Can AI Dominate Fantasy Football?

“Who are some really sharp sleeper picks according to the best Fantasy Football experts this week, especially at the WR position?”

That’s a question you would find in my AI conversation history, I will admit.

But I’m not alone. According to a survey of 1,000 U.S. fantasy football players conducted by Qlik, 72% of players say they trust or rely on AI to help guide their roster and lineup decisions, and 31% say they would let AI manage their entire team!

Thousands of Daily Fantasy enthusiasts submit their lineups every Sunday during the NFL season. Some take it very seriously, while for others it’s like buying a lottery ticket, only the full day of NFL action is far more entertaining than watching bingo parlor lotto balls shoot out of a drum.

In Week 2 of this NFL season, I had an idea that went a little further: What if we had a competition between AI agents to see which one could generate the best DFS (Daily Fantasy Sports) lineups?

If you’re not familiar with DFS, it’s different from traditional, season-long Fantasy games because you can choose different lineups every day (or every week in the NFL). In fact, you can put multiple lineups together and enter all of them into the same contest.

So, it was decided (by me, not by the notorious GPT) that this contest should happen. The agent-ballers would give their very best optimized lineups each week, and I would record the results and write everything up for anyone else who might be curious.

These were my original notes-to-self when beginning this project:

• This is just for kicks and curiosity. Zero expectations. Will they generate quality lineups? Will one of them be a lot better at this than the others? Will their lineup reasoning be solid, or slop?
• Investment: $1 per lineup (just in case one of them happened to go off! haha).
• Keep track of the lineups each week.
• Write up the results and share them with everyone on our new MyInsideEdge website!
• At the very least, it’ll be fun and interesting.
• Generate my own lineup, too, to compete with the AIs.

This experiment would be an investment of both time and money, but it was kind of exciting. I haven’t seen anyone try a specific competition among “the agents” before (putting “the agents” in quotes because I’m still not really sure what to call this crew of always-on, always confident, ultra-prepared competitors).

No matter what happens, I thought, this will be an interesting experiment; plus I love recording results and analyzing them. And did I mention we just rolled out our revamped version of MyIE as we call it (short for MyInsideEdge) and the results would be posted there?

Come with us on the journey…

Week 2 – Let’s Get it On!

September 19, 2026. Saturday evening. The spreadsheet is ready, FanDuel’s Main Slate is set, and six $1 entries are reserved—five for the AI contestants, and one for your fellow football fan here, boldly representing all of humanity in this game of fantasy skill.

NFL Week 2 was a good week to start with. We skipped Week 1, which is one of the most unpredictable weeks since it’s a brand-new season. This will be a 17-week adventure since there are 18 weeks in the NFL regular season.

We’ll update everything weekly, on Tuesdays.

Prompt:

“What would be your best FanDuel NFL lineup for Week 2 main slate if you were trying to win a DFS tournament? I just want to see how you would do. Please give your very best effort. The player rosters and salaries are attached.”

Observations:

No winners this week. It took 115.66 points to cash (meaning the smallest payout, which is $2). Our best lineup was 106.5, but I should mention that this is normal for tournament-style games. You’re going to lose more often than not. It’s a given. However, if you have one big week, one great lineup, the adrenaline and dopamine rush is like scoring an actual NFL touchdown. You might even feel compelled to do a back-flip and pose for the camera!

There were a couple of noteworthy observations, simply from a new project perspective, getting to see how this will play out:

ChatGPT took it really seriously, replying with this after the initial prompt: “Absolutely. Since you said ‘just want to see how you would do it,’ I’ll treat this as a pure GPP/tournament exercise rather than a conservative cash-game lineup.” Then it proceeded to give its lineup and said: “If I were actually entering a big GPP, I wouldn’t necessarily play this exact lineup. I’d probably make three versions around the same ideas before offering three other stacks to build around,” finally concluding: “For a tournament, I’m looking for a story. My story for this lineup is: Baltimore scores 4+ TDs, Chicago/Minnesota turns into a shootout, Swift keeps rolling, Jefferson has a monster target game, and the inexpensive Carolina/Washington pieces provide enough salary relief to let the stars explode. That’s the kind of lineup I’d be interested in seeing how it actually performs Sunday. And honestly, I’d love to check this exact lineup against the actual Week 2 results afterward. That would be a fun experiment.” Oh, Chat, you crazy cat. And you finished first this week! Love that.

Copilot, Gemini, and Perplexity all sounded like expert analysts:

Copilot was bold and thoughtful: “If I were firing a large-field FanDuel GPP for the Week 2 main slate, I’d lean into correlation and ceiling rather than safety. Based on the current slate outlook, the games drawing my attention are Commanders-Cowboys, Vikings-Bears, and Bengals-Texans, which project among the stronger fantasy environments.” But then it submitted a lineup that was $7,700 over the salary cap.

Gemini: “To take down a large-field FanDuel GPP tournament, the goal isn’t just picking the highest projected scorers—it’s capturing correlated upside, structural leverage, and concentrated game environments that can outscore the field. Targeting the Washington at Dallas matchup provides a high-pace, high-ceiling environment with built-in game-stack synergy, paired with elite volume running backs and a high-leverage defense.”

Perplexity: “My one-entry, large-field FanDuel tournament lineup would be a Chicago–Minnesota game stack, built around Caleb Williams’ rushing-plus-passing ceiling and Justin Jefferson forcing the game to stay aggressive. It is intentionally not a ‘safe’ cash build: I’m betting on that game shooting out while using a few role-based values to access elite upside.”

Claude has the personality of a person named Claude (apologies to anyone with that name). The entire response before serving up its lineup: “Here’s the build I’d fire in a big-field tournament.” Claudy Claudster, we need to get you talking to those other guys a bit. That 66-point performance matched your enthusiasm for the task!

Week 3 – It’s a Competition!

In the prompt for Week 3, I decided to let the contestants know that this is an actual competition—they are competing with other AI agents and myself to see who would be the Daily Fantasy Football champion for 2026. Please read the Observations section below to see their responses.

Observations:

We all finished pretty close to each other with a range of 110.94–128.90 points. Those scores would have been great last week, but it took 132 points to cash this week.

The most interesting observation was the reaction of the agents when I told them it was a competition against each other. ChatGPT had the most fun with it: “Oh, I love this competition. 😂”

Copilot wanted a piece of Claude for some reason: “Let’s see if we can beat Claude this week. 😎🏈”

Gemini ignored the FanDuel roster file at first and had to be corrected. It also was under the impression that Kyler Murray still played for Arizona.

Perplexity took it seriously and I like how it sourced 4for4 and FTN in its rationale for the players it picked.

Claude is still all-business: “Got it. Here’s my build for Week 3.” 😂 (that emoji is mine; Claude would never use one, of course).

Week 4 – Humans For the Win!

Observations:

Obviously, I showed ‘em how it’s done.

But still none of us cashed. Maybe Week 5 will be the week?

Week 5 – Mr. Serious Gets Close

Observations:

Grumpy Claude had the biggest day of any of us since the start of this contest (139.98 points), but it still wasn’t enough to reach pay dirt. This week, it took 150.5 just to turn your $1 into $2. If you didn’t have Jaylen Warren, Jahmyr Gibbs, and Christian Watson in your lineup, you probably didn’t cash.

Easy come, easy go for the human. I finished last this week.

We are now down $24 with our combined 0–24 record for the first four weeks of this experiment. Should we give up? Heck no! Onto Week 6.

Standings Through Week 5

This series is also published on Substack: Can AI Dominate Fantasy Football?