This account explains the analysis a reader can obtain today. Coverage figures retain their stated sources and vintages; planned capabilities are identified below.
What you learn
A decision market may price a token more highly under one proposal than another. Before treating that difference as evidence for the proposal, a reader needs to understand the objective, the terms of the alternatives and how the prices formed.
Eidos brings those questions together. An organization can explain its market to stakeholders; a fund can inspect the evidence behind a conditional valuation; an agent can retrieve supported readings with their sources and limits. The result is an interpretation a reader can examine and challenge.
The analysis follows five questions: what is priced, what the comparison implies, what trading supports it, which findings the evidence supports and what remains unknown. The record keeps those answers attached to the decision.
What is the market pricing?
Start with the proposal’s alternatives and stated objective. A market pricing a token under approval and rejection asks a different question from one forecasting a project’s measured outcome. Read the proposal terms beside the selection rule: which price or forecast determines the result, at what threshold and over which window?
Eidos’s automated semantic checks examine disclosed proposal class, threshold sign, tie behavior and measured-outcome rules. Questions about whether a promise can be enforced or whether the alternatives depend on one another need further sourced review. A disclosed term establishes what was stated, not that it will be delivered.
What does the comparison imply?
Read the conditional valuations or forecasts together, with their reference choice and observation time. Their difference is the market-implied impact on the objective being priced. In a token market, it concerns that token. Converting it to enterprise value or total organizational value requires a justified valuation and supply basis for each branch, including changes the proposal would make.
Keep the market implication separate from the window outlook, which forecasts passage under the selection rule. A favorable conditional comparison and a high likelihood of passage answer different questions. Eidos reports both where available and does not recommend what to approve, fund, trade or vote for.
What supports the price?
A price formed through broad trading at substantial depth has a different participation and liquidity basis from one formed through a few trades in a thin pool. Eidos exposes that basis through available instruments:
- Depth and flip-cost scenarios describe sensitivity to trading. The flip-cost scenario is a lower bound for moving a constant-product pool price from parity at disclosed or decoded reserves. Fees, opposing flow and the cost of holding the price through the window are excluded, so it is not a complete attack cost or safety rating.
- Order flow shows trading direction, accumulated volume and how prices moved with trades. Price-response fits are withheld when the sample is too sparse.
- VPIN summarizes how one-sided trading was in equal-volume buckets of indexed conditional trades. Eidos discloses scan coverage and large-trade effects. Its estimator is not interchangeable with every published VPIN measure, and imbalance does not establish informed trading or manipulation.
- Wallet concentration and history show how much volume particular accounts supplied and, where histories qualify, their funding, accumulation and stake recovery. Accounts are not necessarily separate people; concentration alone does not establish coordination.
These instruments describe the evidence behind the price. Any adjusted estimate or aggregate would require a defined target, justified inputs and a method evaluated against a baseline; none is implied by these measurements. Flow instruments require supported conditional-swap records; depth and history checks require their own inputs. Each reading states its coverage or why it is withheld.
How does evidence become intelligence?
The data records observations. The ontology defines the decision, its options, objective, governing rule and relationships to execution and outcomes. Versioned instruments calculate market readings, participation, trading patterns and historical comparisons. Interpretations connect supported results to the question a reader is trying to answer.
Each analytical result names its instrument, evidence and time. Applicability, availability, completeness and freshness are separate: a partial trade history can still describe observed participation, while a detector requiring the full history remains unavailable. An inapplicable check is not missing evidence.
A sourced interpretation states what was observed, what was derived, what is hypothesized and what could change the reading. A pattern match does not establish intent, and a no-match does not establish safety. Historical review records retain their original versions; current analysis is organized around the decision and the available instruments. Methodology explains the calculations.
How can the record improve?
A source-linked correction is reviewed before it changes a result. Acceptance recomputes the analysis and preserves the original input and output. Private contributions keep their sharing scope. User questions can identify missing capabilities, while resolved outcomes and fixed evaluations test whether methods improve. More usage creates opportunities for better evidence; it does not by itself establish better accuracy.
Organizations can add what the chain does not hold. Each organization's record has two columns: what the chain shows (the trades, the prices and the outcome) and what the organization says (what a decision is for, who is bound by it, what was committed and the documents behind it). The second column is attributed to the organization, checked against the documents it cites before it is published, and dated as it stood. An unanswered field stays empty; it is never filled from an older answer. Taking part does not buy coverage or change a finding.
What does a qualified interpretation look like?
Hypothetical illustration, not a finding about a covered decision. Suppose a token is priced higher in the approval branch, but the indexed trades show limited depth and most volume comes from a few wallets. Their earlier funding histories are unavailable.
The useful reading has three parts. The market prices the token more highly under approval at the stated observation time. That comparison rests on thin liquidity and concentrated recorded participation. The missing histories leave some attack checks unanswered.
The reader can distinguish a favorable price comparison from evidence of broad participation, identify the missing checks and explain why the interpretation remains limited. Neither a manipulation finding nor a smaller “corrected” price difference follows from those observations. Eidos retains the comparison and the qualifications side by side.
What happened after the decision?
Execution checks ask whether the approved action was carried out under a supported rule. Of 86 passed MetaDAO decisions, 65 are recorded as executed and 21 as not executed. Missing execution readings belong to neither count. Where execution moved assets, 51 decisions have reconciled capital movements.
Later observations track supported outcomes at stated horizons. Token-price histories vary by organization and fill forward as readings become available; other markets may resolve against a measured outcome. These observations tell a reader what followed. Establishing decision quality or causal impact requires an appropriate comparison, including evidence about the alternative that was not chosen.
How can you inspect the analysis?
Follow a reading to its decision page, proposal terms, source times and findings. Covered decisions share one form: what it was for, what was known, what was done and what followed. Indexed trade replays disclose scan coverage, so the reader can see how much of the window was observed. Retained readings preserve what was known at their recorded times.
The infrastructure serves that inspection: source collection feeds the record, reviews use its evidence, and the app and machine interfaces expose supported readings. The figure shows the implementation behind the analysis; its counts derive from the registries.
The ontology defines the record’s vocabulary. Public decision records also serve at /data/decisions.json; supported public objects can be requested by exact key. The MCP guide specifies tool coverage and access. Metering is not yet switched on, and until it is the metered read does not serve. Once it is, each caller receives ten free requests a day, after which each request costs $0.02 in USDC, payable on Base or Solana.
Coverage and customer use
Our read of MetaDAO's program excludes eleven test organizations that never drew five wallets to a window. On that basis, 22 organizations have run a market, 15 of them more than once, across 66 proposals. Of the 37 closed windows with enough prices to model, 32 passed. Because the verdict forms inside the window, the record is kept continuously rather than as a snapshot; the Thesis shows how agreement with the final decision builds over the window.
Across every engine we can source, at least 34 organizations on five engines have had a real-money decision market decide or select one of their own actions; 102 of their decisions are counted in the record. That census is the adoption the record exists to serve, and we track it alongside our own measure.
For each organization, Eidos answers the same fixed questions — who controls its treasury, its mint and its decisions, what it promised at its raise, and what is held only in documents — each with its date, its source and who asserted it, and marks the rest unknown.
The measure we manage the business to is the number of decisions made by outside organizations with an Eidos record attached before the window closes. It is currently zero. The infrastructure is in place; acquiring the first customers is the work ahead.
What still needs work
- What followed, at depth. Outcomes fill forward as each horizon passes, so they are thin today, thinnest where an engine's token prices have only begun to be recorded. Learning, the record's lesson across decisions, is defined and waits on enough measured outcomes to say anything.
- A customer-filed review. The pre-window review exists in the ontology and as a working scaffold, but no organization has yet filed one.
- The organization's own column and dated documents. Organizations can submit answers through the register intake for review; self-service maintenance is not available, and a submission is not a published statement. Proposal-document capture retains dated archive revisions where collection succeeds. Coverage is incomplete: a later capture cannot establish what Eidos knew while a window was open.
- Metering in operation. The payment rails are built and charge nothing until they are switched on, which requires only the operator's receiving address and facilitator keys.
- Every playbook's evidence, further back. All four monitored playbooks now run from chain data, but each runs only where its evidence is held: stake-origin tracing is bounded and can withhold where transfers depend on an unread sender history, while retained token prices after a close begin on 19 July 2026. Earlier decisions wait with that reason stated. Solana's governance votes follow.
- Execution where no rule applies yet. MetaDAO's execution is checked against the Squads transaction and Umia's against its own contract. Combinator and Butter carry out decisions outside their engines, so each reading there names the rule it used and how strong that rule is: a Squads transaction the proposal cites, a redemption vault opened after the close, or a grant payment verified onchain. A decision that matches none of these is published as not checked; covering it needs a rule not yet written. futarchy.fi's results bind nothing on chain, so its decisions are published as not checked by design.
The customer can inspect available analysis in the app or scope an organization’s first window through a pilot. Pro is on sale; fund subscriptions remain planned. Capital explains the offers and the controls that keep paying clients and traders from shaping findings.