Boards›Esports

I made my own Power Ranking System with Math

xayahs bich·5/10/2015, 5:54:41 PM·7 votes·2,437 views

First off here is a picture of the equation http://imgur.com/AT0EfZ6 I used the top 6 teams in the NA LCS and used their results from the regular season to calculate their rank using this equation. I only used the top 6 because it takes about an hour to calculate a single teams result. What I got was... TSM- 5.80 Cloud 9- 4.66 CLG- 4.65 TiP- 4.45 Liquid- 4.44 Gravity- 4.43 As you can see the results are pretty tight in some cases and then there is TSM. TSM is the result of having several outlier games of just demolishing teams and having no games in which they themselves got demolished. The equation isn't perfect but it is a pretty accurate placement of the teams and I think it could be used as an alternative to the "voting by a panel of experts". Reasoning for parts of the equation: The (5(114xtime)) is to get rid of all gold earned passively (not including passive gold on items) by players and counts only earned gold. The reasoning for the +1s by (Team KDA/100) and (team deaths/team kills) is to be a stablizer and prevent punishing teams for having a worse KDA but being rewarding teams with negative KDAs.

22 Comments

delonix5/10/2015, 7:47:11 PM1 votes

Dumb question, but is there a way for me to rotate the image here or at imgur without having to make a copy on my computer and rotate it?

The advantage that the panel of experts has is that it can take into account things like variable opponent strength and weird team drama stuff like KEITH and Piglet switching back and forth. Your formula would have been a decent predictor of the playoffs, but it obviously underestimates Liquid, which beat both TiP at 0.01 advantage and CLG at 0.21 advantage.

There's probably also some analysis that can't be done by math. The playoffs are best of 5 instead of best of 1, which might favor certain teams. Some play styles might beat others.

I'd be interested to see the results of the rest of the teams or the EU LCS top 6 or both to see if there are any outliers.

xayahs bich5/10/2015, 8:30:37 PM1 votes

Here is the flipped picture for all those who can't read sideways

http://imgur.com/LMNSD10

MasterSeth20255/10/2015, 10:13:09 PM1 votes

This would be pretty cool for each individual region as a season summary, however the panel of experts can notice changes in teams faster and adjust them accordingly where the math would take more games to put out similar results. However we could never use this for the international rankings because the teams come from different regions and thus have different competition to gain statistics against. For example, if you compared SKT1 with Fnatic, the numbers could be similar, however the teams have very different skill levels (I actually thought Fnatic did pretty good though). But this is pretty cool, and it does reflect the teams pretty accurately.

ratminer5/11/2015, 12:43:53 AM1 votes

why 114?

passive gold over a game is

(5(475+(total time-90)1.9))

5 for all players 475 for starting gold total time measured in seconds -90 because you dont generate gold until 1:30 1.9 because passive gold generation is 19g/10s

Delioth5/11/2015, 12:56:27 AM1 votes

KK, to try and make your job easier I wrote a program to do this. I'm getting awkward answers due to my lack of actual stats- where are you getting stats from? I can probably give you a working thing if I can make sure that it gives proper answers. I don't think there's an error in it, pretty sure it's due to random numbers being used.

EDIT: I'm also kinda wondering on each of your figures- are you counting damage as total, or as thousands? Same with gold. I'm getting big numbers, but they come down into reasonable numbers when I divide made-up average numbers by 1000. And are these numbers averages or totals?