Teamfight Analysis

 We all know that less 'Deaths' (same as more 'Final Blows') from a team brings a map win (https://owimt.blogspot.com/2021/01/overwatch-league-owl-provides-player.html). 

It is quite obvious and simple, but the underlying process to get final blows is not that simple. I also analyzed the important stats of each hero for winning  (https://owimt.blogspot.com/2021/01/important-stats-of-each-hero.html), but it is almost impossible to say that the "Dragonblade Kills" stat is always the most important stat for Genji every single time. It is likely that the important stat is different for each final blow, which means that 'Damage - Swift Strike', 'Quick Melee Accuracy', 'Deflection Kills' or even 'Healing Received' could be the most important one depending on the situation.

Therefore, we need to break down the contents of a match for analyzing in detail. Currently, I believe that a "teamfight" is the finest level of gameplay we can break down in order to evaluate the effect of the stats.

Then, what is a teamfight? 

Every coach, professional player, or even every casual Overwatch player may think they can easily distinguish the start and end point of a teamfight, but it's not. This is because there is no strict definition of a teamfight. Interestingly, this issue has already been discussed elsewhere (https://overwatchleague.com/en-us/news/22919644/what-is-a-teamfight) by Ben "CaptainPlanet" Trautman from Blizzard Ent. He used the 'Burke Model' to define the end point of a teamfight, but unfortunately, it was impossible to get the model, nor the relevant data from Blizzard. 

So instead, here I used the model I developed to define teamfights. I call this model the 'Teamfight Detector' (version 1.0) and will be updated if critical flaws are found. The model is based on my perspective as an OWL coach, and the perspectives of other OWL coaches and players as well. Writing about this model will be a bit complicated, so I will touching this matter in a later post.

Following is a simple example of a teamfight with two stats, Hero Damage Done (HDD) and TPB (team power balance, which will be discussed in the next post). Here, HDD is the summation of both teams' HDD.

The green line indicates (HDD / 10 seconds), and the black one indicates TPB. The grey-shaded area indicates that there was a teamfight.
You can see that there are 5 teamfights in round 1 and 7 teamfights in round 2.

Here I present a sample of a teamfight summary on one of our 2020 season Countdown Cup match vs Guangzhou Charge (match video: https://youtu.be/0_GroZeCjqQ). I remember that this match was meaningful because GZC was the previous Summer Showdown champion and it was a revenge match (we lost 0:3 vs GZC at the Summer Showdown). Haksal got MVP for this series, and I'll analyze his performance in the next post.

The result was a 3:2 win, but the contents were not very satisfying. Let's take a look at the teamfight result summary.

You can find the detailed information when you hover your mouse over the icons. 

You can see each teamfight (TF) and its winner, win rate in the form of a pie chart, map winner and the match winner. You can also find the team power balance (TPB) and TF duration info when you hover your mouse over the TF order icons. 

As you can guess, each team icon means the winner of the TF. The size of team icons were determined by TPB values of the TF. TPB is a simple indicator I made to evaluate the team's dominance during the TF and I will cover this in detail on the next post. But here, it is safe to understand that if the TPB is high, a one team was dominant in the teamfight whereas if the TPB is low, the teamfight was a close affair. 

Now you may see the reason why I mentioned that the match contents were not very good for us because the number of TF wins was high at the first two maps (Busan, 8:4 and Eichenwalde, 10:2), but it was terrible at the third and fourth maps (Hanamura, 1:9 and Route 66, 2:10). Surprisingly, in the last map, we were behind in the number of TF wins (Lijiang Tower, 6:7) but still won the map. This is because GZC C9ed the first round. 

Overall, we played really well on maps 1 and 2, played badly on maps 3 and 4, and for map 5 it was close but we won.

You can also find the major stat reports for every teamfight below. I only posted 12 major chosen stats because there are almost a thousand stats that can be categorized. 

please enlarge the table to see the whole data. (it will take some time to load because the data contains stats from all TF).

I will show you an example of a teamfight analysis with these stats next time :)