Select Page

The 2024 PGA season is here. Get the industry’s best projections

Our founders, Evan Silva and Adam Levitan, have always believed in connecting with talented people and sharing their work. Sometimes it leads to formal working opportunities, but it’s often about expanding their own knowledge while helping move the industry forward.

With that in mind, and with the recent addition of ETR Director of Analytics Mike Leone, ETR is creating a private, informal Slack group for those interested in the intersection of fantasy sports, gambling and analytics.

There are no strings attached. It will simply be a place to discuss ideas, projects, job opportunities, and meet and get help from others who have relevant skills and experience, including our fantasy sports analysts and business team. We plan to keep the bar high as far as admitting people to the group, though there are no official qualifications required. There are plenty of places to publicly read and discuss ideas (and we have nothing against those — we participate there, too), but this will be a place for people who have serious aspirations of working in the industry or for participating in relevant research or projects.

To be transparent, one of our goals as a company is to connect with more people who are interested in these topics, as we believe it could lead to finding contributors or ideas for our business. We plan to post paid and unpaid projects to the community. One initial project we are posting internally each for NFL and NBA is described below.

If you’re interested in joining, please send [email protected] a resume/CV, short bio about yourself, and any previous work or skills that you think are relevant. We will respond to everyone who submits an application.

-The ETR Team

 

Sample Research Projects

NFL Breakout Finder

One of the most interesting dynamics in seasonal fantasy football is finding the appropriate balance between median projections and upside. At ETR, we’ve developed a system for median projections that we’re proud of but would like to more scientifically approach upside.

Based on player profiles (age, rookie status, athletic profile, previous year performance, offensive line strength, team win total, etc.), we’d like to accomplish one or more of the following, ideally for all the skill positions but at least for RBs:

– Place player into buckets based on breakout potential (i.e. Jonathan Taylor may be in the highest breakout bucket, while someone like Leonard Fournette may be in a limited upside bucket)

– Place player into buckets based on anticipated role (i.e. bellcow, pure handcuff, third-down back, etc.)

– Predict a player’s ADP next season

– Give each player a breakout score

After the season, we’ll be able to combine this more macro player analysis regarding a player’s breakout potential/role with our median projections. Analyzing the relationship between the two will hopefully allow us to find the proper equilibrium when drafting seasonal fantasy football teams.

Sean Koerner of Action Network recently wrote an article regarding his RB Upside Ratings, which is very insightful on this topic: https://twitter.com/the_oddsmaker/status/1294782378305445888?s=21

 

True NBA Player Positions

As the NBA moves towards “positionless” basketball, traditional positions lose some of their value. We want to use a player’s position to help predict that player’s baselines appropriately and to get more applicable DvP. One of the best ways to do this is to cluster NBA players by different statistics. These clusters then serve as a better position designation than a player’s default assigned position.

It’s important to try and cluster players by their roles on the court – shot selection, usage and assist rates, tracking data (dribbles, touches, rebounding chances), rebounding data, defensive data, and talent (efficiency), along with other things. Ideally, we can have two sets of clustering, one that is broader (6 – 8 clusters) to ensure a large sample size, and one that is narrower (12 – 16 clusters) to ensure the appropriate level of specificity.

References:

https://towardsdatascience.com/using-k-means-clustering-algorithm-to-redefine-nba-positions-and-explore-roster-construction-8cd0f9a96dbb

https://dribbleanalytics.blog/2019/04/positional-clustering/

https://github.com/acheng1230/NBA_Player_Types

https://www.wired.com/2013/03/basketballs-hidden-positions/

https://towardsdatascience.com/clustering-nba-playstyles-using-machine-learning-8c7e8e23c90c