Effectively Wild Episode 1853: What Are the Odds?

by
May 26, 2022

EWFI
Ben Lindbergh and Meg Rowley talk to Kelly Pracht, the CEO and co-founder of predictive analytics startup nVenue, which has provided the real-time probabilities displayed on this season’s MLB Network-produced Friday Night Baseball broadcasts on Apple TV+. They discuss nVenue’s origin story, its sports-betting ambitions, its 100-plus-input machine-learning model, which factors are and aren’t predictive of performance, Ben and Meg’s misgivings about some of the displayed probabilities, and much more. Then (1:04:37) Ben and Meg bring on FanGraphs writer Ben Clemens to discuss the results of his study about how nVenue’s odds compare to a simplistic, one-factor model, and why they think the accuracy of the system matters.

Audio intro: Remember Sports, “Odds Are”
Audio interstitial: Sunflower Bean, “Beat the Odds”
Audio outro: The Rock*A*Teens, “Count in Odd Numbers”

Link to Friday Night Baseball details
Link to nVenue’s website
Link to article about nVenue fundraising
Link to SportTechie on nVenue
Link to SportTechie on nVenue again
Link to D Magazine on nVenue
Link to InnovationMap on nVenue
Link to nVenue YouTube video
Link to nVenue on PitchBook
Link to nVenue on Crunchbase
Link to Emily Bender on AI
Link to machine learning wiki
Link to overfitting explainer
Link to SABR on machine learning
Link to “reach base probability” tweets
Link to Ben Clemens’s nVenue study
Link to Ben’s study data
Link to Brier score wiki
Link to league count splits

 Sponsor Us on Patreon
Subscribe to Stathead (Code: WILD20)
 Facebook Group
 Twitter Account
 EW Subreddit
 Effectively Wild Wiki
 iTunes Feed (Please rate and review us!)
 Get Our Merch!
 Email Us: podcast@fangraphs.com

Podcast (effectively-wild): Play in new window | Download

Subscribe: RSS





 

5 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
68FCmember
1 year ago

Listening to the Kelly Pracht interview was painful, it was a perfect example of someone knowing just enough to be dangerous. It’s wild that someone would have the audacity to build a model to predict baseball and just completely disregard and not even consider the last 40 years of baseball research and understanding. A complete lack of understanding of sample sizes.

It is so clear that the entire project is just to manipulate people into microbetting by feeding them incomplete or misleading data.

13
tomerafan
1 year ago
Reply to  68FC

Apologies if I missed it in the interview, because it was indeed painful, and I tuned out and then dropped out entirely.

But is the algorithm computing the probability of an event happening, or computing the level at which bets on either side of that number would even out? That to me is the key question. Think about a line on a football game… saying a team is favored by 7 points doesn’t mean that the oddsmakers think that there is a statistical probability will score seven more points. Rather, it’s the point at which the house evens their own odds on the bets that are made.

0
marechalmember
1 year ago
Reply to  68FC

What was particularly striking was her arrogance (often typical in self-described “technologists”). When asked whether they had consulted any baseball expert, she brushed it aside because she and her colleagues are “baseball fanatics”. Truly incredible stuff. The types of things that she pointed out as being relevant (whether a batter went 0-2 in his previous two at bats, whether he is in a slump, etc) were so shocking. Totally agree with the conclusion above.

Good job by Ben and Meg to get that out there in the open.

Last edited 1 year ago by marechal
5