Certified v4.2

The method

A ranking is only worth reading if you can see how it was built. Here is the formula, the eighteen things it scores, the four terms on top and the four source defects repaired underneath, how much research sits behind any one number — and the list of things this model is not able to know.

The base score

Three groups, eighteen scored factors, one weighted sum. This is the only layer of the model with published exact weights, and it is the only layer this site can re-derive from the factor scores:

Each group is the mean of the factors listed under it.

Craft — 56% of the base score

What happens in the ring, and whether it holds up.

• In-Ring Execution
• Psychology / Storytelling
• Selling
• Opponent Uplift / Carry Factor
• Peak Match Quality
• Match Catalog Depth
• Big-Match Reliability

Star — 29% of the base score

Whether anybody cared, and how far past the arena it carried.

• Mic Work
• Character
• X-Factor / Aura
• Crowd Connection
• Drawing Power also feeds business impact
• Peak Mainstream Crossover also feeds business impact

Career — 15% of the base score

What was still true after the run was over.

• Sustained Career Relevance
• Adaptability
• Influence / Legacy also feeds business impact
• Major-Promotion Title Equity
• Era Dominance / Recognition

What sits on top of the base, and in what order

The published ranking is not the base score. Four terms sit on top of it, and this site computes every one of them — the algorithm is ours, it is published in full, and anyone who disagrees with a placing can point at the line that produced it. In the order they are applied:

• Elite craft. A bounded bonus, at most 1.6 points, for the genuinely extraordinary complete worker — it starts at an in-ring craft score of 93 and saturates fast. It is applied first and in full, because it is a ceiling correction: its whole job is to keep an extraordinary worker distinguishable from an attraction at the very top of the scale, which is exactly where there is least room left to say so.
• Era-adjusted business. Built from the reference model’s one published set of business weights — 70% Drawing Power + 10% Mainstream Crossover + 10% Influence + 10% Crowd Connection — with two changes stated below: drawing power is de-saturated before it enters, and the whole is scaled by how much of the house the wrestler personally drew in their peak decade (1.00 in the 1950s down to 0.75 in the 2020s, when the brand draws more of it). Only the elite end earns anything, and what it earns is a fraction of the headroom left above base-plus-craft, so a wrestler already judged 98 on the factors has very little left to be paid for drawing, and nobody can pass 100.
• The interrupted career. A small credit for a career that was demonstrably at or near its peak and was stopped by a documented event — an arrest, a substance-related release, a career-ending injury or a death at the top. Twelve per cent of the remaining headroom where the run was cut short and the career continued diminished; eighteen where it ended outright; six where it was interrupted and largely recovered. The credit is scaled by how clearly the record shows a peak — a top PWI 500 placing or a world title — so the table that names a career cannot on its own pay it anything, and it can never exceed the thin-evidence penalty. Allegations do not qualify; nor, ever, does violence against another person. The event is stated on the profile in one neutral clause, and no stance is taken on it.
• Thin evidence. A flat 2.4-point penalty, not a cap, on every row whose factor set was never individually researched. It is the only term that reads the evidence tier, and it is why the top of the list is the top of the researched set.

What is deliberately not here. The reference model also carries a “future potential” term for careers in progress. This ranking does not: it is a claim about what a career has done, and a guess about runway would be the one number on the page nobody could argue with, because nobody could check it.

Four defects in the source data, repaired at the input

Each of these was measured against the snapshot rather than assumed, and each is fixed where it arises rather than papered over with a correction on top — because a fudge factor moves the numbers without fixing what produced them, and the next reader has no way to tell which.

1. The group step ignored its own weights. The workbook publishes eighteen factor weights summing to 100 and to 56/29/15 by group, then averages each group as a plain mean — 16,830 checks, no exceptions. So “Historical Match Peak”, stated at 10, really carried 8, and “Title Equity”, stated at 2, really carried 3. Weighting as stated also moves the model measurably closer to its own published anchors, so this is the arithmetic the source meant to do.
2. Title equity knew about three promotions. The factor is named for WWE, AEW and TNA and it is the only championship signal in the model, so eight NWA World Championships scored below one modern midcard run. Replaced, where the editorial table has a standing, by a world-title equity across every lineage this site tracks, with diminishing returns: the step from no world title to one is worth far more than the step from a fourth reign to a fifth.
3. “Era dominance” was a recency proxy with a gender problem. It tracked inverse PWI 500 rank, and the archive starts in 1991 and is the men’s list — so 2,904 of 5,610 wrestlers sat on its floor, including 155 of the 166 women who have held a women’s championship. Replaced by a dominance signal that rewards peak seasons where the archive covers the peak, and falls back to the wrestler’s own crowd, draw and influence where it cannot. A missing archive is our gap, not theirs.
4. Drawing power saturates. The column tops out at 100, and the people who genuinely moved the industry are all pressed against it: Hogan, Austin, Rock, Cena and Sammartino at 100, Dusty Rhodes and Flair at 99. A column that puts Hulk Hogan one point above a man who headlined one good year cannot carry the fact a business term exists to carry, and with the era table on top of it the eighties were being paid for being the eighties. De-saturated — squared, so 100 stays 100 and 95 becomes 90 — before only the business term reads it, and the era table was compressed to match. The base score still takes the column as judged.

Nothing here is tuned to produce a desired order. There is no per-wrestler override and no term whose purpose is to move one name. The two hand tables that name people — world-title standing and interrupted careers — supply facts the source lacks, and each is gated by the record before it moves a score. The reference model’s published scores are used for exactly one thing, the thing its own method prescribes for outside lists: a sanity check, never a target.

How much research is behind any one score

Not every row is equally researched, and the ranking says so on every row rather than in a footnote. Three tiers travel with the numbers:

• Calibrated deeply researched, argued over, and the tier the leaderboard is built from.
• Provisional indexed and scored, not individually re-examined. Treat a provisional score as a placeholder with a number on it.
• Historical Index careers that peaked before the PWI 500 began printing in 1991, so the archive this model leans on cannot see them at their height. Placed from what the historical record supports. It is not a lower grade than provisional — it is a different gap, the record’s rather than ours.

A ranking that prints thousands of scores in one undifferentiated column is claiming a depth of research that exists for a fraction of them. The tier is the correction for that, and it is the reason this page exists.

Championships

Title equity is weighted by promotion, tier and era, with diminishing returns — so a twentieth reign is not worth what the first was, and belt proliferation in one decade cannot outweigh a career in another. World titles carry full or near-full weight; secondary, tag, trios, specialty and novelty championships are discounted. Where two promotions recognised the same reign, it is counted once.

The lineages tracked:

• WWE/WWF/WWWF
• WCW
• NWA
• AEW
• TNA/Impact
• ROH
• ECW
• NJPW
• AJPW
• NOAH
• AWA
• CMLL
• AAA
• Stardom
• Original World Heavyweight lineage

What the PWI 500 is used for

Thirty-five years of Pro Wrestling Illustrated’s annual list, 1991 to 2025, sit behind this project — 17,500 placements. It is used as a discovery and career-recognition input: it is how a name enters the universe at all and evidence that somebody was visible in a given year. It is not a vote on greatness, and its density in recent decades is precisely the modern-era bias the historical correction exists to undo.

The rules the model holds itself to

• No wrestler-specific rank override.
• External GOAT lists are sanity checks, not targets.
• Jim Cornette's public philosophy is a calibration input, not the ranking authority.
• PWI 500 is a universe/career-recognition input, not a direct GOAT vote.
• Modern data-density bias must be corrected.
• Drawing power, crossover, X-factor, crowd connection and sustained relevance remain distinct concepts.
• Jointly recognized title lineages should be de-duplicated for scoring when appropriate.
• Broad database evidence tiers must remain visible; provisional rows are not equal-confidence to calibrated rows.

What this model cannot know

Worth saying plainly, because a two-decimal score projects a confidence no ranking of an art form has earned:

• It cannot watch a match. Every craft factor is a human judgement entered as a number. The arithmetic is reproducible; the judgements are opinions, and opinions with decimal places are still opinions.
• The record is thinner the further back it goes. Gate receipts, television ratings and match results survive unevenly, and territories that never had a national television deal are underrepresented in every source this draws on.
• Drawing power is contaminated by everything around it. A wrestler on top during a boom and a wrestler on top during a bust are not being measured on the same axis, however carefully the era correction is built.
• Careers in progress are guesses. That is exactly why this ranking carries no future-potential term at all: a wrestler still working is scored on what the record shows so far, and the record will move the score when it moves.
• A ranking is an argument. This one is published with its formula, its factors and its uncertainty attached so the argument can be had properly. Disagreeing with it is the point.

How the ranking has changed

Every revision below moved a published score, and each did so because a specific defect was measured and fixed, or a term that had shipped as a scaffold was populated with verified data — never because a name needed to move. The formula is public; disagreeing with a row means pointing at the line of code that produced it, and this table is the record of which lines changed and what happened when they did.

This is the formula’s own history, sourced to the commits that changed it — not a tracker of how any one wrestler’s rank has drifted release to release. goat.json.gz is a static, dated snapshot regenerated from scratch on every change, not a stream with a saved past state, so a wrestler’s exact rank a year ago is not a question this page can answer yet. Say so rather than imply otherwise.

Sources

The scored model is The Workrate Report, certified version v4.2. Career and championship records draw on Pro Wrestling Illustrated, profightdb and thesmackdownhotel. Every page in this sector is generated from one data file and nothing is hand-typed into the HTML, so a corrected score is one regeneration away.

Facts only, and deliberately: names, reigns, dates, placements and scores. No photographs, no promotional artwork, no logos, no lifted biographies — naming a promotion and a championship to say whose title it was is nominative use, and reproducing what they made is not.