requires energy specific candidate term
✓ candidate hints remain zero weight
✓ beta61 regression guards active

batch_5_v0.5_beta7_economic_capture_evidence — PASS (11/11)

✓ economic capture layer generated
✓ selected research candidates receive economics
✓ stock operating economics present
✓ stock snapshot not falsely normalized
✓ crypto network economics attempted
✓ crypto network economics source operational
✓ chain economics not tokenholder capture
✓ no token value capture fabricated
✓ economics zero thesis and portfolio weight
✓ economic capture regression guards active
✓ no new discovery score probability or trade path

batch_5_v0.5_beta8_formal_thesis_readiness — PASS (12/12)

✓ formal thesis readiness generated
✓ all selected research candidates assessed
✓ essential blockers optional enhancers policy
✓ discovery score not readiness gate
✓ ready means review not capital
✓ ready candidates pass all essential checks
✓ blocked candidates preserved for research
✓ crypto requires verified token capture
✓ no calibrated probability or buy signal created
✓ formal breaker not auto selected
✓ readiness regression guards active
✓ no new position sizing or portfolio path

batch_5_v0.5_beta9_targeted_capture_resolution — PASS (12/12)

✓ targeted resolution generated
✓ selected crypto only assessed
✓ primary document budget bounded
✓ primary source operational
✓ targeted finding observed
✓ exact project routes only
✓ source provenance and failures visible
✓ reports not verified capture
✓ followups preserve zero weight
✓ candidate specific evidence actions present
✓ stock normalization followup not new readiness blocker
✓ targeted capture regression guards active

batch_5_v0.5_beta10_op_custody_reconciliation — PASS (12/12)

✓ op custody reconciliation generated
✓ exact op chain contract wallet only
✓ chain request budget bounded
✓ chain provenance and failures visible
✓ dated monthly claims not cumulative totals
✓ balance pinned to finalized block hash
✓ rpc transfer observation operational
✓ indexed rpc conflicts held for review
✓ lot consistency not purchase attribution
✓ bounded window not complete wallet history
✓ custody preserves unverified capture and zero weight
✓ custody regression guards active

batch_5_v0.5_beta11_financial_basis_reconciliation — PASS (10/10)

✓ financial basis diagnostics generated
✓ primary financial basis operational
✓ fresh net revenue policy preserved
✓ primary tags periods and components visible
✓ mismatched cash flow windows not combined
✓ publisher summary dates not measurement dates
✓ unresolved differences remain blockers
✓ no publisher override or independent confirmation
✓ no new normalization readiness requirement
✓ financial basis regression guards active

batch_5_v0.5_beta12_source_backed_thesis_review — PASS (12/12)

✓ selected candidates receive review or followup
✓ selected primary operating evidence observed
✓ primary filing identity dates and hash visible
✓ operating comparisons scoped to issuer period and usd
✓ cash spending kept separate not fcf resolution
✓ blocked candidates not promoted to formal drafts
✓ missing current filing visible not fabricated
✓ draft does not complete review or select breaker
✓ draft has no probability capital or portfolio weight
✓ existing scenarios remain conditions not underwriting
✓ annual followup request bound preserved
✓ formal review and relation regression guards active

batch_5_v0.5_beta13_cash_obligations_and_fcf_definition — PASS (12/12)

✓ targeted cash reviews generated
✓ publisher statement request bounded and cached
✓ reported cash and accrual same balance date
✓ cashflow adjustment not assumed from expense
✓ accrual sensitivity not free cash or case resolution
✓ definition resolution requires two dates and six components
✓ dated statement identity scale and hash visible
✓ standardized primary basis and threshold preserved
✓ research valuation uses issuer derived spending basis
✓ remaining or failed comparisons keep conflict blocker
✓ cash review has no override confirmation or equity claim
✓ cash obligation and definition regression guards active

batch_5_v0.5_beta14_paired_stock_thesis_findings — PASS (12/12)

✓ paired review findings generated
✓ issuer exhibit requests bounded and cached
✓ completed findings match exact source provenance
✓ completed findings preserve formal draft state
✓ sam signed litigation sensitivity not normalized
✓ sam demand and selling margin evidence scoped
✓ cwen four historical cafd cash bridges reconcile
✓ cwen interest maturity and restricted cash not double counted
✓ dated llc ownership and share scope remain sensitivities
✓ cafd does not replace fcf or supply per share value
✓ analytical completion does not approve thesis or capital
✓ paired thesis regression guards active

batch_5_v0.5_beta15_cwen_share_capitalization — PASS (12/12)

✓ cwen share review attached without forced selection
✓ share review reuses selected primary without new requests
✓ verified share scope matches exact primary provenance
✓ two dated class counts reconcile without unit or average substitution
✓ dividend and sponsor llc distribution paths preserved
✓ dated share ratio only corroborates rounded ownership
✓ original publisher inputs and effective date gaps preserved
✓ class scope arithmetic recalculated not provider confirmation
✓ current capitalization and equity cash remain unverified
✓ missing or later share sources visible and per share outputs held
✓ share resolution does not approve thesis or capital
✓ specific capitalization regression guards registered

batch_5_v0.5_beta16_common_cash_and_portal_contract — PASS (12/12)

✓ distribution review scoped and missing sources visible
✓ distribution split matches primary and cash flow payment
✓ cafd growth and debt sensitivities do not double subtract
✓ historical distribution does not approve recurring cash or capital
✓ specific common cash regression guards registered
✓ portal contract is versioned and strict json
✓ portal candidate ids survive selection rotation
✓ portal never derives approval from readiness
✓ portal preserves withheld prices and unknown effective dates
✓ portal is read only and omits internal source bodies
✓ portal generation timestamp is separate from source dates
✓ specific portal regression guards registered

batch_5_v0.5_beta17_cwen_financing_liquidity — PASS (12/12)

✓ financing review attached without forced selection
✓ financing reuses exact selected primary
✓ financing observations preserve primary values
✓ repayment residual is recalculated not recurring allocation
✓ cafd principal match retains arithmetic policy
✓ revolver repayment compares separate cashflow line
✓ restricted liquidity not common cash
✓ debt maturities not full forward schedule
✓ issuer covenant statement is dated
✓ financing does not approve cash or capital
✓ financing keeps per share hold and portal boundary
✓ specific financing regression guards registered

batch_5_v0.5_beta18_cwen_cafd_exclusions_contractual_claims — PASS (12/12)

✓ cash claim review scoped and missing evidence visible
✓ cash claim review reuses both matched hashes
✓ cash claim values preserve primary and exhibit scopes
✓ tax credit exclusion matches reported cafd line
✓ partner net cash exclusion not ordinary dividend claim
✓ issuer principal exclusion matches existing cafd
✓ contract aggregates use inception and lifetime scopes
✓ remaining and annual contract payments not invented
✓ llc distribution qualifications preserved
✓ exclusions and contracts do not create new cafd deduction
✓ cash claim review does not approve common cash or capital
✓ specific cash claim regression guards registered

batch_5_v0.5_beta19_cwen_dated_market_inputs — PASS (12/12)

✓ market review scoped and incomplete evidence visible
✓ market review reuses existing sources
✓ market review preserves publisher hash
✓ market review preserves primary hash
✓ regular close session and timestamp preserved
✓ labelled exact and rounded values preserved
✓ missing share and cap effective dates not inferred
✓ dated class comparisons are observations
✓ same page arithmetic recalculated not forced
✓ original valuation inputs and price difference preserved
✓ market review does not approve capital or common cash
✓ specific market input regression guards registered

Forecast skill remains blank until dated predictions have resolvable outcomes; this prevents fake calibration.

ADA Cardano

crypto

RESEARCH — insufficient evidence

Fundamental success24.5%

Directional confidence7.6%

Expected multiple*1.14×

P(5×+)*6.6%

Permanent-loss risk*54.5%

Market cap: $9.58B · DeFi TVL: $69.61M · MCap/TVL: 137.6× · Stablecoins: $67.32M · Stablecoins/MCap: 0.70% · DEX 30d: $83.85M · Fees 30d: $506.20K · MCap/TVL vs peer median: 15.75×

Positive evidence

None yet

Negative evidence

Market-cap/TVL is 15.75× the median of configured crypto peers. Cross-chain TVL comparability is imperfect, so this signal is deliberately low-weight.
1-year token price change: -69.3%

Primary-source event intelligence

Atomic active: 22 · prioritized/clustered: 12 · new since prior run: 0 · bootstrap seeded: 0

4 related governance change updates — committee schemagovernance change · 2026-08-10T10:47:37Z · materiality 0.80 · relevance 0.59 · 4 related atomic events

IntersectMBO/governance-actions: Merge pull request #175 from IntersectMBO/revert-174-update-intersect-schemas Revert "Committee schema update"; IntersectMBO/governance-actions: Revert "Committee schema update"; IntersectMBO/governance-actions: Merge pull request #174 from IntersectMBO/update-intersect-schemas Committee schema update; plus 1 related update(s).