What Stats are Important [Pt.1] – Ground Stats

June 17th
written by Karahol


With Blizzard having announced an API for Overwatch and with the 10-minute averages being in the PTR -soon to be implemented, it’s the perfect time to start discussing what stats do we need the API to provide us with, so as to create infographics that represent the true potential and performance of players/teams and also to replace the stats presented in-game, which most of the time are useless (because they create delusions) with stats that clearly show to players if they are underperforming, overperforming or doing as expected.

In this first piece, I am gonna lay out the framework of my thought process and focus on the Ground Stats (G-stats).


Ground and WOW stats!

Before I move on and break down what I would like to see for each hero stats-wise, I would like to make a distinction between two sets of numbers that can be used to measure a performance: the ground stats and the WOW stats.

Ground stats (G-stats) are stats that you need to gather for every single hero’s performance, such as Kills / Deaths / Ultimates used etc and you always need these numbers in a fighting-oriented game. You still have to weight them differently for each and every hero, but you can’t disregard them.

WOW stats show the potential of a player and his expertise in some areas of the game and heroes, but they don’t tell much on their own. Sometimes are even misleading. For example, opponents flashbanged by a McCree is a WOW stat, or Hook Accuracy for Roadhog. However, as I will explain in my next pieces where I will discuss stats on a per hero basis, it’s more important to know how many flashbangs led to a kill, for McCree, or hooks, for Roadhog, than how many opponents a McCree flashbanged or a Roadhog hooked.

I make this distinction because in some cases, especially for teams that lose a match, their G-stats will always be lower than the winning team’s. So, WOW stats will be there to give us an idea of how well each player on the losing team did and his potential for future matches.


Stats are not PATCH-resistant

The other thing one needs to keep in mind is the fact that patches in Overwatch have the inherent capability of upsetting which stats are important and which are not, each and every time.

Stats that were important in one patch may be not as important in another patch and therefore, before someone goes on and starts evaluating the numbers or gathering them, one must always formulate a model in his mind about which things are crucial and which are secondary per case.

For example, back in the Quadruple Tank meta, Ana’s healing was probably at its peak since she could constantly heal up all those beefy targets with the biggest health pools in-game. Right now her healing is significantly lowered in averaged numbers since she needs to heal targets with small health pools, but her effectiveness remains the same. So Healing Done for Ana should be evaluated differently in every patch.

Or Roadhog’s Hook Accuracy, before the changes to his hook mechanism, could indicate how great a player was at securing picks, since a hook was translated in 95% of the cases into a kill, but now that’s almost irrelevant, because it’s much harder to get the kill even if you get the player hooked.

In other words, reading stats and finding the combinations of them that give us the proper picture is an ever-going process that will never cease, as long as patches shift the balance.


Stats in-game that are useless

In the profile of a player, one should expect to see the general stats. So, let’s go through what stats Blizzard provides us with right now for all heroes and what stats need to be added or to be changed.

First of all, after the patch we will have the averaged per 10 mins stats that will get us more accurate numbers, in general, so keep that in mind.

Right now in the career profile, a lot of stats try to just make players feel good about themselves. I have colored red all the stats which I deem not very useful.


Total stats, as in the sense of total damage done, total deaths, total eliminations etc. are “totally” useless. These numbers mean nothing, because they are affected by how many games a player has played in general and how well he/she did in them. Someone with more games will, unsurprisingly, have a higher number.

Hero Damage and Barrier damage are on the way to be added, so I won’t mention these. Damage done / Healing done are fine as they are and they will be even more insightful once we have them averaged per 10 minutes. Assists are also fine as they are.

1) Eliminations is a weird stat for me. This number should be replaced with the Kill Participation one. Eliminations is like a glorified Kill participation, where a player gets credit for every kill he contributed to but his level of participation in the kill is ignored. It could be anything from 1% to 99% and we still can’t know about it. Why is this important, you may wonder? Because in Overwatch everyone does damage and solo kills, especially in the top competition, aren’t so common. So, we need a metric that shows us how kills are distributed properly so as then to evaluate damage done by a hero.

2) Final Blows are essentially what we consider Kills in WinstonsLab.

3) Objective kills is another glorified stat that doesn’t provide much info, because of the way the objective works. You can secure 15 kills around the objective that are of great importance, but only kills on the objective (close to the payload or on the cap box/circle) will count towards this metric. It doesn’t provide much insight at all.

4) Objective time also has little to no value for many heroes, even in the case of pushing the payload. First of all, this metric doesn’t give us any info on whether a player stayed on the objective defending it or attacking it (pushing the payload, for example). Secondly, this metric means nothing for many heroes because of the way teams approach fights and allocate resources to capping objectives. It needs to be replaced with a metric that counts how long a player has contested an objective while defending it, at least.

5) Time Spent On Fire is a WOW stat but it really suffers from the way a player gains fire points and therefore from a competitive point of view, it’s not that important. I would prefer a metric indicating how long on average a player stays alive on a hero between two consecutive deaths. I would call it “Hero Uptime”.

6) Melee Final Blows is another weird stat. It’s definitely a WOW factor, because meleeing someone is always a sign of prowess and clutch situations, but it’s definitely not a general stat. It doesn’t provide much insight. Some heroes can’t even melee and some don’t need it even.

The rest of the stats in the “All Heroes” tab don’t hold much value, like cards and medals. They can be there, but again they are dependent on the number of matches one has played.


Which stats need to be included as G-Stats (Phase 1)?

1) Ultimates Used should be alongside deaths and kills, despite having a wide deviation because of the heroes’ ultimates charging differently. Even though at first glance it won’t make much sense, it will definitely help combined with the info of time played per hero.

2) Although Headshot Accuracy isn’t a metric for every hero in-game, it’s still a metric that one would like to see when viewing a player’s stats, right off the bat. And a step further, Headshot Kills %: how many of one’s kills were headshots is also another nice indicator. Then, in the heroes tabs, one will see the breakdown of this number.

3) PTK or Percentage of Team’s Kills can give great insight and has been used greatly in WinstonsLab lately to show how well-rounded a team is. In a single player’s case, this number can instantly tell you if the player is DPS-oriented or not, because DPS players have higher PTK on average. If a player is not maining DPS heroes, this number will tell you how aggressive he is and how well he implements his aggressiveness either as a Tank or a Support. A low PTK will tell you that this player is playing passively mostly.

4) First Death (FD) & First Kill (FK) is a great pair of indicators about the player’s performance during the first seconds of every engagement. A high value in First Death means that said player lacks positioning or good movement in most cases. A high value in First Kills means that he’s a great opener so as to gain the early advantage in a fight for his team. This can apply to all heroes, especially in the current Dive meta.

5) Kills per Ult (KPU) is another great stat indicator for most classes as to how well or badly they do with their ultimate, since most Ultimates in-game are attack-oriented. Obviously it doesn’t work in cases where the Ultimate doesn’t have attack power.


Stats Visualization

And another thing that Blizzard needs to do is to visualize the stats so that people don’t need to read text lines when they browse their stats. The whole stats page should be visualized like the overview page. Icons for every stat category should be introduced that will make stats-browsing much easier.


I have already introduced some of my own, with the help of Hexo, a gfx artist, so as to get to understand really fast, at a single glance, what one needs to understand. You can find them in my infographics.picks


Closing thoughts

This is just the first wave of stats that need to be included, the ones we know for sure that have value. There are more that one can gather, but for the first overhaul, I believe these changes will suffice.

In the next piece I will be focusing on the Attack category and which specific stats for each hero in that category I would like to see, that I think will help us evaluate a player’s performance better.