How Blueprint MMA works

See the pick. Understand the fight.
Check the result.

Follow official weekly picks for scheduled fights, or build a Custom Matchup Report—powered by the Oracle Engine—for a current, future, or hypothetical fight. Official picks are graded publicly; custom reports stay separate.

See the free picks Check the resultsBuild my first Custom Report free
Free official pickTwo opposing casesPublished pre-fightPublic results

The simple version

Four steps. No technical background required.

Choose a path, see the value first, pay only for the depth you want, and verify every official result afterward.

  1. 01

    Choose how you want to use Blueprint

    Follow a real fight week for official picks, or choose two supported fighters for a current, future, or hypothetical Custom Matchup Report.

  2. 02

    See the official call free

    For published scheduled fights, the winner, win chance, publication time, and public result are open to everyone. No payment details required.

  3. 03

    Choose only the depth you want

    Use an Event Pass for one published card, Sharp for every official full breakdown, Whale for market tools, or a report credit for one fresh Custom Matchup Report.

  4. 04

    Verify the result afterward

    Official scheduled picks become wins, misses, or voids on the public Track Record. Custom Matchup Reports stay separate and never inflate it.

See value before you buy

What is free—and what paid access adds

Always free

The call and the proof

  • Official winner and win chance
  • Publication time and current status
  • Public Track Record, including misses
  • Fight schedules and fighter records

Choose your access

The reasoning and custom reports

  • Likely method and round
  • Primary and alternate paths to victory
  • Full matchup reasoning and evidence
  • Fresh Custom Matchup Reports for current, future, or hypothetical fights

A free account includes one starter custom report after email verification. No card required.

Event Pass covers one published card. Sharp covers every official full breakdown. Whale adds the Betting Desk. Report packs add Custom Matchup Reports without a subscription.

Compare access

Behind the pick

What happens before a call goes live

Official picks and Custom Matchup Reports use the same two-advocate and adjudicator approach. Official fights are published and graded; custom reports stay private and separate. The schedule shows exactly which official fights are ready.

Confirmed Card

the right fighters and rules

The schedule, fighter names, bout order, rounds, and publication deadline are checked before analysis begins.

Saved Research File

the same sources for both cases

One saved evidence file records the research cutoff, keeps sportsbook prices out of the official pick, and marks missing information instead of inventing it.

OpenAI Debate

two Luna advocates, one Sol adjudicator

Two OpenAI Luna advocates work in parallel, each building one fighter’s strongest case from the same evidence. Sol weighs both cases, counterevidence, and uncertainty, asking which fighter can actually establish their preferred fight. Blueprint checks the raw model probabilities without learned calibration, lens weights, or market adjustment.

Final Check

one coherent fight report

Blueprint checks the evidence and probabilities, then publishes only when the free pick and every paid section agree.

100%of each published fight must agreeNo mixed lineups. No free winner disconnected from its full breakdown. Unready fights stay clearly labeled.

For curious readers

Open the exact seven-step process

+

The plain-language details from card confirmation through public grading.

  1. Check the schedule

    After the previous card is settled and reviewed, Blueprint can begin the next confirmed card. It checks the lineup, bout order, rounds, main event, and deadline, with a 72-hour backstop.

  2. Save the confirmed card

    The lineup becomes one saved version. Duplicate fighters, substitutions, or incomplete identities stop the process before analysis.

  3. Build the research file

    Each fight gets a separate, source-backed evidence file with a clear cutoff, citations, disagreements, and missing categories.

  4. Test both fighters’ cases

    Two OpenAI Luna advocates work in parallel, each making one fighter’s strongest supported case. Sol then weighs both arguments and conflicting sources, asking who can establish their approach in open space, against the fence, and on the ground.

  5. Check the final call

    Blueprint verifies source citations and probability consistency. The raw model forecast remains unchanged; method and round chances are labeled separately from the overall chance of winning.

  6. Publish each ready fight

    The official pick, full breakdown, and proof record go live together. The schedule shows exactly how many fights are available.

  7. Freeze and grade

    The final pre-fight version enters the public Track Record. Cancelled fights are void, and completed fights are graded after the event.

What the analysis looks at

Blueprint organizes research into 7 evidence areas. The models assess their relevance to this matchup without fixed category weights. Missing statistics do not become zero performance, and unknown tactical mechanisms remain unknown.

See how the evidence becomes a forecast

Two OpenAI Luna advocates work in parallel, each building one fighter’s strongest case from the same evidence. Sol weighs both cases, counterevidence, and uncertainty, asking which fighter can actually establish their preferred fight. Blueprint checks the raw model probabilities without learned calibration, lens weights, or market adjustment.

  • Style
  • Physical traits
  • Film
  • Statistics
  • Career history
  • Fight context
  • Preparation

These are evidence areas, with no fixed category weights. Missing tactical detail stays unknown; verified career history can still inform the forecast.

What Blueprint refuses to chase

More information is not always better information. Hype, unsupported rumors, and noisy signals stay out when they cannot be checked well enough to improve the call.

Zero weightWhat Blueprint MMA refuses to weigh
  • staredown body language
  • unsupported motivation narratives
  • highlight reels as primary evidence
  • training clips as proof of competitive expression
  • unverified medical or psychological diagnosis
  • gym reputation without verified camp involvement
  • betting-market movement
  • unsourced referee and judging narratives
  • live/current-event scale weight
  • post-lock weigh-in reruns
  • unverified hydration speculation
  • raw intuition

Too unreliable, too situational, or too low-yield to move a serious forecast. No staredown ever changed who has the better chin.

When the picks appear

Blueprint can start the next confirmed card after the previous card is settled and reviewed, with a 72-hour backstop. A real lineup change can trigger a refreshed pick before the deadline. Odds movement alone cannot rewrite an official pick.

01Confirm card
02Review matchups
03Publish picks
04Grade results
  1. After the previous card01

    Confirm the card

    T−72 hours backstop

    Once the previous card is settled and reviewed, Blueprint can check the next confirmed event. Its start time, fighter identities, bout order, rounds, and main event must be complete.

  2. Card ready02

    Review each matchup

    Same process for every fight

    Blueprint saves the current card and sources. Two Luna advocates examine both fighters, then Sol adjudicates their cases. Missing evidence and source disagreements remain explicit.

  3. Before picks close03

    Check and publish

    Each ready fight

    Automated checks verify names, sources, and percentages. A fight publishes only when it passes; the official pick, free win chance, and available breakdown stay tied together.

  4. Picks close04

    Freeze and grade

    No quiet edits

    Official picks stop changing. Cancelled fights are marked void, completed fights are graded publicly, and Betting Desk activity stays on its own separate record.

Readiness is checked every five minutes, with a 72-hour backstop for the next event. Verified lineup changes can be processed automatically before picks close; unchanged picks retain their original versions. Betting odds alone never rewrite an official pick.

What you can hold Blueprint to

A prediction is useful only when you can tell what was called, when it was called, and what happened next. Blueprint follows these rules for every official card.

  • The free and paid views use the same official winner and win chance
  • Sportsbook odds do not decide or rewrite the official pick
  • Fighter and card checks catch substitutions and mismatched bouts
  • An incomplete or contradictory fight does not publish
  • Custom Matchup Reports never pad the official Track Record

One saved pre-fight version connects the free pick, full breakdown, public feed, and later result. Source checks do not guarantee that every explanation is correct; published corrections make identified errors visible.

Do not take our word for it

Check whether the call existed before the fight and whether it was right afterward.

  • Official predictions are published before fights
  • Every eligible win and miss remains in the public Track Record
  • Accuracy shows how often the winner was right
  • Win-chance tracking shows whether the percentages match real outcomes

This architecture has not yet demonstrated a measured improvement in predictive accuracy. Inspect the settled sample and its size before drawing conclusions or paying.

Start with proof, not a promise

See the pick free.Unlock the why only if you want it.

Browse the next card, check the public record, or build a Custom Matchup Report. Then choose only the access you need.

See free picks Compare accessBuild my first custom report free

Deeper Reads

Optional reference material for readers who want the technical definitions, extended method, and detailed performance data.

FAQ

What can I see before I pay?

For every published official fight, you can see Blueprint’s winner, win chance, publication time, and public result. Inspect those first and decide whether the deeper explanation is useful to you.

What does paid access add?

Paid access adds the likely method and round, main and alternate paths to victory, full reasoning and evidence, and Custom Matchup Report credits. It does not reveal a secret, different winner.

What factors shape the analysis?

Blueprint considers style, physical traits, available film evidence, statistics, career history, fight context, and preparation. Both advocates use the same research. The adjudicator weighs their opposing cases and which routes are supported; missing measurements never become zero ability.

What do the percentages mean?

The win chance is the adjudicator’s raw AI estimate, without sportsbook blending or learned calibration. Method chances are conditional on that fighter winning. A historically documented winning method retains at least 1% of that conditional distribution; this is a guardrail, not a measured accuracy boost or a minimum chance in every round.

Can the engine assess a UFC debut or limited tactical evidence?

Yes, when both fighters have enough verified career or regional fight history. The report can use that history while saying which tactical phases remain unknown. It must not invent UFC statistics or specific technical advantages. If a supported forecast is not possible, the report waits or gives no pick.

Do subscription tiers change prediction behavior?

No. Everyone sees the same official winner and win chance. Plans change how much explanation and tooling you can open, not which fighter Blueprint picked.

What counts in the Track Record?

One final official pre-fight pick per active scheduled fight counts. Custom Oracle reports for current, future, or hypothetical matchups never enter the official record.

How is performance verified?

Predictions are saved before the fights and compared with actual results afterward. Track Record shows wins, misses, accuracy, and whether the stated win chances have matched reality over time.

What happens if the engine cannot produce a valid read?

The new release waits and the existing public page stays unchanged. Blueprint does not fill a gap with a guess, publish contradictory sections, or quietly substitute an older answer.