Finding Edge in Hot Dog Eating Markets

Sign up HERE!

Last week we covered my foray into betting into an unorthodox market (the NBA draft) with the help of AI.

This week I’m at it again (!!), taking the absurd July 4th tradition of Nathan’s Hot Dog Eating Contest and working a betting angle to make those 10 minutes more fun and profitable!

Author’s note: I realize what follows will stretch the definition of “sports betting” and reads like satire. It is not. Still, it’s probably not what my Mom pictured when I graduated law school.

This week I’ll cover my framework for modeling a hot-dog-consumption outcome, how I again leaned on AI to supplement my work, and the betting angle I hope will pay off on Saturday (it’s already got big time closing line value).

As always, we’ll close with our bets of the week, this time on hot dog props that show major value.

Let’s get after it.

My Framework to Handicap Eating

I’ve watched the Hot Dog Eating Contest before, and I’ve probably even bet on it, but I’d never set out to understand what variables contribute to the outcome.

Here are what I learned to be the important things to consider (ranked most influential to least):

1. Participants: Competitive eating’s “modern era” (2001 to today) has been dominated by two men and one woman: Takeru Kobayashi, Joey Chestnut and Miki Sudo. Most relevant for the markets I’m looking at currently: 10 time champion Chestnut (pictured below) will be back this year. He missed last year due to a conflict of interest, and the winning total dropped to a 14-year low of 58 in his absence.

2. Weather: In simplest terms, cool or climate controlled weather is ideal for high scores. Any deviations from this environment seem to bring down consumption. That includes extreme heat, humidity, wind or rain (which can cause delays). The importance of weather is backed up by comparing the outdoor record (76 by Chestnut at Nathan’s contest in 2021) against the climate controlled record (83 by Chestnut on an indoor soundstage on the Netflix eating special).

3. Unusual events: Protests, contestant injuries, and other black swan events have led to lower totals in the past. This category doesn’t deserve the weighting of the other two, but I’ll bring estimates down a bit to account for any such tail variables.

In the next section I’ll walk through how I used AI tools to gather data on those variables and turn them into probability scores for different betting markets.

Gathering Data and Making Predictions

To turn the three key variables into probabilities I again leaned on ChatGPT to collect and merge public data. Here is the approach:

Step 1 - Pull the raw numbers automatically

• Historical results – ChatGPT queried the contest archives and built a table of every total from 2016‑2024, focusing on this year’s participants.

• Weather archive – In the same run, the model pulled July 4th Coney Island temperature and humidity from National Weather Service (NWS) records.

• Forecast snapshot – Using the NWS API the model grabbed the latest July 4th 2025 outlook: 77°F - 82°F, 50% relative humidity, light wind, no rain.

• Current lines – I supplied prices on the Kalshi markets I was interested in. These were a rolling set of Over/Under lines for the winning total: 69.5, 74.5, 79.5, 84.5.

Step 2 - Fit a simple distribution

ChatGPT fit a normal curve to likely-winner Joey Chestnut’s last eight healthy 10‑minute performances.

The baseline mean (μ) came out 69 Hot Dogs & Buns (HDB) with a standard deviation (σ) of 4.5.

The model then adjusted μ for forecast weather:

With the current weather outlook the working mean (or over/under line) becomes 70.5 HDB.

ChatGPT then produced probabilities for each Kalshi threshold:

Step 3 - Convert probability to fair price

Because Kalshi prices contracts in cents, probability can be directly converted into “fair price.”

For example, a 44 % chance of the winning total being less than 69.5 would imply a fair “NO” price of 44¢. 

Having built this value model, I could now assess if any of these markets offered bets with a meaningful edge.

Fortunately, they very much did!

Bets I Made

When I placed my bets last weekend, here’s how each total looked in comparison to my model.

I was able to place sizeable bets at every price mentioned above, and my modeling has been validated by the market, as each bet has already accrued meaningful growth in value.

Take a look!

Nathan’s Hot Dog Eating Contest is happening tomorrow at 12:30 Eastern, so I’ll make some recommendations below as to how you can snap up whatever edge remains.

Bets of the Week $$

Last week we went 0-1 and lost 0.5 units.

Salzburg was not able to keep up with Real Madrid and our long-shot (+550) "draw” bet flamed out when they lost 3-1.

Even so, since starting the newsletter, bets given out in this section are ahead 14.9 units, at a positive 18% ROI. We’ll update this regularly.

Based on the research detailed above, I am recommending the following bets this week:

• Hot Dogs Eaten by Nathan’s Contest Winner UNDER 69.5 @+200 on Kalshi for 0.5 units

• Hot Dogs Eaten by Nathan’s Contest Winner UNDER 74.5 @-200 on Kalshi for 1 unit

• Hot Dogs Eaten by Nathan’s Contest Winner UNDER 79.5 @-400 on Kalshi for 0.5 units

My rationale for these bets was covered in depth above, but the key thing to monitor tomorrow is the weather!

If the temperature and/or humidity spike, and pricing stays the same, up your unit size by 50%.

If it’s an epically beautiful day, I’ll forgo the top and bottom rungs of the ladder and just bet “Under 74.5” for 0.5 units (50% reduced).

Here’s to hoping for some NYC heat!

Please fill out the survey about today’s newsletter and let me know your thoughts!