Below is our Model 284 consensus bracket for the 2021 NCAA Tournament. As you will see from our Game-by-Game Predictions and Tournament Simulation results, our bracket does not necessarily advance our model’s predicted winner for every single game. Rather, we use a combination of (1) our model’s individual game predictions, (2) our tournament simulation results, (3) injuries / other factors not captured by our models, and (4) consideration of public picks – to make sure we are differentiating our bracket enough from the most popular choices. For those interested, here is a bracket filled out purely using raw model output. Continue reading Model 284 Bracket | March Madness 2021
Tournament Simulation | March Madness 2021
The article below shows how our models view this year’s tournament field from a simulation perspective. Each table shows the % chance that each team advances to a given round of the tournament (e.g., in the first table, Gonzaga has a 92.5% chance of advancing to the 2nd round). These figures are calculated based on 1,000 simulations of this year’s tournament, which were performed using win probability and spread predictions from our models. See our related posts for our Full 2021 Tournament Bracket and our Game-by-Game Predictions. Continue reading Tournament Simulation | March Madness 2021
2021 Model 284 March Madness Pool
Can you beat the Model? Enter a March Madness bracket into the group Model 284 for a chance to take on The Model.
2021 NFL Big Data Bowl Submission
A Framework for Accessing Individual Defensive Performance in Coverage
1. Introduction
Current defensive coverage metrics are only calculated on plays where that defender’s assigned offensive player is the one targeted. This approach is flawed and incomplete. On any given play, there are around five available receivers, and the best coverage may prevent the quarterback from even throwing the ball to a given receiver. Further, these metrics identify coverage assignment simply by considering the responsible defender to be the nearest one at the time the ball arrives. Thus, this methodology only gives us coverage assignment for one player per play, and even that assignment is questionable; the closest defender at one particular moment isn’t necessarily the assigned defender.
The work presented here illustrates a framework we’ve developed to address these shortcomings. Our approach begins by identifying the responsibilities of each defender on a given play. If a player is in man (everywhere) coverage, he is assigned a specific offensive player whom he is to cover until the ball is thrown. On the other hand, if a player is in zone coverage, he is instead responsible for an area and any receivers who enter that area. Because the responsibilities of a player differ in each coverage type, the evaluation of performance should be different for each. Once we identify the type of coverage being played by each defender, we then consider two separate time points for evaluation. For one, we consider the time frame after the pass is thrown. This is the most straightforward but limits us to only the targeted receivers, as players tend to abandon their assignments and converge to the ball once it is thrown. To include all defenders in our analysis, we also consider the time frame between the snap and the pass. Figure 1 below illustrates our proposed framework…….
Read our full paper on Kagge here: https://www.kaggle.com/model284/a-defensive-player-coverage-evaluation-framework
2020 Peak NBA Statline Projection Model
Peak NBA Statline Projection (PNSP) is a model used to project NBA success for college and International basketball players. PNSP considers each player’s individual and team statistics, physical measurements, high school scouting ranking, and age/experience. The PNSP model returns a single rating value from 0 to 100, with higher values indicating a “better” NBA prospect. We provide a detailed article outlining how PNSP is formulated here, and PNSP rankings from previous years can be found here. Below are a few highlights for the 2020 NBA Draft Class, and a listing of each player’s PNSP.
2019 NHL Draft Prospect Sentiment Analysis
Continuing our analysis of 2019 NHL Draft Prospects, this article looks at Scouting Reports and tweets on 2019 NHL Draft eligible prospects. In hopes of capturing some additional information outside of the numbers as seen by the “public” scouts. If you follow our NBA Draft content, you are already familiar with our sentiment analysis on 2019 NBA Draft Prospects. For those that are unaware, sentiment is a form of NLP (Natural Language Processing) or more formally defined as “the process of computationally identifying and categorizing opinions expressed in a piece of text, especially in order to determine whether the writer’s attitude towards a particular topic, product, etc., is positive, negative, or neutral.” Using Professor Bing Liu’s graded sentiment dictionary, we can analyze text to and identify who is more positively written about. For more information, please read our My Model Monday series article.
2019 NHL Draft Role Probability Model
Our NHL Draft Role Probability Model uses prospects’ statistical production, physical measurements, league prominence, and team statistics to predict the likelihood that a players assume a specific NHL Role (i.e., First Line / Top Pair, Second Line / 2nd pair defensemen, 3rd Line / 3rd Pair, Fourth Line / 7th defensemen, Two-way player, and Non-NHL player). This Model expands on our NHL Translated PPG model in order to account for variables outside of Points Per Game, such as size, scouting ranks, caliber of leagues played in, etc. Continue reading 2019 NHL Draft Role Probability Model
2019 NHL Draft Prospects Translated PPG
This Friday, June 21st, 217 NHL prospects will be drafted with the hopes of becoming the next Sidney Crosby, Jamie Benn, or Evgeny Kuznetsov. Teams are faced with the difficult task of projecting young adults to the NHL, with some of those players coming from leagues that don’t even track second assists, and others coming from the highest levels of hockey. Analysts, Scouts, GMs, etc. will scour eliteprospects and hockeydb, checking out how many points a prospect produced in each league, and then try to use their small DB of memory to remember what NHL players also played in that league, and how they performed at the NHL level. Instead of using our 0.1 GB RAM human brains, we will use our 16GB RAM MacBook Pro to reference historical data necessary to translate point production across leagues for 2019 NHL Draft Eligible prospects.
2019 NBA Draft Prospect Sentiment Analysis
If you are a frequent reader of Model 284, you can recall we did a sentiment analysis on 2018 NBA Draft Prospects based on scouting reports at The Stepian and NBADraft.com. For those that are unaware, sentiment is a form of NLP (Natural Language Processing) or more formally defined as “the process of computationally identifying and categorizing opinions expressed in a piece of text, especially in order to determine whether the writer’s attitude towards a particular topic, product, etc., is positive, negative, or neutral.” Using Professor Bing Liu’s graded sentiment dictionary, we can analyze text to and identify who is more positively written about. For more information, please read our My Model Monday series article.
2019 Similarity Score Tool
Select any 2019 NBA Draft prospect from the drop-down menu below to view their top 10 most similar college basketball players, as measured by our Similarity Score model. Similarity Scores provide insight into how a player will translate to the NBA based on how their historical comparisons have performed in the NBA. This model considers a player’s college production, physical measurements, and age/experience to generate their most similar historical players. For more background on the calculation of the similarity scores, see this article. This model is one of three pieces that we use to evaluate the NBA potential of college players, with the other two being PNSP and NBA Role Probability Model.
Continue reading 2019 Similarity Score Tool