Belief, Provenance, and Outcome Methodology

A BeliefState record needs an attributable identity or explicit identity uncertainty, a security or thesis, a source, a publication or observation time, and exact supporting evidence. Commentary that cannot support a specific interpretation remains unstructured or is excluded from scored research.

Every interpreted field records its evidence, confidence, model version, prompt version, and methodology version. Public discovery also requires an accepted review state and a separate append-only publication event whose latest scope is public.

• Every record stays linked to its source identity, canonical URL, publication time, observation time, and exact permitted evidence span.
• Source evidence remains separate from model interpretation. Author confidence, model confidence, and measured historical skill remain separate values.
• Point-in-time queries exclude evidence unavailable after the requested cutoff. Unknown, ambiguous, stale, and insufficient data remain explicit.
• Evidence, source revisions, model output, belief revisions, publication, methodology, outcomes, tombstones, and audit history are append-only.

A scored position enters at the first eligible market price at or after the source timestamp. During market hours, that is the captured quote when available; otherwise the first eligible split-adjusted session price is used. Each market refresh updates the security return, SPY return over the same active period, direction-adjusted excess return, and favorable and adverse excursion.

Each position receives equal weight in its creator’s score. An add attaches a new equal-weight entry lot to the existing position; hold and trim events remain linked lifecycle receipts. A confirmed sell closes a bullish position at its first eligible market price. A position also closes at a direction-adjusted loss of 100%; gains are not capped.

The public leaderboard ranks every creator with at least one measured position by the average of each position’s peak observed return, with every position weighted equally. It does not classify positions as wins or losses. Each row exposes the share of tracked market sessions above entry, maximum running drawdown, and path efficiency. SPY-relative excess and the latest return remain supporting context. Unpriced calls remain visible as awaiting entry price, and uncertain trim or sell language can carry a small possible-exit flag without closing the position.

Discord scoring begins with a human-authored alert from a channel selected by an authorized server manager. The parser must resolve one long, single-leg U.S. option contract, one author identity, and a timestamped entry or exit event. Bot and webhook messages are rejected. Ambiguous or unmatched events remain needs-review or unscored records.

Version 1 public records require one active, nonsuperseded entry and a standard contract representing 100 shares with no additional deliverables. Open and closed trades enter the public option leaderboard. Pending or incompletely priced trades remain visible but do not change rank; disputed, corrected-away, or nonstandard records remain excluded.

Methodology discord-option-nbbo-v1 selects the first eligible positive bid-and-ask quote at or after each Discord event and no more than 15 seconds later. A long-option entry uses the ask; each partial or final exit uses the bid. Partial exits are weighted by their recorded exit fraction to calculate the closed trade’s percentage return.

Every entry and exit needs exact timestamped quote proof. Missing proof remains pending market data. Provider failure to resolve the exact historical contract, adjusted deliverables, excess exit fractions, or more than one active entry makes the trade unscored. Quote marks are a reproducible convention, not proof of an actual fill or account return.

The public Discord option leaderboard groups all canonical trades by person. Total tracked return adds every verified fresh endpoint return; records with pending, stale, or unverified pricing remain visible but do not enter the total. This cumulative call-performance measure is not account P&L because source evidence does not include position sizes or capital allocation. Any profile with at least one measured return receives a numeric rank. The public record also exposes the share of trades whose observed peak reached the opportunity threshold, open-portfolio return, and path quality. Verified bid-side realized return remains separate and determines payout evidence. For an unclosed historical trade, the endpoint return is the return at the end of its stored market window—not a current quote.

Opportunity paths use one-minute executed-trade OHLC bars. The bar high records a real reported trade, but does not prove that the size was available or that a follower could execute there. Each record exposes provider, exact provider contract, bar count, active window, provider observation time, and data delay. Path analytics include running drawdown, entry crossings, profitable-to-nonprofitable round trips, profitable minutes, time to first profit, peak and drawdown times, and path efficiency. Path efficiency compares the absolute endpoint move with total bar-to-bar travel; a direct gain or decline scores higher than a choppy path covering the same net distance.

A captured statement does not prove that its author acted on it, maintained it, or achieved the measured result. Benchmark, timing, corporate-action, liquidity, and methodology choices can materially affect results. Discord option marks also omit commissions, fees, portfolio sizing, taxes, assignment, and liquidity beyond the observed quote. Historical outcomes do not predict future performance.

Originality, influence, confidence, and repeatability are bounded, probabilistic findings. BeliefState reports indexed-corpus limits, sample size, uncertainty, and methodology version instead of making absolute claims from one record or outcome.

Questions about a record or this methodology: antvacation@gmail.com.