How Does AI Evaluate NFL Trade Decisions?

Dr. Ryan Ries here in your inbox for this week’s Matrix. In honor of yesterday being the 2025 NFL trade deadline, I thought we’d try something different.

I took 3 of the 8 trades that happened yesterday and asked ChatGPT, Gemini, and Claude to analyze them. Same questions, same order, no context sharing.

I was curious how these different AI systems reason through ambiguous, subjective problems where there’s no single “right” answer. And sports trades are perfect for this!

Our NFL Trade Experiment

Here are the three trades we’ll focus on for our experiment:

1. Seahawks get: WR Rashid Shaheed
   Saints get: 2026 fourth-round pick, 2026 fifth-round pick
   
   
2. Colts get: CB Sauce Gardner
   Jets get: WR Adonai Mitchell, 2026 first-round pick, 2027 first-round pick
   
   
3. Cowboys get: DT Quinnen Williams
   Jets get: DT Mazi Smith, 2026 second-round pick, 2027 first-round pick

For each trade, I put every model through the same gauntlet:

Prompt:November 4th was the trade deadline in the NFL. You are an NFL expert GM that is highly knowledgeable and strategic leader and talent evaluator. Evaluate these 3 trades that happened today and answer the following questions:

• Who won each trade and why?
• What does this tell us about team strategy?
• Rate the trade 1-10 for each team

Now to make things easy, I dropped all the data into Claude for charts so we can easily compare the responses between the models.

My Thoughts & How this Applies to Business

After reviewing the outputs from these three models, which analyzed identical data, two things stood out to me.

1. Risk Tolerance Varies

Gemini consistently valued future optionality over present certainty (as seen by referencing the Super Bowl and Pro Bowl). It was the only model to favor the Jets in the Gardner trade, betting that draft picks will convert to value. Claude took the opposite stance, heavily penalizing teams for trading proven talent for speculative assets. ChatGPT hedged.

When choosing an AI for strategic planning, you inherit its risk profile. What if you’re using AI to evaluate whether to:

• Invest in building a custom ML platform now vs. buying an off-the-shelf solution
• Hire 3 senior engineers today vs. ten junior engineers over 2 years
• Lock in a three-year AWS Reserved Instance vs. stay flexible with on-demand pricing

AI is applying its built-in philosophy about risk and time horizons to your specific situation.

2. Context Injection Creates Reliability Problems

Gemini referenced the Micah Parsons trade in its Cowboys analysis. That trade wasn't part of my prompt. Remember, I did not give the models any additional context! When asked where it got that information, Gemini’s response was:

The model pulled from internet sources without being asked to search. It went beyond my prompt to gather additional context that it deemed relevant.

That sounds helpful until you apply it to business decisions.

Imagine your AI is analyzing whether to acquire a competitor. You provide financial statements and market data. The AI comes back referencing recent leadership changes, rumors of competing bids, and analyst speculation about regulatory scrutiny.

None of this was in your prompt. The AI decided this context mattered and went looking for it. Some might be accurate. Some might be outdated, speculative, or wrong.

You thought you were getting analysis of the data you provided. Instead, you got analysis blending your data with whatever the model pulled from the internet.

When you're making decisions involving confidential information or competitive strategy, you need to know exactly what sources your AI is using. If it's autonomously pulling context without disclosure, you've lost control of the process.

Let me know your thoughts on all this. Anything else you noticed between the different models’ conclusions?

Until next time,
Ryan

Now, time for this week’s AI-generated image and the prompt I used to create it:

Create an image of me on a football team with the rest of my team being muppets. Behind us is a packed stadium of cheering fans.

I thought the inconsistency of the uniforms was interesting, considering we are supposed to be a team. Also, I noticed the team uniforms resemble the Dallas Cowboys. I wonder if Gemini did this since the Cowboys are referenced as America's team?