Blizzard, please have Mercy on WinstonsLab’s PTK model!

October 05th
written by Karahol

ptk model mercy

The reworked Mercy has brought forth lots of discussion about Blizzard enabling a low skill ceiling hero in the game, breaking a lot of things in the process with SR gained/lost being at the top of most people’s list. Another thing that most note is that it affects the fights in a very negative way, lowering the impact of First Kills and also making any scenario where one of the teams does not run a Mercy, an uphill battle, a 6v7.

Still, the problems with Mercy’s design ran deeper than just SR gains and First Kills and I have written another piece here, on WinstonsLab, where I examined what stats are important for her and my analysis concluded that her design, so different from that of the other Healer heroes, actually makes it very difficult to evaluate her performance and balance her. This time around, since now we know exactly the direction Blizzard intends to take, I will approach the subject from a different angle, that of Mercy lacking kill potential.


The Enabler

Out of all heroes in the game, she is the only one where kills as a stat tells us pretty much nothing about her performance. She’s not designed to kill. And in-game there are many stats categories for kills, such as Objective kills, Solo kills, Final Blows, Melee Final Blows etc. In all of these categories, her stats are pretty close to 0 on average the higher you go in ranks, because the only times you want a Mercy to shoot at someone is when she has no one else around to heal. The following stats are Faze Clan’s Rawkus’s from Contenders S1 NA.

Contenders NA S1_FaZe Clan_Rawkus_Mercy copy copy

Every other Support hero that heals, like Zenyatta, Lucio & Ana, also deals damage and when we look at a player’s stats, we can get a pretty good idea if that person plays aggressively or passively. If he focuses on healing his teammates and prioritizing enabling them than getting frags. In other cases, based on the kills he gets, we can estimate his impact once his team’s DPS/Tanks are down and Supports are left alive to bring back the game in favor of their team. Finally, during the opening phase of engagements, where not a lot of damage has been dealt, all of these heroes (and the players that play them) prioritize dealing damage, while Mercy can’t really do that, because she loses utility of her healing/damage boosting when she switches weapons.


No kills, no star plays

On WinstonsLab, while examining the data and stats gathered by tournaments we started trying to find ways to translate a player’s stats on the team level and evaluate his contribution. One of the stats I invented was PTK (percentage of team’s kills attributed to a player) that Barroi then took and created his PTK model that estimates the kill distribution on a team basis for Tanks, Suppors and DPS players. Obviously, some times there is an overlap role-wise, as in HarryHook playing Soldier76 on certain maps while being mostly the Lucio player for Team EnvyUs. However, these instances are few and far between.

What this model managed to capture was something very interesting. When put in the proper context, that of tough matches for teams -strong competition- it could tell us if a team was carried by their DPS players, their Tanks and how much the Supports contributed to the team’s overall performance. The more kills the Supports had on their names, the better the team’s standing in the competition was, because in very close matches, where DPS players can be taken out of the engagement equation quite fast (even for slight mistakes), it’s up to the Supports to make the plays and get the kills while Tanks try to provide cover and add damage. Everyone needs to step up on a team, when the matches are very close, like those of APEX S3. Below you can see the PTK distribution for the season for the grand finalists, Lunatic-Hai & Kongdoo Panthera.


With the reworked Mercy in-game and her must-pick nature, Supports’ PTK will drop significantly. I am not 100% sure as to which Support heroes combos we will see mostly, if the changes to Mercy are to persist, but if we take the example of a Mercy-Lucio combo, only Lucio will be able to secure kills. The overall support PTK will plummet even more, while Tanks’ and DPS’ PTK will get boosted. This will also affect screen time of the Supports players, especially the Mercy one’s, due to the fact that her playstyle is really passive and her most “wow” ability is now something that no one cares to see in first person. At least with the 5-man resurrections beforehand, due to the Resurrection being an Ultimate and its charge time, and its role in a team’s strategy altogether (suicide on point), observers were prepared for it and could get a proper angle to show it. Valkyrie now will be so hard to spectate and measure its impact on the battlefield outside resurrections.


Closing thoughts

In the end, it’s not only about whether Resurrection’s cooldown needs to be adjusted or whether her passive regeneration rate needs to be lowered, so as to make her more bearable in ranked. Mercy with her kit and playstyle effectively moves the game in a direction where 5/6 members play ball and score, while the 6th one just constantly passes the ball to his teammates and never even attempts to shoot. And I am not sure if Blizzard intends to introduce more heroes like her, pure Supports, but I would advise them not to. It is better to have the same principles design-wise when you are building an esport, at least on the role level, than constantly alternating the kits and introducing new elements allowing for every possible play style. It is my opinion and understanding, that in sports and competitions you find optimal ways to maximize depth of strategy and let skill shine, by minimizing chaos as much as possible. It can work in MOBAs, because they have standard elements like map layout and overall game mode strategy remaining pretty much the same, but in a game like Overwatch, with FPS elements and a multitude of game modes and map layouts, it can get quite problematic.

So, Blizzard, we humbly ask you: Please, have Mercy on our PTK model!