Since hockey season is starting soon I decided to do some ideas on hockey and sport related Ideas. Food related ideas will appear in between don’t worry if that’s what you’re in for.
Sport analytic is a super interesting field and with all the api available nowadays you could probably create your own analytic program to help you pick the right player for your fantasy teams or your betting needs. There are the standard stats that you can find almost everywhere, which is game played, goal, assist, point, shot, plus/minus, time on ice (toi), etc. However, you might not know that there are advanced stats like corsi, fenwick, etc. (See https://www.hockey-reference.com/analytics/).
Using an available api and a web scraper for stat not available via api to create a more complete portrait of the players and the team they are on. With past data and current trends it could help you find if a player is on a downward trend or an upward one.
How to build your own analytic software (bot?)
The first step would be to gather the data about the team and the players. In (Idea #1: Manage your fantasy hockey draft or any other sport via facebook ) we discuss different api to get the data so this is a good start. Once you get data from the previous year and daily updates you can find a way to display the trends efficiently to be usable for your fantasy hockey and betting needs. Example, when a player is injured, it can help you pick the right player to replace him from waivers. It could identify underperforming players that usually bounce back later in the season.
If you’re feeling fancy and have some knowledge of AI, you could feed the data to a neural network to see what kind of selection the analytic software will pick. With enough training, you might just let the bot do the job for you.
What will the bot help you do ?
They say that in general the peak performance of an athlete is around 27 years of age. There are exceptions to this like Kadri 87pt career high at 31 and Letang career high at 35.
You could probably predict if the player will have upward or downward trends according to past statistics, age and line composition. Not sure any analytic tool would have predicted Kadri or Letang career year though. Your model should be reevaluated at the end of the year if your analysis were right and correct itself. If you have access to a competitor you could verify who was right or wrong and find out who year after year has the most accurate prediction.
Why build that kind of analytic bot?
The original idea came to me while reading about robo advisers and bots picking stocks. While there is a whole other world on stock picking, balance sheet, forecast and all jazz that I don’t think I’m qualified to give you advices about. The base idea is similar, you have a set of data that help you predict whether the stock should go up or down and you (or the bot) buy or sell appropriately.
My first usage was going to be for a frozen pool I do with friends every year. The goal was just to help us with the pick, we would be mostly making the decisions ourselves at the end. However, with the rise of online betting recently with Draft king and others betting platforms, I wondered if it could be used for that too. Therefore, stacking the odds in your favor.
And finally, with the plethora of pools only, I thought building a bot to manage them all would be a good idea, helping finding gems, managing it instead of myself and helping the bot learn.