WHAT THIS IS

This skill runs one method: the way a fight analyst with fifteen years of
watching MMA reads a card. Not "an AI that knows who wins" — a framework that
applies the same fifteen-year judgment to every fight, the same way, and shows
its work.

Give it one matchup. It scores the fight across six weighted factors, checks it
against a short list of reasons for caution, and delivers a confidence tier
with the reasoning laid out line by line. It covers UFC and RIZIN cards (RIZIN
is where third-party English analysis is thin — that's on purpose).

HOW IT READS A FIGHT (the six factors)

1. Proven level — not the win-loss count. The level each fighter has
   demonstrably beaten, judged from who they beat and how.
2. Momentum — are you winning in your own style, or scraping by.
3. Style matchup — technical fit only: striking vs grappling, finish
   patterns, takedown defence. Numbers, not reputations.
4. Physicals — weight cut, age, reach, height.
5. Setting — home vs away, travel, altitude, crowd.
6. Short notice — a late replacement changes the fight for both men.
   It is scored, not ignored.

On top of that, it applies "confidence caps." Heavyweight bouts, five-rounders,
and light-heavyweight fights with no readable skill gap all get their confidence
capped — because those are the fights where a clean read is hardest.

One more rule the framework lives by: if an opponent's level cannot be verified,
that scores zero — never a minus. Not being able to see is our limitation, not
the fighter's weakness.

THE OUTPUT: FOUR TIERS, INCLUDING "NO CALL"

• High Confidence — the structure is close to decided.
• Moderate Confidence — one fighter is clearly ahead.
• Low Confidence — it leans, but there's a live path the other way.
• No Call — this one isn't readable cleanly. This is a real answer,
  not a failure. A framework that never passes on a fight
  is a framework that isn't being honest.

WHAT IT DOES NOT DO

• It never touches money. No prices, no market quotes, no links to any site
  that takes money on a fight, no sizing, and it never combines two fights
  into one call.
• No win percentages. It will never promise to be right — no honest fight read
  can promise that — and it never talks about money.
• It is fight analysis and reasoning. It is not betting advice and not
  financial advice.

INPUT

Fighter A and Fighter B, weight class, scheduled rounds (3 or 5), event, and
any late context (weigh-in result, short-notice replacement, layoff, camp
change, home vs away). The more complete the input, the sharper the read —
and where data is missing, it says "insufficient data" instead of guessing.

OUTPUT

A fixed-format breakdown: the six-factor table, the weighted total, the market
cross-check (noted when the read diverges — it never moves the tier), any
confidence caps that fired, a "why this could be wrong" line, and a final tier
— with the specific facts behind each score. Two worked examples below (all
fighters and cards are fictional).

DISCLOSURE

AI-assisted analysis. This tool applies one human analyst's fifteen-year
framework; it is not an oracle — it can be wrong, and it promises nothing about
results. For entertainment and analysis only. Not betting advice. Not financial advice.
If gambling is a problem for you, please seek local support resources.

SAMPLE OUTPUT — two fictional cards

Sample 1 — a clean read on a card with no market data (RIZIN-style)

Rui Tanaka vs Marcus Vela — Meridian FC 12 (illustrative / fictional)

Fight analysis only. Not betting advice.

Read: Tanaka is favored. The level he has actually beaten, the shape of his
wins, and the grappling numbers all point the same way — and none of it depends
on a market that is not there.

W: +11.0 (favored: Fighter A — Tanaka)
Band (by |W|): High
Market cross-check (7): no market data (scored on elements 1-6 only)
Hard Veto: none
Bias check: B1 — if this fighter were not Japanese, would element 1 read the
same? Yes: it rests on the named opponents' records at the time, not on his
nationality.
Demotion applied: none
Self-checks: C-1 / C-2 / C-3 / C-4 — all pass
Why this could be wrong: Vela's power is real — three of his wins are
first-round stoppages — and Tanaka's entries take him through the pocket. If the
clinch is denied early, the shape above does not survive the first ten minutes.

Tier: High Confidence

Sample 2 — a favorite the framework still will not call (heavyweight, five rounds)

Bruno Alkan vs Viktor Sable — Vanguard MMA 7 (illustrative / fictional)

Fight analysis only. Not betting advice.

Read: Alkan is ahead on what he has proven — but this is a heavyweight fight
over five rounds, and nothing in the grappling or submission numbers separates
them. This is a No Call. That is the framework working, not failing.

W: +4.0 (favored: Fighter A — Alkan)
Band (by |W|): Low
Market cross-check (7): no market data (scored on elements 1-6 only)
Hard Veto: hits (V1 heavyweight, V3 five rounds) -> ceiling: No Call /
V1 exception: not applied — no clear wrestling, submission, or proven-level
advantage on either side to relax it
Bias check: none (flags.japanese not set for either fighter)
Demotion applied: Hard Veto ceiling, 2 hits (effective: Low -> No Call)
Self-checks: C-1 / C-2 / C-3 / C-4 — all pass
Why this could be wrong: both ways at once — either man's right hand can end
it inside a round. That is exactly why this one is a pass and not a lean.

Tier: No Call

HOW TO USE

1. Ask for this week's card: "Break down this week's card." Every fight comes
   back as its own block, plus a closing list of the fights we pass on.
2. Or name one fight: "Read Fighter A vs Fighter B." One block, same format.
3. Add late context if you have it — a weigh-in result, a late replacement, a
   long layoff. Facts sharpen the read; the skill never invents what you
   don't give it.
4. Read the tier AND the reasoning. The "Why this could be wrong" line is part
   of the product, not a footnote.
5. Expect a No Call sometimes. That is the framework being honest about a
   fight it cannot read cleanly — not a malfunction.