Pick more UFC winners.
When the brain is most sure, the pick wins 76.6% of the time. Proven in public, locked before every fight.

Stop betting on hype, gut feel, and YouTube confidence. Cagemath runs on the brain: a prediction engine built from a million rated fights by a UFC veteran who has been writing code for 30 years. Tested blind on 897 fights it had never seen. Every pick is timestamped into a public, permanent record book that nobody can edit, not even us. You don't have to trust the story. Check the receipts.

Here is the next upcoming fight card. The picks are already locked:

LOCKED
UFC - Road to UFC Season 5: Shanghai Semifinals
FREE PICK Rabindra Dhant over Kasib Murdoch 57.0%
See the board: 2 cards, 17 picks

New to betting or experienced? You're covered.

One thing the betting world never tells beginners: winning is not just about picking winners. The sportsbooks already know who the favorites are, and they set their prices accordingly. The real opportunity is finding fights where the odds underestimate a fighter's true chances. That is exactly what the brain was built to find, and you don't need to know any betting math to use it. Every pick simply shows:

  • Who the brain says wins, and how confident it is (a plain percentage)
  • What the sportsbooks think, right next to it
  • A VALUE flag when the sportsbooks are underrating our fighter
  • A suggested bet size, in your own dollars. Type your betting budget once, say $500, and every pick shows a real number: put $15 on this one, $9 on that one. No math on your end, ever.

You can read any pick in about five seconds. And if you're an experienced bettor: every advanced number you'd want (implied probabilities, line movement, best available odds at every book) is still there. Nothing was dumbed down; it just explains itself now.

Guessing is expensive.

Every weekend, sportsbooks profit because most fans bet on feelings: a highlight reel, a podcast take, a pick-seller on social media who deletes his losing weekends and reposts the wins. The betting apps employ teams of mathematicians. Most bettors bring vibes. That gap is where losing lives, and no amount of fight knowledge closes it without numbers.

The fix: take the emotion out and put the receipts in.

Cagemath's engine, we call it the brain, evaluates every fight the same way regardless of hype: who actually wins these matchups, at what rate, based on everything measurable about both fighters. Then it does the thing no tout will do: it locks every pick into a public ledger before the fight and grades itself after. Here is what makes that work:

01 / THE MODEL

Built on every fight, not vibes

Built from how every fighter connects to every other fighter across a million fights of MMA history, plus round-by-round striking, takedown, and control data from every UFC fight ever. No narratives, no favorites played. See how the brain works.

02 / THE NUMBER

Percentages that mean something

When the brain says 75%, picks like it have historically won about 77% of the time. We prove that on fights the brain had never seen, and we publish the full table below so you can check the honesty of every percentage yourself.

03 / THE EDGE

We show you the market too

Each pick shows what the sportsbooks think next to what the brain thinks. And because the same bet pays differently at different betting apps, every pick also names the app currently paying the most for it, so you never take a worse deal than you had to. When the brain and the sportsbooks disagree hard, the pick gets flagged. Those flags are where the money gets made; the full story is just below.

04 / THE LEDGER

Graded in public, forever

Wins and losses both go on the permanent record automatically when results come in. No deleted tweets, no "we went 9-1" with a straight face. Check us any time.

Where the money gets made: the disagreement.

The simple version. Every sportsbook's odds are really an opinion about each fighter's chance of winning. The brain forms its own opinion, from the data. Most weeks the two roughly agree, and those picks are simply solid. But some weeks the brain sees something the sportsbooks have not priced in yet, and the two disagree hard. Those picks get a VALUE flag on the board. That flag is the sweet spot of this whole service.

A real one, from a card everybody watched. July 11, UFC 329: the sportsbooks had Mario Bautista as a coin flip against Cory Sandhagen, 50/50 (even money). The brain had Bautista winning 77% of the time and flagged the gap. Bautista won. At even money, a $100 bet returns $100 in profit, so that one pick paid +$100. Here is the receipt, timestamped before the fight.

And the whole truth, because we sell receipts, not fantasies: no pick is guaranteed and some flagged picks lose. So here is the number with every loss counted.

Imagine you had placed the same simple $100 bet on every single pick we have ever graded. Winners and losers both. Nothing skipped.

+$675
where you would be sitting right now
7.9%
return on every dollar staked, losses included

For every $100 wagered, about $108 came back. To put that 7.9% in perspective: professional sports bettors grind entire seasons to clear low single digits, and they call that a career. That number lives in public, updates every card, and you can replay the whole history yourself at the odds we recorded. Analytics, not betting instructions; past picks never promise future ones.

It's Saturday night and you're watching the fights with your buddies. They're grumbling about the bets they just lost, and you called the upset before the walkouts. They're going to ask how you knew. Whether you tell them about the brain is up to you.

The proof, in numbers we can't take back.

1M+
fights rated across all of MMA history
897
blind-test fights, no hindsight, no cherry-picking
65.1%
winners called on fights it had never seen
76.6%
hit rate when we say 70-79%

The public ledger is young: 62-25 (+$675 at flat $100) since launch. The math behind it is not: a million rated fights, 897 blind-test bouts, receipts on everything. Watch the ledger grow.

Do the percentages mean what they say? Simple test: take every pick where the brain stated a confidence level, and check how often those picks actually won. Here is that test across 897 fights the brain predicted blind. This is the table every picks service should publish and none do:

When the model saysThe pick actually wins
50-59%56.9%
60-69%68.8%
70-79%76.6%

Read it like this: take every fight where the brain said its pick had a 60 to 69% chance. Those picks went on to win 68.8% of the time. Same story in every row. Here is why that matters to you: it means the percentage on a pick is a number you can actually plan around, not hype. A service that exaggerates would show a right column much lower than the left one, which is exactly why no other picks service will publish this table.

Bring your skepticism. We built for it.

Three ways to check us before a dollar moves, no account required:

A / READ IT

The public ledger

Every pick ever locked, timestamped, with the market price at lock and the result stamped after. Open the record.

B / REPLAY IT

The simulator

Pick a bankroll and a strategy and replay our whole graded history at the prices we locked. Losses included. Run the receipts.

C / PULL IT

The raw data

Our complete history as raw, computer-readable data, so anyone can audit the math themselves. Pull it here. Plus free calculators for converting odds, seeing true prices with the bookmaker's cut removed, and sizing bets.

I spent 26 years learning fights from the inside. Then I built the brain.

My unfair advantage is that I've had every job in this sport.

I was writing code before I could drive. At 19 I walked into an MMA gym, and by 23 I had a UFC contract fighting at middleweight. While I was still fighting I opened my first gym and started building fighters instead of just being one: I've trained UFC athletes since 2006, hold a 4th degree black belt in Brazilian jiu-jitsu, and own American Combat Gym.

But the years that taught me the most weren't in the cage. They were as an agent. I represented some of the top talent in the UFC, and my edge was always the same: I could watch a kid on a regional card and know he was real before anyone else did. Then I'd book him the exact right fights, in the exact right order, until the UFC had no choice but to call. That's what matchmaking really is. It's prediction, with a fighter's career as the stake. I got very good at being right.

The whole time, I never stopped writing code, and five years ago I went all in on AI. Then the two lives collided. I built Cagefile, a tool that helps gyms and managers find the right opponent, the way I used to do it by instinct. To make it work, I had to turn everything I knew into math: how records lie, how strength of schedule compounds, why a champion's decline shows up in the numbers before it shows up on his record. Somewhere in that build I realized what I was actually holding. The engine that finds winnable fights, run in reverse, picks winners.

That accident is Cagemath.

Twenty-six years of fight IQ, encoded, and graded in public every single week.

Charles McCarthy
Charles McCarthy
UFC veteran, middleweight Training UFC fighters since 2006 4th degree BJJ black belt Former agent to top UFC talent Owner, American Combat Gym AI engineer, 30+ years writing code
Most pick sellers show you their wins. I show you everything. Every pick locks before the card, grades after it, and stays on the ledger forever, wins and losses both. That's how it worked when my fighters' careers were on the line, and that's how it works here.

The fight world has looked under the hood.

I've spent years inside the Octagon, and I know how unpredictable this sport looks from the inside. That's what gets me about CageMath: it keeps finding value the betting lines miss, and the number is right more often than it has any business being. It's like having a world-class analyst in your corner before you place a bet.
Charles RosaCharles RosaUFC veteran
As a fighter, I obsess over tape and stats to find the winning edge. CageMath does that at a scale no human can, and the results are right there in public for anyone to check. The picks hit. If you want to stop guessing on fight night, this is the only platform you need.
Lucas RosaLucas RosaUndefeated MMA prospect
Training fighters means breaking down every advantage and vulnerability in a matchup. CageMath does exactly that, then shows you where the betting line got it wrong. I've watched it call underdogs the whole industry wrote off. As a coach I respect the science behind it; as a fan I just like being right more often.
Matthew WagyMatthew WagyMMA trainer and fighter

Quotes from fight people are worth more than quotes from strangers, but we hold ourselves to a harder standard than either: every claim on this page is checkable against the timestamped public ledger, and members who outpick the brain earn their spot on the public Beat the Model leaderboard, graded by the same rules as the machine. Receipts first, reviews second.

The picks are half of it. Here is the rest of what you get as a Cagemath member.

A membership isn't a weekly list of names. It's the whole fight-week experience, built around the brain:

EVERY CARD

The picks, fully loaded

Every UFC fight with the brain's win probability, the sportsbooks' number beside it, VALUE flags on the big disagreements, the best available odds at every book, and bet sizing converted to your own dollars.

FIGHT WEEK

The pre-fight briefing

Before every card, a briefing video from Charles himself: late intel, what changed since the picks locked, and where his fighter's eye agrees or argues with the brain. No other prediction service puts its founder on camera every week.

GAME NIGHT

Fight-night live chat

A live room for every card, open through fight night. Talk the card as it happens with people watching the same numbers you are.

COMPETE

Beat the Model

Call every fight yourself before the card locks and get graded in public, right next to the brain. The best record on each card wins a free month of membership. No purchase or bet ever required to play.

COMMUNITY

The forum and DMs

A members' forum with a thread for every card, polls, and direct messages. Fight talk with people who look at numbers, not just highlights.

EVERY WEDNESDAY

The weekly newsletter

Last week's graded results, what's on the board next, the forum's best threads, and member offers, in your inbox once a week.

YOUR MONEY

Bet tracker and tools

Log what you actually bet and the site settles it automatically when cards grade, so the math, not your memory, tells you how you're doing. Plus the full calculator toolbox.

See the full membership tour.

What is all of this worth? Look at what the alternative charges.

First, meet the competition. They're called pick sellers, or cappers: people selling their fight bets through Instagram pages, Telegram groups, and websites with no verifiable track record at all, the ones who delete their losing weekends and screenshot their winning ones. That industry routinely charges $100 to $300 a month for a list of names and a "trust me."

Now stack that against what a Cagemath membership actually includes: every pick with a calibrated probability and the reasons behind it, the sportsbooks' own numbers alongside for comparison, VALUE flags on the mispriced fights, bet sizing in your own dollars, Charles's pre-fight briefing video every fight week, the fight-night live rooms, the forum, the Beat the Model contest with free membership months for winners, the weekly newsletter, the bet tracker, and the one thing no pick seller on earth will show you: a public, permanent, timestamped record of every pick we have ever made, losses included.

By the pick-seller industry's own price sheet, that package is easily worth the top of their range, the full $300 a month. We priced it where almost anyone serious about fights can comfortably try it.

$49/month
or $39/mo billed annually
  • Every pick for every UFC card, locked before fight night
  • Win probability, sportsbook comparison, VALUE flags, sizing
  • Charles's pre-fight briefing video every fight week
  • Fight-night live chat, forum, and DMs
  • Beat the Model contest with a free month for each Card Champ
  • Weekly newsletter, pick alerts, bet tracker, all the tools
  • The full graded record, forever
  • Cancel anytime in two clicks
Get this week's picks

$49 is introductory pricing, and it will not last: the price climbs as the public record grows, and it could double before long. But join now and you are locked in. The rate you join at is your rate for as long as you keep your membership, monthly or annual. Late joiners pay the new price; you never will.

Questions people ask before joining

Can picks be changed after the fights?

No. Every pick is locked into the database with a timestamp before the fight, and the result is stamped next to it after. The ledger is public, permanent, and available as raw data so anyone can verify nothing was edited. This is the entire reason Cagemath exists.

Is the win rate really 76.6%?

On high-confidence calls, yes, and here is the honest, full version: across 897 fights the model predicted blind, it called 65.1% of all winners overall, and when it stated 70-79% confidence, those picks won 76.6% of the time. We publish the whole calibration table above, including the lower buckets, because a cherry-picked number you can't verify is exactly the kind of thing we built this site against.

Do you guarantee winners?

No, and nobody honestly can. A pick at 70% loses three times out of ten, and losing nights are part of the math. What we offer is a process: calibrated probabilities, market comparison, disciplined sizing, and a public record of how it performs. Cagemath is statistical analysis for entertainment and information, not gambling advice.

Why is the live record so short?

Because we launched in July 2026 and we refuse to backfill. The 897 blind-test fights were predicted by the machine before it ever saw their results, but the public ledger only counts picks locked live on this site, in front of everyone. It grows every card. Watch it happen.

Do I have to bet to get value from this?

No. Betting is never required. Plenty of members are here for the fight analysis, the forum, fight-night live chat, and the Beat the Model contest, which is free to play and pays out bragging rights and free membership time, never money. If you do bet, you must be 21+ and bet legally. Problem gambling help: 1-800-GAMBLER.

What does it cost, and can I cancel?

$49/month, or $39/month billed annually, and the rate you join at is locked in for as long as you keep your membership, even after the price rises for new members. Cancel anytime in two clicks from your billing page, no phone calls, no retention scripts.

You found this early. That's the best seat.

Right now the ledger is young, the rooms are tight, and the rate is the lowest it will ever be. Every card that grades makes the record longer, the case stronger, and the price of joining higher for the people who show up after you. Picture your fight week as a member: the picks land Friday with the reasons behind every number, my briefing walks you through the card, and Saturday night you're in the live room with us, your own calls locked on the board next to the brain's. That's the club. I built it so I'd want to be in it.

See you in the room.
Charles McCarthy, UFC veteran and builder of the brain

Get this week's picks