Published in

Andbox

# Dominance Index v2.0

I’ve addressed the concept of Dominance Index (DI) in my previous post (https://owimt.medium.com/dominance-index-v1-0-evaluating-a-quality-of-a-win-244c6f79f049).

The concept of DI was made to evaluate the difference in team performances in a map. Simply put, a higher DI means a match was one sided a lower DI means a match was a close affair. However, the previous version of DI was not a perfect index because it was based on the final map score. The final map score might roughly indicate the difference in performance between the two teams, but we felt the need to be more specific.

Therefore, I used Relative Combat Power (RCP) of a teamfight instead of final map score as the standard. If the concept of RCP is not familiar to you, please check out my previous post on the subject (https://owimt.medium.com/relative-combat-power-rcp-v1-0-in-overwatch-league-6eb92bed3f7a).

The purpose of DI is to evaluate the quality of both teams’ performance on a map. There are several methods to evaluate DI, the most simple one being comparing the number of teamfights won.
i.e [(num_Team A_win)-(num_Team B_win)] / (Total num_Teamfights)

For example, if Team A won 6 teamfights while Team B won 4 teamfights, the DI of Team A is [(6)-(4)] / (10) = 0.2, and that of Team B = -0.2.

This method is very straightforward and simple to evaluate teams’ performances based on the teamfights, but it misses out on the details both in and out of the teamfights. For example, this model cannot differentiate between a map where a team won 3 teamfights out of 3, and a map where a team won 6 teamfights of 6 because in both cases DI is 1.0. Also, even though the teamfight results may be one sided in a map, the actual teamfights might have been very close. I wanted to develop an index that can take the quality of the teamfight in consideration.

# DI v2.0

Thus we need to evaluate the content of each teamfight. Since RCP is an indication of how the teamfight went, it is plausible to say the sum of RCPs during teamfights that occurred throughout a map reflects how much a team dominated the other.

However, simply adding up the RCP numbers would be misleading especially when a match has prolonged teamfights because a match with longer teafights have a higher RCP sum. Therefore, diving the total RCP sum by the duration of the teamfight should solve this problem and accurately portray the average RCP dominance of the teamfights.

Let’s take our second match of the 2021 season against HZS (https://youtu.be/HKYCvqSrIHw?t=414) as an example. Here are the first two teamfights of the first map of the match, Busan:Downtown.

You can see that NYXL both won the first (orange) and the second (red) teamfights. There were 6 final blows by NYXL in the first teamfight (none by HZS), and the total sum of the RCP was 51.00. There were 5 final blows by NYXL and 2 by HZS in the second teamfight, and the total sum of the RCP was 31.77. The teamfight durations of both two were pretty similar, 49.974s for the first one, and 47.892s for the second. Thus, the dominance of the first and the second were 1.0205 and 0.6633, respectively. In this perspective, we can say that the NYXL was about 54% more dominant in the first teamfight than in the second one. In order to get the dominance index for each round, map, or match, I can just simply calculate the average RCP_sum/s value for the corresponding level.

According to DI V2, the most dominant round in favor of NYXL was Busan:Downtown against HZS at 1.020, while the round that NYXL was dominated the hardest was Temple of Anubis against CDH at -0.801.

Future possible applications of this include inversing the coefficient of variation ((average of RCP_sum/s )/ (standard deviation of RCP_sum/s)) to evaluate the variation of teamfight dominance.

• English Revised by Minseong Kim (NYXL Player Manager)

--

--

## More from Andbox

Andbox brings people together across the New York Tri-State area to celebrate their passion for gaming with live events, digital content, competitions, & apparel. Home to NY’s professional esports teams & franchises and some of the leading gaming content creators in the world.

## Yong-Cheol "Imt" Jeong

Head of Data @NYXL. PhD in Neuroscience @KAIST, South Korea.