AI isn't going to give you the answers in sports betting. But it can absolutely turbocharge your process.
Back to VideosOne of our viewers built a college basketball dashboard he said was worth 200 units over the course of just one season.
What was it? How did he do it?
Sports betting has an attention problem. The market moves all day, news breaks all night, and you can't stay on top of it all unless you're willing to stare at screens until your eyes melt.
But our viewer—a fellow by the name of Trap God 9—knew that the best way to use AI in sports betting isn't as an oracle. It's as an intern.
It's not hard to find AI skeptics when it comes to sports betting. But that's because people look to use AI the wrong way.
The people who say AI is terrible at finding bets for you are completely right. It's not a shortcut to finding good bets, but it's fantastic for handling all the down-and-dirty work that used to take bettors most of their week.
Today, we're going to look at five ideas for handing grunt work to an AI intern who never gets disgruntled. Hopefully you'll come away inspired with your own ideas on using AI to attack markets.
Because if you aren't, your competition—including some of the people right in these comments below—well, they are.
Before we're done, we'll put those pieces together into the kind of system one viewer used to get live betting alerts at scale.
First, though, let's look at taking advantage of one of the best things about AI:
It doesn't sleep.
The problem with interns is that you have to do things like feed them, let them go home from work, and stop bothering them in the middle of the night.
Jack, the whole country's gone soft.
But AI? Not so much.
It's always there. At 7 a.m., at 4 in the afternoon, at two in the morning after you stumble home from a few cocktails—which is convenient because Shams and Ian Rapoport never sleep either.
Market-moving information could come at any time, but you can put AI to work watching injuries, lineups, and news even when you can't.
In baseball, where a batter hits in the lineup dictates how many at-bats they'll have and the prospect of hitting with men in scoring position. You can pull confirmed lineups from an MLB API and have AI recalculate a lineup's stats on the fly or tell you when certain batters are hitting higher or lower in the order.
In basketball, minutes played are everything.
There are various sources for lineup information as it drops, whether that be Twitter, fantasy sports websites, or even our own news channels in the Unabated Discord.
Ask your agent to pull a website or social media account every couple of minutes looking for actionable information.
Want to know the effect of removing a player from a lineup? You can do that.
Want to be alerted to a batting order change? You can do that.
Compressing information into one easy-to-read dashboard is a great way to put together a system that could be your version of a 200-unit hit.
Using AI to be your eyes and ears on social media can leave you time for keeping up with other areas of social media, like engaging in the conversation on Gambling Twitter.
Then again... maybe you can find a better use of your time.
You can even have AI retrieve and process league-mandated injury reports. Tell the intern to summarize what they find and deliver it in an email digest.
It's the closest thing you'll be able to get to yelling at a nervous 19-year-old:
"Write up a brief summary and have it on my desk by the end of the day."
This is your eyes-and-ears intern.
Not the one making decisions—just the one standing by the door yelling,
"Boss... something changed."
Now, if you want your intern to do all the stuff that you hate doing yourself, you can do that, too.
The worst part of sports betting isn't going on some savage downswing.
It's letting your bet tracker pile up with a month's worth of unrecorded bets.
I'd rather wake up on the first Sunday of the NFL season and hear my wife tell me she just got us tickets to a matinee performance of Wicked than sit down and spend hours backfilling my bet sheet.
But you don't ever have to do that again.
The second project on my list is designed to get you spending less time bookkeeping and more time beating books.
You could build an entire bet tracker that pulls score information from ESPN's API and auto-grades all of your bets.
Or you could just build a module to calculate your Closing Line Value.
Enter the bet at the price you got, enter the closing price, and tell it to use the CLV formula of your choice.
Take it a step further and you could use an odds API—like the one we offer at Unabated—to pull in closing lines.
You could designate certain sharp books as your source of truth and then use the de-vig formula of your choice to compare a fair closing price to the number you bet it at.
This is your feedback-loop intern.
He's not just collecting information. He's telling you whether the processes you're putting in place have long-term value.
All that time you spent doing basic calculations?
Hand it off to the AI intern.
Have them DoorDash your coffee while they're at it.
So now you've got two processes you're handing off to the intern. That's two steps closer to building a complete project the way Trap God 9 did.
First, let's look at what else you might want the intern to handle—the real roll-up-your-sleeves work that nobody wants to do.
It used to be that you had to stare glassy-eyed at endless spreadsheets like you were trying out for the Excel Olympics.
I've been finding ways to beat sports betting for a long time, and when I look back at it, the hardest part was often cleaning data.
Some data might have team names. Others might just have city abbreviations. And player names? Don't get me started—they're a disaster.
But processing and analyzing large datasets is where LLMs excel.
No pun intended.
...Well, maybe a little bit.
You don't even need Claude or Codex to ingest a full spreadsheet or database. Often you can just paste in screenshots and have it read and process the information from there.
Try it yourself.
Gather data from a couple different sources and blend it all together. Maybe use NFL Next Gen Stats from NFL.com along with player data from Pro Football Reference.
Blending data from multiple sources can help you find a sharper signal and build a bigger edge.
This is the intern locked in the file room making sure Jaren Jackson Jr., Jay Jackson, and whatever crime a CSV committed against somebody's name all point to the same human being.
You never have to create another pivot table again in your life.
And that, in my opinion, is worth Anthropic and OpenAI using up all the water in California.
So there's three pieces of the process that you can use as the bedrock of your system.
It's still not quite enough to build the full thing, though.
We've already talked about gathering information from sports-focused sources.
But what about incorporating and automating non-sports sources that still affect your bets?
Nobody wanted to click through every city in their weather app to get conditions for games played outdoors.
For years, our best option was a cottage industry of websites that listed game-time forecasts and other weather information.
Now you can put your own personal AI reporter to work wherever you need them.
The National Weather Service has a free public API you can connect to, and gathering weather information is intern job number four on our list.
The NWS API doesn't just get you weather in a city, either.
You can drill down and pull data from specific weather stations within a city and use AI to figure out which stations are closest to the stadium.
Just make sure you understand how weather impacts the lines in whatever sport you're betting.
Markets have become much more efficient over the past decade in most sports.
It used to be you could still profit from popular misconceptions.
For example, people used to think that when it was muggy outside and the air felt thick to breathe, a baseball wouldn't travel as far.
Humid air feels thick to breathe.
But for a baseball, it's actually thinner.
Water vapor is lighter than the nitrogen and oxygen molecules it displaces.
Humid air is less dense and creates less drag.
Your lungs hate it.
The baseball does not.
Come for the sports betting angles. Stay for the science lesson.
Just call me Dollar Bill Nye.
This is the intern you send outside with a clipboard, a wind gauge, and just enough dignity left to still come back with useful information.
I let Codex get creative in the dashboard mockup it made for this one, but taking advantage of weather effects is something a lot of people are now factoring into their betting.
It's one piece of the puzzle, but it's not the whole picture.
There's more data that goes into every play.
And more data means more work.
And more work means more headaches.
The first four ideas are useful on their own, but together they point toward the bigger idea.
AI is at its best when you give it a narrow job and let it watch that job relentlessly.
And that brings us back to Trap God 9, because his dashboard is the full attention machine we've been working toward building this whole time
The first four ideas are useful on their own, but together they point toward the bigger idea.
AI is at its best when you give it a narrow job and let it watch that job relentlessly.
And that brings us back to Trap God 9, because his dashboard is the full attention machine we've been working toward building this whole time.
A while back, we published a video about how you could vibe-code an NHL shots-on-goal model. That's when we saw this comment.
He told us about a dashboard he'd created that looks for specific triggers in live betting.
He built one for college basketball that alerts him when both teams are in the bonus during the first half.
When he sees that, he looks to bet first-half overs on the live lines.
He said it resulted in something like 200 units of profit.
Pretty brilliant, in my opinion.
This is an example of a logic puzzle that you get paid to solve.
This is the attention problem solved in its cleanest form.
Not AI telling you who to bet.
AI watching something you physically cannot watch all of.
Here's why.
When a team gets into foul trouble, a couple of different things happen.
Obviously, players shoot more free throws.
But the secondary effect is that defenders also try not to accumulate additional fouls, so they play softer defense.
That leads to more scoring.
Now, does the market already react to this?
Possibly.
But your goal is to identify these situations as they occur and get a bet down quickly.
My conversation with ChatGPT to plan all this out was actually pretty straightforward.
It suggested different ways I could ingest the play-by-play data and different ways I might receive alerts.
It was very comprehensive.
After sufficient planning, you can use Codex or Claude to make your app come alive.
A couple of the ideas I tossed out in today's video involved creating dashboards. Typically, that means letting AI code something up.
Data visualization is one of the best uses for AI.
I'm a fan of creating dashboards wherever I can because they help me see the big picture.
Be aware, though, Trap God 9 did warn us that this required a lot of time staring at the screen.
I can only imagine what that's like on a January or February Saturday when there are more than a hundred college basketball games on the schedule.
That's exactly why a dashboard like this works as an AI project.
The AI isn't making the pick.
It's paying attention to a specific condition at a scale no normal bettor can keep up with, then handing that decision back to the human.
The best intern isn't the one you ask to be a genius.
It's the one you give one painfully specific task and tell to keep watching until something happens.
Using AI as an intern is something the sharpest bettors are already doing.
And if you're curious about some of the secrets behind their success and how they stay one step ahead of everyone else, check out our next video.
