[Intro] Welcome to the OWL Role Star Power

Hi, I'm the one that really adore and analyze the Overwatch League like anyone who will be watching this.

I'm majoring in Statistics trying to make something with Artificial Neural Network (ANN). Well, it is always enjoyable to rank sports players. But everyone has different criteria. Can AI help? It already does!

 

OWL has its own Power Rankings just like other professional sports. The league had introduced Player Impact Rating (PIR) in 2019 and started IBM Power Rankings this season. But the problem is that I'm not completely satisfied yet.

1. IBM PR really does not explain how it works.

2. It looks so accurate that it lacks humane nature.

 

Here's my thought of Power model.

1. It only needs to compare the top players appropriately. You know, it takes more than a month to quarrel about 'Messi VS Ronaldo' { Spoiler Alert - Messi Won :) }

2. It should be FUN. It is for 'killing time' not for scientific verification.

3. To be fun, it should exaggerate anything possible and thus make mistakes.

 

10% of people watch IBM PR with awe of an AI. Another 10% listen to various PR news with sincerity. The other 80% watch Plat Chat PR to hope for its entire collapse. To be honest, the exaggerated and wrong but reasonable PR is all what we're looking for.

 

To satisfy my ideology of PR, how can we quantify the power of players? I think the answer can be 'Impact' and 'Role Stars' is what embodies impact.

 

Whole list of Role Stars / Any Complaints?

 

Starting from 2019, 'Stars' are selected by those related to OWL regarding regular season (before Play-In & Play-Off kick off) {There was no such system in Inaugural Season}. How could you be voted as Stars? To win Finals? To kill six with a Rein charge? To finish the season with 0 death? { Iddqd did it tho :( }

 

No, the answer is 'Just get as many as votes whatever it takes'. If a player can show constant and tremendous impact, the player may well be considered as role stars.

 

Assumptions

1. It may be False that 'High-impact Player' is one of the best players. But realistically, it is usually True.

- So it is reasonable to consider high-rank 'Role Star Powers' as good players.

2. Impact consists of multiple elements.

- So I brought all the variables that seem to be related to the 'Impact' from public data of OWL Statslab.

 

Under these assumptions, I made 'Role Stars Power' using Artificial Neural Network (ANN).

I will explain details on the next post. To be continued...