Methodology

Status first, source next, price after that

Orrery turns prediction-market prices into a verification workflow: status first, source next, then price, market quality, flow, deadline, and oracle state.

read-onlydry-run auditv2026-06-06.1

Methodology kit connected

The public page now renders from the vendored source model. Orrery owns layout; the kit owns doctrine order, formulas, caveats, FAQ, and source references.

compiled 2026-06-06 · contract 2026-06-06.1

Engine boundary

100/100100/100

Methodology, signal interpretation, risk gates, autonomous governor/allocator contracts, and dry-run approvals are connectable. Runtime execution remains operator-gated in main Orrery.

Public methodologyread-only

Rendered from the vendored methodology kit; no runtime network or executor dependency.

Signal enveloperead-only

Live Orrery signals can be interpreted against the stable envelope without order placement or wallet signing.

Operator reviewdry-run audit

Order intents, approvals, would-send packets, and receipts can be displayed as audit artifacts only.

Paper executordry-run audit

Approved intents can be staged into paper fills before any supervised canary or live executor path is enabled.

Runtime submit locks6 LOCKED
Executor autostartFALSE

No app surface can start an unattended executor loop by itself.

Network sendFALSE

Venue network-submit packets remain paper or would-send evidence.

Raw secretsFALSE

Private-key/API-secret material is never loaded by methodology or UI.

Unattended live submitFALSE

Live order submission requires explicit operator enablement outside the library.

Policy expansionFALSE

Venue, market, account, and cap expansion stays locked behind signed operator policy.

Self-modifying strategyFALSE

Runtime decisions can be audited; strategies cannot rewrite their own mandate.

fairProbabilityread-only

Can be displayed when produced by the engine/model boundary; never becomes an order instruction by itself.

marketProbabilityread-only

Can mirror the observed market-implied YES probability on public signal surfaces.

edgeread-only

Can be interpreted as fairProbability minus marketProbability before governor vetoes and sizing caps.

confidenceread-only

Can show rule evidence strength from the deterministic public signal.

liquidityRiskread-only

Can be interpreted from liquidity and spread as context, not sizing guidance.

resolutionRiskread-only

Can surface status, source, expiry, and oracle caveats before price interpretation.

executionCostdry-run audit

Cost/slippage evidence can veto a decision and is first wired through paper receipts.

manipulationRiskread-only

Flow diagnostics can be surfaced as a governor veto input without sending orders.

Dry-run audit UINO SEND
Governor decisiondry-run audit

Records trade/no-trade plus stale, calibration, liquidity, manipulation, resolution, and cost vetoes.

autonomous decision governor

Allocator packetdry-run audit

Shows capped paper size by market, venue, outcome, theme, portfolio, drawdown, and correlated exposure.

autonomous capital exposure allocator

Pre-trade approval packetdry-run audit

Binds mandate, risk, kill-switch, reconciliation, exposure, and operator policy evidence.

autonomous trading approval boundary

Paper fill receiptdry-run audit

Transforms approved intent into simulated fills/rejects/slippage/PnL without venue network submit.

main Orrery paper executor

Supervision receiptdry-run audit

Tracks fills, rejects, partial fills, drift, stale data, cancel, flatten, and self-disable evidence.

autonomous supervision self-disable

Live submit receiptblocked

Receipt schema is visible for canary review; liveOrdersAllowed remains false until operator enablement.

executor enablement boundary

Next runtime gate: Main Orrery paper executor wiring. Readiness is complete; live submit stays behind the main Orrery operator boundary.

read-order

Read order

The headline probability is read only after status, source, quality, flow, deadline, and oracle state are visible.

01

Status

Is the market live, pending, disputed, resolved, or only rail-pinned?

Show canonical status before probability or signal language.

Orrery product bridgeResolution riskPrice as probability
02

Resolution source

Which source decides the outcome and how explicit is it?

Expose source type, confidence, URL or text, and matching rule when available.

Orrery product bridgeResolution risk
03

Market-implied probability

What probability is the venue implying right now?

Use market-implied probability; do not call it true probability.

Price as probability
04

Liquidity and spread

Is this price supported by depth, turnover, and a usable spread?

Treat liquidity, spread, and capacity as warnings and context.

Order-book market qualityThin book capacity
05

Flow

Did recent trades align, concentrate, or arrive from repeated wallets?

Call flow an observation; never call it proof of informed trading.

Order-book market qualityWallet flow methodology
06

Deadline

How close is the market to expiry or resolution?

Escalate near-expiry and expired-unresolved states before ranking opportunities.

Resolution riskBelief update loop
07

Oracle state

Is UMA or another mechanism pending, proposed, disputed, or already resolved?

Separate upstream resolved, UMA pending, UMA disputed, and rail-pinned states.

Resolution risk
doctrines

Six doctrines

These are the public guardrails that keep Orrery's interpretation layer observational.

Price is context, not truth

Prediction-market prices can summarize dispersed beliefs, but only under market-quality and resolution-source constraints.

Use market-implied probability and attach caveats when liquidity, spread, or source quality is weak.

Price as probabilityOrder-book market qualityResolution risk

Status before price

The same price means different things in live, near-expiry, pending, disputed, resolved, and rail-pinned states.

Status badges and warnings must appear before probability interpretation.

Resolution riskOrrery product bridge

Signals are observations

Momentum, divergence, flow, and resolution-risk signals describe what changed; they are not recommendations.

Signal cards should lead to source verification, timeline inspection, alerts, watch-only, or ignore actions.

Belief update loopOrder-book market qualityMarket scoring rules

Evidence is not edge

Live confidence, historical edge, aggregate edge, sample quality, and forward-only status are separate concepts.

Keep Evidence, Backtest, Action, sample tier, and forward-only labels visibly separate.

Market scoring rulesBelief update loop

Research Priority is attention ranking

Orrery can rank markets for inspection, but the score is not expected value and not a position-sizing rule.

Use Research Priority in public copy; treat edgeScore as an internal compatibility name.

Market scoring rulesThin book capacityOrrery product bridge

Agents need bounded next steps

A useful assistant says what to verify next and what remains uncertain, instead of declaring a trade conclusion.

Agent cards should include sources, freshness, warnings, and explicit next verification calls.

Belief update loopResolution riskOrrery product bridge
implementation-snapshot

Implementation snapshot

Formula and caveat snapshot audited against 2026-06-06. Stale-prone modules remain listed for drift review.

Momentum

snapshot

Fires when

  • absolute 24h price move is at least 3pp
  • liquidity is at least USD 25k
  • 1h move is nonzero
  • 1h and 24h moves have the same sign

Confidence

  • 0.25 base
  • + 0.35 * min(abs(24h_move) / 0.10, 1)
  • + 0.25 * min((volume24h / liquidity) / 2, 1)
  • + 0.10 when spread is below 4c
  • capped at 0.95

Static snapshot: mixed slice, weak aggregate, high sample.

Action: Verify the source, inspect the timeline, and watch whether the move persists.

Caveats

  • Momentum is not a trade call.
  • Thin books can exaggerate price movement.
  • Historical support is mixed, so live confidence is not edge.
Order-book market qualityBelief update loopMarket scoring rules

Divergence

snapshot

Fires when

  • absolute 1h price move is at least 1pp
  • liquidity is at least USD 50k
  • 1h and 24h moves have opposite signs

Confidence

  • 0.20 base
  • + 0.40 * min(abs(1h_move) / 0.04, 1)
  • + 0.25 * min(log10(liquidity) / 6, 1)
  • capped at 0.95

Static snapshot: mixed slice, weak aggregate, low sample.

Action: Inspect whether new information conflicts with the prior trend.

Caveats

  • Divergence can be noise, stale data, or a real update.
  • Low sample support should keep action language conservative.
Belief update loopOrder-book market qualityMarket scoring rules

Flow

forward-only

Fires when

  • at least 3 recent trades appear in the last 30 minutes
  • at least 3 recent trades align to the same direction
  • total recent trade-window USD value is at least USD 5k

Confidence

  • 0.30 base
  • + 0.35 * min(log10(total_usd) / 6, 1)
  • + 0.25 * min(unique_wallets / 5, 1)
  • + 0.10 * aligned_share
  • capped at 0.95

Forward-only until realized outcome data is accumulated.

Action: Monitor direction, concentration, wallet repetition, and source freshness.

Caveats

  • This is not three USD 5k trades; it is a total-window threshold.
  • Large flow is not proof of informed trading.
  • Direction-source quality must be audited before stronger language.
Order-book market qualityWallet flow methodology

Resolution risk

forward-only

Fires when

  • expiry is within roughly 72 hours
  • UMA status is present and not resolved
  • spread is above 6c
  • explicit or extracted resolution source is missing
  • source is ambiguous or a public-figure statement

Confidence

  • + 0.40 from near expiry
  • + 0.25 from unresolved UMA state
  • + 0.15 from spread above 6c
  • + 0.10 from missing source
  • + 0.05 from ambiguous or public-figure source

Forward-only until labeled resolution outcomes and disputes are stored.

Action: Verify source wording, oracle state, expiry, and dispute risk before interpreting price.

Caveats

  • Resolution risk changes the meaning of the price.
  • Rail-pinned is not resolved.
  • Oracle pending and disputed states require separate language.
Resolution riskOrrery product bridge

News lag

forward-only

Fires when

  • external news or source evidence appears to precede market reaction
  • the market reaction remains incomplete or delayed

Confidence

  • forward-only until source/event matching is audited

Forward-only.

Action: Compare source timestamp, market timestamp, and reaction path.

Caveats

  • Requires careful source timestamping.
  • Should not be promoted without event-source provenance.
Belief update loopMarket scoring rules

Drift audit modules

status precedence or thresholds changeresolution source extraction rules changesignal firing thresholds or confidence formulas changecalibration sample tiers or action labels changeResearch Priority scoring changeswallet sample, concentration, or copy-risk language changespublic methodology copy changes
lib2-runtime-bridge

Lib_2 runtime bridge

The new corpus is exposed as read-only paper diagnostics: integrity, semantics, microstructure, execution markout, and thin-liquidity context.

Kernel packet

read-onlypaper-onlylive disabled

market integrity

deterministic score, tier, reason codes, and paper-only veto/cap candidates

semantic risk

deterministic score, tier, reason codes, and paper-only veto/cap candidates

microstructure quality

deterministic score, tier, reason codes, and paper-only veto/cap candidates

execution markout

deterministic score, tier, reason codes, and paper-only veto/cap candidates

thin liquidity

deterministic score, tier, reason codes, and paper-only veto/cap candidates

Authority boundary

liveOrdersAllowedfalse
networkSendAllowedfalse
rawSecretsLoadedfalse
trainingAllowedfalse
modelPublishAllowedfalse
deployAllowedfalse

Forbidden actions

  • live_order_submit
  • network_send
  • raw_secret_load
  • executor_autostart
  • training
  • model_publish
  • deploy

Source artifacts

Lib_2/wiki/01_manipulation_integrity/market_integrity_layer.mdLib_2/wiki/02_resolution_semantics/semantic_risk_engine.mdLib_2/wiki/03_microstructure/event_market_microstructure.mdLib_2/wiki/04_execution_markout/fill_slippage_markout.mdLib_2/wiki/05_liquidity_design/thin_liquidity_mechanisms.mdLib_2/wiki/06_bridge/orrery_runtime_integration_proposal.md
research-priority

Research Priority

The public ranking is an inspection queue, not expected value and not a position-size rule.

Components

  • signal evidence: max signal confidence * 45
  • multi-kind agreement: 0, 6, or 12
  • liquidity: min(log10(max(liquidity, 1)) * 2.5, 15)
  • tight-spread factor: max(0, 10 - spread_cents * 2), or 5 if unknown
  • resolution-risk penalty: -10 when resolution-risk signal fires
  • final score: clamp(sum, 0, 100)

Internal compatibility name: edgeScore.

Confidence and capacity

  • 0.55 base when liquidity > USD 25k and spread is unknown or below 6c; otherwise 0.35
  • + max_signal_confidence * 0.25
  • + 0.15 when more than one signal kind fires
  • - 0.10 when resolution risk fires
  • clamped to 0..100%

max(500, liquidity * 0.08), shown only as rough context.

Caveats

  • Research Priority ranks inspection order, not expected value.
  • Capacity estimate is not a suggested size.
  • Resolution-risk penalty is a caution axis, not a valuation model.
Market scoring rulesThin book capacityResolution risk
wallet-flow

Wallet and large-trade flow

Wallet activity stays descriptive: sample, concentration, direction source, and copy-risk warnings before any story.

Principles

  • Large trade is not smart-money proof.
  • Activity is not profitability.
  • Wallet specialization is different from wallet skill.
  • Copy-risk should be shown as a caution axis, not a follow instruction.

Sample tiers

  • low confidence: fewer than 3 relevant trades
  • limited confidence: 3 to 9 relevant trades
  • medium confidence: 10 to 29 relevant trades
  • established sample: 30 or more relevant trades

Diagnostics

  • unique wallet count
  • aligned share
  • single-wallet concentration
  • repeated-market or repeated-theme exposure
  • recent trade-window value

Copy-risk warnings

  • Display copy-risk as a risk label, not as an invitation to mirror a wallet.
  • Prefer monitor, inspect, compare, and verify language.
  • Do not imply the user should follow a wallet or automate a trade from wallet data.
Wallet flow methodologyOrder-book market quality
resolution-risk

Resolution and oracle risk

Finality, rail-pinning, pending states, disputes, and source wording remain separate before probability copy.

Status precedence

1. archived2. UMA disputed3. UMA pending or proposed4. upstream resolved5. expired-unresolved6. rail-pinned7. illiquid8. near-expiry9. live

Risk evidence

  • near expiry within roughly 72 hours
  • expired but unresolved
  • UMA pending or disputed state
  • wide spread near resolution
  • missing, ambiguous, or weak source text
  • public-figure statement used as source

Invariants

  • RAIL_PINNED_NO is not RESOLVED_NO.
  • RAIL_PINNED_YES is not RESOLVED_YES.
  • Future-expiry markets cannot be marked resolved without upstream evidence.
  • UMA pending, proposed, and disputed states must be visible before price interpretation.
  • The deterministic source extractor is best-effort and is not an oracle by itself.
LIVENEAR_EXPIRYEXPIRED_UNRESOLVEDUMA_PENDINGUMA_DISPUTEDRESOLVED_YESRESOLVED_NORAIL_PINNED_YESRAIL_PINNED_NOILLIQUIDARCHIVED
Resolution riskOrrery product bridge
limits

Limits

The integration is deliberately read-only and dry-run only.

Page warnings

  • Orrery does not execute trades.
  • Orrery does not sign wallet transactions.
  • Orrery does not provide investment, legal, or tax advice.
  • Forward-only signals must remain visibly labeled until outcome data exists.

Public copy contract

Preferred

  • market-implied probability
  • Research Priority
  • inspection priority
  • source verification
  • priced near YES
  • priced near NO
  • flow observation
  • rough capacity context

Careful

  • edge, only when explicitly separated from live evidence and sample quality
  • confidence, only when the source of confidence is named
  • signal, only as an observation
  • wallet quality, only with sample and concentration caveats

Forbidden

  • true probability
  • guaranteed edge
  • smart money, unless explicitly debunked or quoted as a risk label
  • follow this wallet
  • copy trade
  • suggested size
  • trade recommendation
faq

FAQ

Short answers with source references from the methodology kit.

What does Orrery mean by market-implied probability?

It means the price-implied probability visible in the market, interpreted with liquidity, spread, status, and resolution-source caveats.

Price as probabilityOrder-book market qualityResolution risk

Why does status come before price?

Because live, near-expiry, pending, disputed, resolved, and rail-pinned markets need different interpretation rules.

Resolution riskOrrery product bridge

How does Orrery compute market signals?

It applies deterministic observation rules for momentum, divergence, flow, resolution risk, and forward-only news lag.

Public methodology rewrite specOrder-book market qualityBelief update loop

Why split Evidence, Backtest, and Action?

Live evidence, historical support, aggregate edge, sample size, and next action are different claims and should not collapse into one score.

Market scoring rulesBelief update loop

What is Research Priority?

It is an attention ranking that helps decide what to inspect first. It is not expected value and not a suggested position size.

Market scoring rulesThin book capacity

How does Orrery handle wallet and large-trade flow?

It treats flow as an observation with sample, concentration, and direction-source caveats; wallet data is for monitoring and comparison, not copying.

Wallet flow methodologyOrder-book market quality
sources

Sources

The renderer exposes kit source references without importing wiki files at runtime.

Methodology spine v1

wiki
../../wiki/00_meta/methodology_spine_v1_2026-06-06.md

methodology-spine

Orrery product bridge

bridge
../../wiki/00_meta/orrery_product_bridge.md

product-bridge

Public methodology rewrite spec

bridge
../../wiki/00_meta/methodology_page_rewrite_spec_2026-06-06.md

rewrite-spec

Phase 1 source closeout

wiki
../../wiki/00_meta/phase1_source_closeout_2026-06-06.md

phase1-closeout

Source inventory manifest

docs
../../manifests/source_inventory_v1.json

source-inventory

Price as probability

wiki
../../wiki/01_prediction_markets/price_as_probability.md

price-probability

Market scoring rules

wiki
../../wiki/02_market_scoring/market_scoring_rules.md

scoring-rules

Order-book market quality

wiki
../../wiki/03_microstructure/orderbook_market_quality.md

microstructure

Wallet flow methodology

wiki
../../wiki/03_microstructure/wallet_flow_methodology.md

wallet-flow

Belief update loop

wiki
../../wiki/04_bayesian_forecasting/belief_update_loop.md

belief-update

Thin book capacity

wiki
../../wiki/05_liquidity_execution/thin_book_capacity.md

thin-book-capacity

Resolution risk

wiki
../../wiki/06_resolution_oracles/resolution_risk.md

resolution-risk

M1 methodology policy packet

docs
../../manifests/methodology_policy_packet_v1.json

m1-methodology-policy-packet

M2 methodology validator

docs
../../manifests/methodology_validator_v1.json

m2-methodology-validator

P30 human-grade explanation

docs
../../manifests/p30_human_grade_explanation_v1.json

p30-human-grade-explanation

P31 deterministic dry-run report

docs
../../manifests/p31_deterministic_dry_run_report_v1.json

p31-deterministic-dry-run

P82 library coverage readiness

docs
../../manifests/library_coverage_readiness_v1.json

p82-library-readiness

P84 validator readiness bundle

docs
../../manifests/p84_methodology_validator_readiness_bundle_v1.json

p84-validator-readiness-bundle

P81 Hinkley changepoint source card

wiki
../../wiki/08_paper_reading_notes/p81_014_hinkley_changepoint_inference_full_application.md

p81-hinkley-changepoint

P81 exchangeability martingale source card

wiki
../../wiki/08_paper_reading_notes/p81_016_plugin_martingales_testing_exchangeability_online_full_application.md

p81-exchangeability-martingales

P81 confident OPE policy-selection source card

wiki
../../wiki/08_paper_reading_notes/p81_026_confident_ope_policy_selection_hcope_full_application.md

p81-confident-ope-policy-selection

P81 KSWIN source snapshot

wiki
../../wiki/08_paper_reading_notes/p81_015_scikit_multiflow_kswin_source_snapshot_full_application.md

p81-kswin-source-snapshot

Need machine-readable references? /llms-full.txt covers the agent-facing reference, and /openapi.yaml remains the OpenAPI 3.1 endpoint map.

Methodology - Orrery verification workflow | Orrery