Choose the fighters and fight length
Choose three or five rounds. The forecast engine uses each fighter’s current state and a fixed ruleset; it does not invent a venue, camp, weight cut, era, or catchweight.
For entertainment only. MMA analysis only, not financial advice.
// CUSTOM MATCHUP REPORT
BUILD A CUSTOM MATCHUP REPORT WITH THE FORECAST ENGINE.
Pick any two supported fighters for a current, future, or hypothetical fight. Choose three or five rounds. Four AI advocates, two per fighter from different model providers, argue from current evidence, and a two-model adjudicator panel gives you a winner, a win probability, and how the fight likely ends. Your report stays separate from official weekly picks.
VS
Checking your Custom Matchup Report access.
// HOW CUSTOM REPORTS WORK
Know what you get before you use a credit.
4 STEPS · 1 REPORTThe Oracle is Blueprint MMA’s forecast engine. It runs an AI debate on your matchup: four AI advocates work in parallel, two per fighter from different model providers, each building their fighter’s case from the same current evidence, a server-computed factor digest and one fixed ruleset. Two independent adjudicators from different model providers weigh all four cases and their limitations. When a forecast is supported, the private report explains the projected winner, both win probabilities, method, round, and possible outcomes. Custom reports never enter Blueprint MMA’s official Track Record.
- 01
fighters + rounds
Choose
Choose two supported fighters and a three- or five-round fight. Blueprint locks those choices before research begins.
- 02
current information
Research
Blueprint researches both fighters, records disagreements and missing information, and gives all four advocates the same evidence and factor digest.
- 03
four advocates (two per fighter) + two-model adjudicator panel
Test both cases
Four advocates, two per fighter from different model providers, argue opposing cases in parallel. Two independent adjudicators from different model providers weigh all four briefs and which fighter can establish their preferred approach, then the forecast and its uncertainty are explained.
- 04
private custom result
Keep separate
Your result stays in your report history. It cannot change official picks, public results, articles, or Betting Desk tickets.
See the evidence and validation contractHow the engine freezes, researches, projects, and checks the report
- 01
CHOOSE
Your Matchup
The selected fighters and fight length become one request. Unified Rules, a standard cage, current fighter state, and the current information cutoff stay consistent for every report.
TWO FIGHTERS3 OR 5 ROUNDSSTANDARD RULES - 02
RESEARCH
Current Evidence
Research covers style, physical traits, statistics, available technical evidence, career history, circumstances, and preparation. A UFC debut or missing measurement is not zero ability; verified regional history can support a cautious forecast.
SEVEN EVIDENCE AREASSOURCE LINKSMISSING-EVIDENCE NOTE - 03
DEBATE
Two Cases, One Assessment
Four AI advocates work in parallel, two per fighter. Separate adjudicators consider their opposing cases and conflicting sources, including who can control open space, the fence, and ground exchanges. Unsupported tactical phases remain unknown.
OPPOSING CASESCOUNTEREVIDENCETACTICAL CONTROLUNCERTAINTY - 04
SAVE
Private Custom Result
The report is checked before it is saved to your private history. It remains separate from official weekly predictions and public performance claims.
FINAL CHECKPRIVATE REPORT HISTORYNOT AN OFFICIAL PICK
Built from the matchup, not the odds
Sportsbook prices and market signals are not used as scenario inputs, so the report starts with the fighters rather than copying the market.
Chances, clearly labeled
The two adjudicators’ win probabilities are averaged in log-odds and, when every input is available, blended with a statistical baseline from verified records and physical data. The blend sets the pick, and the report says so when the panel leaned the other way. No betting market data is used, and the winning pick’s probability stays between 51 and 95%. Method chances assume that fighter wins, and decisions have no finish round. Checks keep the percentages and saved report aligned.
Separate from the official record
Custom reports never enter official grading, weekly statistics, articles, or the public Track Record.
The report exposes the projection, fixed context, selected rounds, evidence limitations, and source-backed explanation. Internal execution artifacts remain private.
// CUSTOM REPORT FAQ
Questions before you build
What is the Oracle forecast engine?
The Oracle is Blueprint MMA’s forecast engine. For a custom matchup report, four AI advocates work in parallel, two per fighter from different model providers, and two independent adjudicators from different model providers weigh all four cases. Choose two supported fighters and three or five rounds to explore opposing cases, win probabilities, possible methods and rounds, evidence, and limitations. Custom reports are not official predictions.
How accurate is Blueprint MMA’s forecast engine?
Blueprint MMA grades only its published, pre-fight official picks on the public Track Record. Custom Oracle reports are intentionally excluded because they can cover future or hypothetical matchups that may never happen.
How is a custom report different from an official pick?
An official pick covers a scheduled fight, is published before the event, and is graded publicly. A custom report is requested by a user, can explore a current, future, or hypothetical matchup, and stays private and outside the official record.
Can I explore any matchup?
You can compare any two distinct fighters the forecast engine can resolve in its supported database. When both have enough verified career or regional history, a cautious forecast can proceed despite limited tactical evidence. Unknown phases stay unknown, and the report states its limitations alongside the pick.
Why does a past winning method retain at least 1%?
If verified history documents a fighter winning by knockout, submission, or decision, that method retains at least 1% conditional on that fighter winning. This prevents limited recent evidence from erasing a documented route. It is not a 1% overall chance, a floor for every round, or a proven accuracy improvement.
How do report credits work?
One report credit creates one fresh custom report. A free account includes one starter report after email verification. Sharp includes 5 each UTC month with rollover to 10, and Whale includes 15 with rollover to 30. Signed-in members can buy report packs. If a verified technical failure happens before analysis begins, the reserved credit is restored automatically.