OVERWATCH HERO META REPORT #1: MONKEY BUSINESS

What's up guys and gals, CaptainPlanet here and this week I have something exciting to present: my first Overwatch Hero Meta Report for the competitive scene of Overwatch. This project will be a weekly work-in-progress, as I plan to take community suggestions to optimize how I generate this report. Before I get to the interesting trends, the Tier List, and conclusions we can take from it, let's break down the system I used to collect the data, possible outliers that could appear, and the raw data itself.

THE RULES:

This is only half of the data I would end up tallying...(excuse the poor handwriting)

Because I'm only one person, I had to implement a set of rules to effectively count Hero Picks without going insane. To cut down on time and hopefully still paint an accurate picture of the state of Competitive Overwatch, these are the rules I followed while tallying Hero Picks for both Offense and Defense during this weekend's tournaments:

At the beginning of the game, tally each Hero on each Team
At the end of the game, tally any Hero change that has occurred from the beginning of the Match
Do not count Obvious, opening cheeses (the only time this came into play was the 6-Hanzo Cheese on King's Row)
Do not count Duplicate Heroes
Do not count Heroes on King of the Hill Maps, because they have no delineation between Offense and Defense

Now some of you may notice, there are some obvious limitations and outliers that will occur from these rules. I will discuss them line-by-line:

At the beginning of the game, tally each Hero on each Team: This rule is obvious, and generally accurately portrays a team's overall strategy for the first leg of a Payload or Checkpoint Map. The only issue I ran into were obvious Cheese strats, but this was addressed in a later rule.

At the end of the game, tally any Hero change that has occurred from the beginning of the Match: This rule also seems obvious, and would capture each team's final Push, or Defending strategies. Occasionally, this would result in strange outliers as teams tried various desperation-Hero Swaps in order to tag the payload either on Offense or Defense, rarely with any success. Case in point, all but one of the Tracer picks on Offense, and all but one of the D.Va picks on Defense were unsuccessful, desperation-based Hero Swaps. Finally, by only tallying Hero picks at the beginning and end of games instead of cataloging every Hero-swap that occurred throughout the match, I miss out on second-leg or other mid-map strategic picks. However, I noted that teams generally didn't switch up their lineup too much in the middle of most maps (Gibraltar's Hangar section being the obvious exception), and I would have lost my mind trying to keep up with every Hero swap throughout a game.

Do not count Obvious, opening cheeses (the only time this came into play was the 6-Hanzo Cheese on King's Row): I decided that 6-Hero, single-action Cheese like the Hanzo Scatter Arrow strategy on King's Row were not worth counting, and this was the only such strategy that teams used anyway. I took initial counts for these teams only after they had all swapped off of Hanzo.

Do not count Duplicate Heroes: This was the rule I struggled with the most, primarily because EU players tended not to run duplicate Heroes and I didn't see it impacting the count as much anyway. Then Team EnVyUs happened. EnVyUs preferred to run a Double-Lucio, Double-Winston, Double-Reaper lineup and somehow word got out among the NA teams that this was some sort of Uber-Comp and many teams rushed to copy their strategy. By this point, I had collected about 75% of the data, and I decided to continue not counting duplicate Heroes. I may remove this restriction in the following weeks as these strategies continue. This means that the percentages you will see later indicate the percent out of all Picks a particular Hero accounted for, however this means that the percentages of Lucio, Winston, and Reaper (the most commonly doubled-up Heroes) Picks are a bit lower than actually practiced.

Do not count Heroes on King of the Hill Maps, because they have no delineation between Offense and Defense: This rule was primarily used because I wanted to have a concrete delineation between Offense and Defense to analyze different Hero choices on each side, and King of the Hill maps do not have that. This meant throwing out Hero tallies on King of the Hill maps, although the primary difference I noticed on King of the Hill maps was only an absence of Reinhardt picks.

THE DATA

Raw, just like I like it

The first thing you'll notice is the red highlighting: I added these to indicate where I saw significant* differences between the EU and NA Meta. I also converted Picks to Percentage of Total Picks to better compare the differences between EU and NA, as the total number of picks I tallied were not equal across regions. This data was collected from both GosuGamers weekly tournaments (NA and EU) on Saturday and Sunday, as well as the OneHitClub NA tourney which occurred on Saturday. I was able to collect data from every single GosuGamers EU tournament match thanks to their four channels simultaneously streaming the matches (twitch.tv/gosugamers, /gamersoriginow, /cleverpress, /esl_alphacast), so major thanks to them for doing so. I was only able to catch about half of the matches in OneHitClub and the GosuGamers NA tournament because only half of them were streamed, but the data was still meaningful. Finally, since I don't claim to be a great statistician and I know some of you may want to run "real" comparisons on the data, I've included a link to the Google sheet here.

*significant via eye-test, not statistically.

HERO TIERS

With 21 Heroes, I decided to split them into D,C,B,A and S tiers, with a single Hero filling the S spot. Here's how that looks:

S "You should always have this Hero in your lineup" Tier: LUCIO

This comes as surprise to no one, Lucio was far and away the most picked Hero on both Offense and Defense, accounting for a whopping 14.38% of all Hero Picks across all regions. Because I didn't count duplicates, his pick rate likely much higher than this. Lucio is the healer you should always have at least one of, and as EnVyUs showed us this weekend you often want two of. Is Le Balanced Disco Man too strong? Or are other Supports too weak?

A "It's a pretty good idea to include these Heroes in your lineup" Tier: ZENYATTA, REINHARDT, REAPER, MCCREE, WINSTON

Now that Zenyatta's shields may be healed, and due to the popularity of the Genji Zenyatta combo, Zenyatta was the second-most picked Hero in competitive scenarios. Reinhardt was this weekend's most likely Tank to be picked by a team, however this is likely because EU teams were not implementing EnVyUs's double Winston lineup yet. As much as the pros like to complain about Genji, it was actually Reaper who took the top spot among DPS Heroes, proving that sometimes you just have to press Q and spin to win. McCree appears this tier, primarily due to his popularity among EU's Hitscan-leaning players. This may change as the Double-Winston, Double-Reaper, Double-Lucio Meta spreads across the ocean, however. Speaking of this strategy, it is entirely the reason behind Winston's appearance in Tier A, as well as the title of this Report, as NA teams rushed to copy EnVyUs's Meta-Defining strategy.

B "The Specialist Heroes and NA favorites" Tier: GENJI, SOLDIER 76, ZARYA, PHARAH

As much as the Pros may like to complain about Genji, he actually falls into Tier B of Hero picks, although he had similar numbers as McCree and Reaper. This is mostly due to the disparity in how Genji is used: he's much more utilized on Offense than Defense, at a roughly 60:40 ratio. Soldier 76 marks the first significant drop-off in pick totals, and was similarly used much more on Defense than Offense. Zarya was almost never used as a Main Tank, but would often be swapped to mid-game in order to get a big team wipe against a tricky defense. Finally, Pharah falls to the B tier due to the majority of EU teams simply not using her anymore and some NA team's refusing to swap off their favorite rocket-firing Egyptian.

C "The Single Map Area Heroes, Panic-Picks, and Mercy" Tier: MERCY, WIDOWMAKER, SYMMETRA, D.VA, TRACER, ROADHOG

Mercy's pick rate stands as the next significant drop-off in the Tiers, and I believe her pick rate was artificially inflated by NA players' resistance to change from playing Heroes they're comfortable with. Currently, Zenyatta and Lucio seem to have obsoleted Mercy except in extremely dire situations where teams need a huge push into team resurrection to capture a difficult point. Widowmaker was used extremely effectively in very specific situations by very specific people: Coolmatt69 used her to great effect inside Watchpoint: Gibraltar's hangar, Twoeasy was a demon using her Offensively on Hollywood's final checkpoint, and Pluppie mounted an amazing defense on Hanamura's final checkpoint. Symmetra similarly was only ever utilized Defensively on the first checkpoint of map, and primarily by the EU region. D.Va and Tracer picks (as stated above) were mostly panic-picks to either defend a payload from capping or tap a pushing payload to preserve the Overtime timer. Finally, Roadhog was used by a couple of players like Twoeasy who believed they could hit enough hooks in specific maps to make a difference.

D "The I'm not sure these picks were intentional and rest in peace, Mei" Tier: HANZO, TORBJÖRN, JUNKRAT, BASTION, MEI

These Heroes just straight up stink. Outside of the Cheesy 6-Hanzo Scatter Arrow King's Row strategy, Hanzo only saw play on Defense by a few stubborn NA teams. The rest of these picks occurred so rarely that I'm pretty sure they were random, panic picks at the end of games. Finally, out of 1175 total picks, poor Mei wasn't picked a single time. RIP Balanced Ice Woman.

INTERESTING OUTCOMES, CONCLUSIONS, AND BALANCE IMPLICATIONS

Defense-Oriented Heroes Stink

The first conclusion I draw from this data is the simple statement: "The best Defense is a good Offense." Heroes supposedly valued for their Defensive specialty -- Bastion, Torbjörn, Mei for example -- were completely non-existent in the competitive Meta of Overwatch. This can be attributed to the value placed on Hero deaths as ways to accomplish the goals in Overwatch's various game modes. In the current state of Overwatch, there's no better way to gain an advantage over your opponent's team than getting a pick off one of their team -- giving your team a Hero advantage and often snowballing your assault or defense. A Hero like Mei whose damage isn't very high and whose abilities are based on controlling/altering map space only slows down an assault, it doesn't stop it. Bastion and Torbjörn -- at least at the competitive level -- seem to be too fixed in one spot and thus too easily countered by more mobile or more long range Heroes who can get behind them or snipe them from range. Ideally, you'd see Heroes of this classification be picked at least at the same rate as Symmetra -- who I think is in a great place -- making their mark in a meaningful way on certain parts of certain Maps. The problem then is how to balance this such that it doesn't overwhelm casual players (as we've already seen with Bastion and Torbjörn in the past), but that's for Blizzard to figure out, not me.

EU loves their Hitscan Heroes

There was a huge disparity in how often McCree and Soldier 76 were picked by EU players compared to NA. This can primarily be attributed to NA's adoption of EnVyUs' Double-Winston, Double-Reaper, Double-Lucio strategies which ate up slots that would otherwise be occupied by such Heroes, as well as NA's insistence on running Pharah in a Meta that seems sub-optimal for her. It is very interesting to see the two Regions' Metas shift in completely opposite ways: EU insists on playing Hitscan Heroes which makes Pharahs non-existent in their scene, while NA just turns into a Monkey on Monkey, Reaper on Reaper slugfest half of the time. We will have to see which strategy wins out next week.

The problem with Support and Tanks

There's a balance issue looming with Support and Tank classes and it is blatantly apparent from this data. The smaller the amount of total Heroes make up a "class type," the greater the disparity in picks among that class type will be as players min-max their team compositions. We can see this immediately in the Tank class in particular: teams tended to pick Reinhardt or Winston as their Main Tanks, use Zarya only for her game-changing Ultimate as a change of pace, and almost never picked Roadhog or D.Va for any reason. Support classes -- Healers in particular -- have seen a similar increase in pick disparity. By now most teams have realized that without a strong Pharah, Mercy has become less viable and Zenyatta and Lucio eagerly eaten up all of the team slots she would otherwise have occupied. Fortunately for Blizzard, I do have a good suggestion for how to combat this problem: add more Tanks and Support classes. Adding more classes will smooth out the Pick rates like we see with the DPS classes, and give people more options to swap to in different situations which support different Hero movesets.

FINAL THOUGHTS, AND A QUESTION TO MY READERS

This was a very interesting exercise to do, and something I want to continue doing in the future. Unfortunately, I'm going to be travelling the next four weekends and will have to do this entirely via VOD rather than as the matches happen, so the reports will be a little delayed. My question to you all is: Do you think I need to change anything about my methodology? I'm no statistician, but I would be glad to implement any ideas you might come up with, provided it's still time-efficient for me to do. Please let me know in my reddit post here.