Track Record Methodology

A scorecard reading 99 resolved calls, then 67 of 99 winners, then 53 of 67 naming a method and 34 of 53 getting it right, describes one narrowing sample, not four separate ones. The 53 is how many of the 67 correct calls also named a method on a bout whose result records one; the 34 is how many of those 53 named the right method. Round narrows the same way and is counted independently of method, but what makes a round call eligible is not the same test — see Round called above.

Every bar on a scorecard is drawn against the top of its own funnel — the resolved-call count — so a bar’s length is its share of that one number, and bars shorten as you go down a branch. Method and round are two separate branches, each starting again from the correct calls, so a round bar can be longer than the method bar above it without either being wrong. A bar measures how many calls, never what share were right; the share is always written out with its denominator beside it. This is deliberate: 34 of 53 and 67 of 99 are different rates of different populations, and drawing them at their own scales would invite a comparison that means nothing.

Where a step shows a sentence instead of a number, its denominator is zero, and the sentence says which of two different things happened. “Calls the winner, not the round” means the analyst has never named a round — a style, not a failure. “Named rounds — none on a correct call yet” means they do name rounds, but none has yet landed on a correct call whose result could settle it. Neither is shown as “0 of 0” or as a zero bar, because a zero would state something false in both cases. A genuinely earned zero — a 0 of 1 on someone who named a method on their one qualifying call and got it wrong — keeps its real numbers and is never hidden.

On 19 August 2026, across the 26 analysts with a shown record, the resolved-call counts ran from 5 to 99, with a median of 27.5. A record of 67% correct over 28 calls is consistent with a true rate anywhere between roughly 49% and 82% (a 95% Wilson score interval; the ranges on the board apply this same interval to an analyst’s hit rate and then scale it by the credit they earn per correct call, so it is an approximation rather than an exact interval on the Score) — a span of about 33 points. The same 67% over 99 calls narrows to about 57% to 75%.

The practical consequence: at these sample sizes most differences between two analysts are not distinguishable, and a higher percentage on a smaller sample is not evidence of a better analyst. This is why every figure on the site is published with its denominator, why records below five resolved calls show no ratio at all, and why a larger sample is worth more than a higher percentage when comparing two records.

For a bout to qualify, the consensus must be captured before its event starts, frozen at that boundary, and be less than two hours old. The selected fighter must have a strictly lower consensus chance of winning than their opponent. Tied markets do not create an underdog.

We use the frozen pre-event consensus so later market movement cannot change a historical underdog result. The metric does not use a price at the moment an analyst made a pick.

The analyst directory’s Market lean row splits an analyst’s picks into underdog calls and favourite calls, each with its own hit rate. A pick qualifies once it has a pre-fight price: the frozen pre-event sportsbook consensus described above when one was captured, or, when it was not, a price reconstructed after the fact from Polymarket’s public trading history for that fight. The row appears once at least five of an analyst’s picks, underdog and favourite combined, qualify this way.

This is a wider net than Correct underdog calls and the Score’s dormant underdog credit below, both of which use the sportsbook consensus only. Market lean exists to show a split as broadly as the available price history allows; it does not feed the Score.

Start with the winner. Call the winner in 34 of 48 resolved fights and you start at 70.8%. That is the floor; everything else only adds.

Add for detail that lands. A correct winner call that also named the right method adds 0.25. Naming the right round adds 0.25. One call can earn both. Getting the detail wrong costs nothing at all: naming a method or a round can only ever help you, never hurt you, so an analyst is never punished for being specific.

Divide by every resolved call, then cap at 100%. Bonuses can push the arithmetic past 100; we stop it there. If two analysts both reach 100% they share the position, because the published numbers are identical; the one whose uncapped total was higher is listed first inside it.

Twenty resolved calls minimum to hold a position. Below that an analyst is listed with their record and no Score at all — a short record is a noise floor, not a verdict, and at these sample sizes the cap below starts doing the work. We use 20 rather than 10 because at 10 the 100% cap starts rewarding the smallest records, which are the ones we know least about.

score = the smaller of 100 and 100 × ( winners + 0.25 × method hits + 0.25 × round hits + 0.50 × underdog credit ) ÷ resolved calls

The underdog term is currently zero for every analyst, so today the sum is the first three terms only — which is why the worked example below has three lines and not four. What it will do when it is switched on.

TheWeasle — 48 resolved calls, as their record stood on 19 August 2026. Their live record is on their scorecard and will have moved on since; the arithmetic below does not change.

The Score reserves credit for correctly backing a fighter the market priced below an even chance, and that credit scales with how unlikely the market thought it was. Backing a fighter given a 30% chance is worth far more than backing one given 49%, because a 49% pick is a coin flip in all but name. The scale is (0.5 − chance) ÷ 0.5: a 49% call earns 0.02, a 30% call earns 0.40, a 10% call earns 0.80.

We do it that way because counting every underdog call equally would pay the analyst taking the least risk almost as much as the one taking the most, which is the opposite of what the credit is for.

This component is currently inactive and contributes nothing to any published score. It needs a market price frozen before the fight for every call it touches, and the scoring path does not yet read the price history that would supply them. Rather than pay the credit out on the fraction of calls that happen to carry a price today, we hold it at zero for everyone. The rule itself was settled on 19 August 2026; switching the component on is a separate, dated step. Whether this credit is active is a different question from whether an underdog/favourite split is shown at all: the directory’s Market lean row appears once at least five of an analyst’s picks, underdog and favourite combined, carried a qualifying pre-fight price — and unlike this component, that display also accepts a price reconstructed after the fact from Polymarket’s trading history. When the component is switched on, it will be a dated version change and no existing figure will be restated.

It measures accuracy, not edge. In the UFC the betting favourite wins roughly two thirds of fights. An analyst who simply picks the favourite every time will therefore score well without any skill at all. A high Score means an analyst is reliably right and willing to be specific. It does not mean they are beating the market. An analyst who beat the market decisively can score below one who merely matched it, because picking favourites is the higher-accuracy strategy. We do not currently publish a measure of edge, and this number must not be read as one.

No analyst on the board is currently distinguishable from any other. Across the analysts who clear the 20-call floor, every plausible range overlaps every other, including the top and bottom of the table. Statistical tests agree: the spread between these analysts is no wider than chance alone would produce, and on one test it is narrower. The ranking is the best estimate available from the record; it is not evidence that the analyst at the top is better than the analyst at the bottom. This is stated on the board itself, not only here.

Accuracy and edge are never combined. The Score is a blend, and the formula above says exactly how: winner, method, round and underdog calls are components of one published, defined number, not separate metrics quietly averaged into a rating. What it never blends in is edge. Any future measure of whether an analyst beat the market will be its own board with its own denominator, and will never be folded into this number.

Until August 2026 this page stated: “There is no composite Analyst score or overall ranking today. If FightCardIQ publishes one in future, it will explain which published metrics it uses, how it handles sample size and market context, and when any version changes take effect.”

A ranking now exists, so the first sentence is retired with this dated note. The promise in the second sentence is kept in full above, and it is worth being explicit that it was met on its own terms: every input that feeds the Score today — winner, method and round — was already published on this site with its own visible denominator before it was allowed to feed a ranking. The fourth, underdog, is defined here but currently feeds nothing into the Score. A related but separate display, Market lean, does show an underdog/favourite split on the directory; it is not this component and does not feed the Score either. Sample size is handled by the 20-call floor and by publishing the range every score is consistent with. Market context is handled by keeping it out: the underdog component is switched off, and edge is not measured. Version changes take effect as described below.

When a rule changes — a new metric, a changed eligibility test, a different floor. The version number increases and the change is dated here. Figures published under an earlier version stand as they were; they are not retrospectively restated, because the record of what we said at the time is itself part of the record.

When a data correction is made — a mis-logged pick, a wrong result, a source that turns out not to support a claim. The affected figures are restated in full and the correction is marked on the analyst’s own page. Corrections are not versioning; they are fixes, and they are never quietly absorbed into a new version.

Version 3, 19 August 2026. Published the FightCardIQ Score, its formula, weights, floor and worked example, and retired the statement that no overall ranking exists. No existing metric definition changed and no previously published figure moved: the Score is derived from figures already on the site. The underdog component is defined but inactive.

Version 2, 18 August 2026. Added the scorecard funnel to the analyst directory and defined how to read it; documented the two reasons a metric can have a zero denominator; published the sample-size arithmetic above. No metric definition changed and no published figure moved.

Version 1. Winner, method, round, perfect and underdog calls as defined above.