Identify newly desired data and potential sources

[michalekb11]

The player stats data we currently have is not enough to be effective at modeling.

Ideally, I want to have the following data (and be able to continuously retrieve this data before upcoming games):

• Injury data (severity, when did it happen, how many games missed, etc.) https://www.rotowire.com/hockey/injury-report.php
• News article performance (How do sports writers think they have looked in the last few games? Fast? Slow? Favoring his right leg? Could he be injured but still playing?). Think of NLP transformer applications. It is unlikely that DK uses information like this, which makes it valuable to us. https://www.hockey-reference.com/players/news.fcgi?id=ahose01. https://www.dailyfaceoff.com/hockey-player-news
• Linemate data (performance of linemates in the past 5 games)...(https://www.dailyfantasyfuel.com/nhl/projections/draftkings???)
• Mid season trades and offseason move data
• Biometric data (how fast is his wrist shot?)
• Corsi (https://www.hockey-reference.com/leagues/NHL_2023_skaters-advanced.html), power play TOI per game (https://www.hockey-reference.com/leagues/NHL_2023_skaters-time-on-ice.html),
• Etc...

[michalekb11]

When we find data sources, we need to get the data ASAP or build a way to continuously scrape it before every game. I want to go into next season with as much working data collection as possible.

[michalekb11]

From issue 2:

• % time in offensive zone?

• Number of offensive zone faceoffs

• Point streak. Number of days in a row with a point or without a point. (ex: 2 = last 2 games he had a point, -3 = last 3 games without a point)

• Linemates / team chemistry (this one will be difficult...)

  • One thing we could do as an estimate even if it isn't that great is create an overall score for the team's. For example, could just be the teams points relative to average or something. In general, better team standings means better team which means better players and thus better offensive production around you (assuming "you" is referring to any individual player on the team). Idk, still haven't sorted this one out.
  • Also could check this out (https://www.hockey-reference.com/players/a/ahose01/scoring/2022). We could do the following steps: rank every player in the NHL (maybe just by points or something, just come up with a metric). Then, take a weighted average of the ranks of the players that each NHL player collected his points alongside of. For example, Sebastian Aho collects a lot of his points from Teravainen, who would probably be slightly above average in rank. This means that Aho often plays alongside a good player (someone who gives Aho a lot of chances to score points).

• Traded in last X games

  • We should be able to tell since when we scrape day of lineups, they will not be on the same team anymore. Should get flagged and status set as traded (or just "moved teams")