Define the contract
Record the exact outcome, cutoff, resolution source, cancellation terms, and settlement rules before comparing venues or prices.
Methodology / version 1.0
FORM is designed to make probabilistic judgment inspectable. This document states what qualifies as evidence, when an idea becomes a published position, how results are measured, and how errors are corrected.
01 / FORECASTING
Record the exact outcome, cutoff, resolution source, cancellation terms, and settlement rules before comparing venues or prices.
Establish a relevant base rate or comparable set. If no defensible reference class exists, say so and widen uncertainty.
Separate evidence for the forecast from the strongest countercase. Every source is classified, timestamped, assigned an independence group, and scored for authority, relevance, and freshness under FORM-source-quality-v1.
Publish a probability and stress it downward for model error. Confidence describes evidence quality; it is not a substitute for probability.
Convert the contract price into an implied probability, add estimated execution cost, and require the edge to survive the stated stress test.
Set a maximum-loss ceiling before action. Related markets share a correlation group so apparently separate positions do not masquerade as diversification.
02 / PUBLICATION
A sourced idea that has not cleared every position gate. It expires within 24 hours and remains in the audit trail as an expired watch.
A timestamped position with a fresh quote, market and rules sources, probability, thesis, invalidation condition, confidence, and disclosed price.
A settled position linked to public resolution evidence and, when available, the original execution receipt. Closed records are cryptographically sealed against silent edits.
No public position. A positive base-case edge is insufficient when costs, uncertainty, evidence quality, settlement ambiguity, or concentration fail the discipline gate.
Active positions and watches must use a quote observed no more than 15 minutes before publication. FORM does not label manually entered prices as live.
Supporting evidence must score at least 8, counterevidence at least 5, and the complete brief must span at least two independent groups and distinct source domains. An active position also requires an official or primary source. Scores qualify research structure; they do not prove a claim is true.
03 / THE RECORD
Settlement payout minus verified cost. No hypothetical fills and no unverified entry prices.
Verified profit divided by verified cost. It is not annualized and does not represent portfolio return.
The squared difference between the pre-outcome YES probability and the binary result. Zero is perfect; one is worst.
The count-weighted absolute difference between mean forecast and observed YES rate within disclosed probability bands.
Change in the YES midpoint between immutable snapshots of the same contract. Every snapshot preserves bid, ask, last, spread, volume, open interest, timestamp, collection method, source digest, and chain hash.
Excluded from win/loss and profitability totals, retained in the ledger, and labeled with the venue's resolution evidence.
Fewer than ten resolved forecasts is explicitly labeled too small for performance claims. Longer records may still fail to generalize.
04 / GOVERNANCE
A material factual or calculation error is corrected promptly with the prior value, corrected value, reason, and correction timestamp preserved. A losing position is not an error merely because it lost.
FORM discloses material positions, sponsorships, affiliate relationships, or other incentives connected to a published market. Compensation never converts a watch into an active position or removes a result.
Contract rules and official resolution sources outrank summaries, social posts, and commentary. When sources conflict, FORM withholds publication or states the disagreement.
Forecasts can be wrong. Prices can move before a reader acts. Fees, liquidity, limits, taxes, venue access, and settlement risk vary. FORM provides research and educational tools, not individualized financial advice or execution.
Material methodology changes receive a new version and effective date. They apply prospectively; historical records are not rewritten to fit a newer standard.