Pack Your Knives

Pack Your Knives

Share this post

Pack Your Knives
Pack Your Knives
Can we predict tonight's winner from past seasons’ final three chefs?

Can we predict tonight's winner from past seasons’ final three chefs?

Carly Levitz's avatar
Carly Levitz
Jun 12, 2025
∙ Paid
3

Share this post

Pack Your Knives
Pack Your Knives
Can we predict tonight's winner from past seasons’ final three chefs?
Share

Last week, Tom asked me to look into if there’s a way to predict a winner at the final three. There are many things that I think could be related to the winner: their score, how many wins they had, what percent of the wins were elimination wins, what % of wins were team wins, how many times they were at the bottom of challenges, how recent their most recent elimination win is, how different the number of their confessionals is from what would be considered an equal amount across chefs… honestly, it’s overwhelming how many things I could look at.

Let me take you through a few things. I’ll be using the term “odds.” Odds is a ratio of success and failure (in this case, winning and not winning). Odds are different than a probability of winning; probability would be the number of successes out of total successes and failures. I chose to look at odds because winning is a binary thing: you either win or you don’t.

I also show lower and upper confidence intervals. These numbers mean that I am 95% confident that the true odds are between the lower and upper values. If the lower and upper values fall below and over 1, respectively, then the odds are not statistically significant.

I did what isn’t the most well-regarded strategy in terms of an analysis…a totally exploratory analysis testing many different things. But stick with me - we’re all learning here.

All of the data shown here are all up to and through having four chefs in the competition, as a way to see if the winner can be predicted when there are three chefs left in the competition. I unfortunately only have complete data on confessionals (the individual interviews with contestants) for seasons 1, 2, and 21. Analyses with confessional data will then only be for those seasons - so that data should be interpreted with caution, since the number of data points is low compared to the number of seasons of Top Chef.

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Count The Dings
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share