An inspectable record

The evidence comes first

A useful track record explains what was said, how it was interpreted, and what the available evidence can support.

See coverage and freshness Report a correction

Start with the source

We process selected public videos and their available transcripts. Each call links back to its source. Publication time is separate from the time TubeRank first recorded it.

Separate words from interpretation

Quoted text is extracted from captions or a transcript. The asset, direction, target, deadline and thesis are our interpretation. Captions and AI extraction can be wrong; use the source moment to check context.

Evaluate a specific condition

A result describes a stored claim checked against available price observations. It does not prove that a creator made a trade, that a trade could be executed, or that the wider thesis was correct.

Keep corrections visible

Review status, recorded timestamps and any available correction history belong beside the claim. Missing historical provenance is shown as missing; a current rule version is not proof of how a legacy result was calculated.

Current outcome rules · 2026-10-06.5

A deadline must come from the claim

A horizon such as “short term” does not create a deadline. An otherwise eligible call without a stated deadline remains open and can expire after 12 months without becoming a miss.
AI-extracted claims need a source publication time. Prices used to establish an outcome must be observed after the claim and inside its stated window; a reference observation must come from at or before publication.
A directional target must be on the correct side of its reference price. Suspected instrument mismatches, unsupported conditions and unclear neutral claims are left unscored.
For a directional range, entering the band means reaching its near edge: the lower bound for a bullish call or the upper bound for a bearish call. The record retains the original range.
An observed target crossing can establish a hit for a touch target. Daily closes alone cannot prove that a target was never touched during a trading session.
A claim about the price at a future date, such as where a stock ends the year, cannot be proved by an earlier touch. These terminal-price claims stay unscored while that condition is unsupported.
Negative results and an unbroken ceiling or floor require evidence matching the claim. Complete daily closes can evaluate an explicitly close-based claim. Other claims remain unverifiable when the necessary intraday evidence is missing.
For an eligible close-based claim after its stated deadline, movement at least halfway toward an unreached target receives partial credit when the required coverage is present. Neutral boundaries need explicit language in the source quote.

These are the current rules for new evaluations. Check the methodology version stored on each result. Older outcomes without a version cannot be assumed to have used these rules. Daily-close conclusions cover full days after publication, excluding the partial publication day. Unsupported exchange calendars or incomplete evidence prevent a score.

Reading the numbers

Always keep the denominator

The weighted hit rate is (hits + 0.5 × partials) ÷ (hits + misses + partials). Pending, expired, unverifiable and withdrawn calls do not enter that denominator. Total calls and scored calls answer different questions.

Example: 2 hits, 1 partial and 2 misses give a weighted hit rate of 50% across 5 scored calls. Another 95 pending or unverifiable calls would make 100 total calls, but would not make this a 100-call scored sample.

A hit rate is shown — for a creator or a ticker — and a creator is ranked only from 10 scored calls; below that the record reads “3 resolved, too few to rate”. Every rate names its denominator, such as “50% of 12 resolved”. That is a display threshold, not statistical confidence: 10 scored calls can still be a small, unrepresentative sample. Coverage is selected, related calls may repeat the same thesis, and historical results do not predict future performance.

Outcome vocabulary

What each label means

Hit: The recorded target or condition met the evaluation rule. Read the stored explanation and price window; this does not validate every argument in the video.
Miss: The stored evaluation marked the condition as not met. An agent preparing a “missed after deadline” brief must require an actual stated deadline; a broad horizon or still-open target is different.
Partial: The rule awarded partial progress rather than a full hit. It contributes one half to the weighted numerator. See the individual record for the applied rule.
Pending: No final result is recorded. This can mean the evaluation window is open or evidence is incomplete; it is not a promise that the target is still attainable.
Expired or unverifiable: The claim was not scored as a hit, miss or partial. Unclear conditions, unavailable prices, unsupported instruments or missing timing can prevent a defensible result.

Limits of the evidence

Historical checks use stored price observations, including daily close backfills. They do not reconstruct every intraday high, low or available execution price. Provider quotes may be delayed.
Price movement is a comparison between observations. It excludes position sizing, fees, spread, slippage, dividends, borrowing costs and whether a trade was made.
Stock splits and asset aliases require interpretation. An index, an ETF tracking it and a similarly named company are different instruments.
Ranges, conditional claims and ambiguous deadlines need care. Legacy records can reflect earlier rules, including inferred horizon windows; the stored explanation and correction history take priority over assumptions.
Published records may not have been reviewed by a person. A moderator review is an editorial check, not independent assurance or a certification.
We do not cover every channel or video. Failed transcript retrieval, unavailable sources and processing delays create gaps. A missing call is not evidence that no call was made.

Challenge the record

If an asset, quote, deadline or outcome is wrong, use Report an error on the call, or include the call URL, source timestamp and proposed correction in your report. We can review and correct the record. Corrected outcomes are listed on the public corrections log and in each call’s history. We do not promise an independent audit, complete historical change tracking, or a response deadline.

Report a problem Use the agent interface

Agent access does not grant ownership or redistribution rights in source content or market data. See our terms. TubeRank provides research context, not investment advice.