Home, Hearth & Heroes
My Builds My Blog My Tiers
Impact of the Third Ban in HGC 2018

View my last HGC post here.
View all of my previous blog posts here.

It's now been several weeks since the third ban took the Nexus by storm.  While the first three weeks of Phase 2 didn't have third ban, the professional scene did have a third ban for the two weeks that followed, as well as both Clashes.  As such, now is probably a great time to look at the impact of the third ban on the overall meta of the HGC.

I've split the data I've gathered for Phase 2 into two parts.  The first data collection contains North American, European, and Korean games from Weeks 1-3 (189 games total), which should give us an idea of how the pros handled two bans.  The second set of data contains all of the North American, European, and Korean games from Weeks 4-5, as well as both Clashes (200 games total), so it is roughly the same sample size and should help give a diverse view of how much impact the third ban wave had.

Hopefully, this article will cover absolutely everything that you could want to know about the third ban's impact on professional play!


Hero Popularity

Let's start out by looking at how the third ban impacted general hero popularity.  Did it yield greater levels of diversity, as we had hoped?  Well, sort of.  Here's a table showing the number of unique hero selections made during the first three weeks.

And here's that same table, looking at the hero pools during Weeks 4-5, plus the data gathered from both Clashes.

As you can see from the tables, the overall number of heroes in the pool has increased....however, this isn't necessarily a very meaningful statistic on its own.  When we look at the context of the stats in question, we can quickly find out why these numbers have changed.  The third-ban patch released alongside the changes to Azmodan and Raynor, who had not seen play prior to this point.  They account for 2 of the new heroes on this list.  Additionally, Arthas and Auriel are only counted on that overall list due to being picked/banned by China during the Eastern Clash.  Lastly, two new additions to the list of picked heroes are Gazlowe and Kael'thas.  The latter received a slight buff in the 35.2 patch, which may account for his sudden appearance in Clash play, and the former....might not have been the most serious of picks to begin with.

When you account for these individual pieces of data together, they make up the entire difference between the first and second tables in terms of overall popularity figures (at least, at first glance....but more on that in a moment).  The number of unique picks in NA, EU, and KR decreased slightly, while the overall popularity remained relatively constant.  The ban count, however, did increase slightly in all three regions, suggesting that at least the additional ban is having a direct impact in the area you'd expect it to: there are slightly more unique hero bans than there used to be.  This shift can also be seen a bit more easily when looking at how much the draft slots have changed via the table below.

There were a total of 23 unique hero bans during the first pair of bans in Weeks 1-3, but with the addition of a second pair of bans, that number has ballooned to 35 unique hero bans.  The impact on the draft of such a massive increase in bans is fairly obvious when you look at the rest of the numbers.  A third ban in the early part of the draft typically means more generalist heroes banned out, and more room to try and choke a specific player or role early in the draft.  As such, many teams resort to scooping up the handful of generalists that remain in the pool (and still cover these choked-out roles) much earlier, leading to a smaller, more condensed pool of hero picks in the first half of the draft.  This then leads to marginal improvements in the diversity of the back-half of the draft in response to those earlier chokes.  While the margin is fairly slim, it is nonetheless there and has made at least a minor impact in overall diversity.


Hero Ban Rates

Of course, just looking at broad-spectrum data will only tell us so much.  It doesn't show us the details of how each hero has changed.  So let's look at that next.  This table separates each hero out into categories depending on their ban rates.  Let's see how it looked like during Weeks 1-3.

And here's that same table, looking at the data from Week 4 and beyond.

Now that we are looking at the data in a different light, we can see a lot more individual points of change, and the impact of the third ban becomes a lot clearer than before.  The largest impact of the third ban is towards the center of the chart, with heroes that were originally somewhat popular in the meta to begin with, but the lack of an additional ban made them a bit of a lesser priority.  Generalist assassin heroes such as Fenix and Hanzo were always fairly common picks, but the decline of Tracer and the addition of third ban allows them to be somewhat more meaningful in the draft.  You can also see this in the tank role, where Diablo has increased in ban priority in part due to the presence of a third ban in draft.

You can also see some of the third ban's influence on meta shifts.  For example, when Maiev quickly rose back up to prominence during the Clashes, her ban rate skyrocketed far more than her pick rate did.  The presence of a third ban gave teams a chance to ban her out freely and still have an additional "spare" ban for more specific draft concerns.  We can also see that specialized picks like Sgt Hammer and Medivh, who have been shining more on the international tournament stage, have both seen upticks in their ban rate as a result of having an extra ban available.


Individual Hero Assessments

Now that we've looked at some relatively generalized data on the subject of picks and bans, I'd like to look at individual performances and see which heroes have seen statistically significant shifts in their ban rate.  This will make it easier to tell what sort of impact the third ban might be having.  In order to make as much of this information available to my readers, I created a separate spreadsheet to hold this data, which you are welcome to view here.  I will not be going over every single detail in that spreadsheet, as that would expand this article's length far more than I care to, so feel free to review my work for yourself!

Of course, when looking at this sort of data, it's important to remove changes that might not be meaningful.  So I've done a simple 95% confidence interval calculation to help remove ban rate changes that are less likely to be meaningful, and compiled my findings in a pair of tables that show every single statistically significant change in ban rate between Weeks 1-3 and Weeks 4+.  And here they are:

Now it is worth reminding you that just because a change is statistically significant, it does not mean that the change was directly caused by the third ban.  Many of these changes are the result of more mundane elements, such as buffs/nerfs, overall meta shifts, and plain old experimentation.  We can remove a number of names from this list based on those things.  Raynor's sudden appearance on the list is the result of reworks that made him into a meta pick.  Garrosh went up as Johanna's priority went down, indicating an obvious meta shift.  Similarly, Malfurion received a nerf that removed his Ice Block, pushing Deckard Cain further up the ladder in pick rate.  Tracer lost some of her damage, decreasing her emphasis in pro play and likely increasing the stock of other assassins.  And so on.

However, as you may have noticed, the number of statistically significant increases in ban rate is a much, much longer list than the number of decreases.  This helps to really showcase just how much things have changed with the presence of the third ban.  Ordinarily, a single hero's increase will lead to another's decrease, such as the relationship between Garrosh and Johanna.  But there are far more increases than decreases because of the fact that two more heroes are getting banned each game.  This is relatively obvious, of course, but what's less obvious (at least, when you don't have the data in front of you) are the benefactors.

Generalists like Yrel, Diablo, or Fenix were already fairly meta, but didn't usually draw an early ban.  Now these heroes are drawing many more bans than they did before.  We can see a similar effect on "tournament heroes" like Maiev, Sgt Hammer, and Medivh, who were also pretty ordinary meta, albeit mostly for the international stage.  But beyond that, some heroes that almost never got banned, such as Alexstrasza, Zeratul, and Thrall, saw spikes in their ban rates due to the handiness of having an additional ban to deal with more specific brands of heroes that work well with a specific comp or map.


The Final Word

The third ban has not created a massive upheaval in terms of how many heroes are picked or banned.  However, its effects on the meta have been significant.  A significant number of highly "meta" generalist heroes are now more frequently banned than usual, allowing some of the less popular "off-meta" picks a chance to shine.  There are two distinct "styles" that the third ban has been used for:

1) to eliminate additional highly-desirable meta heroes
2) to further restrict a team's choices via a traditional "role choke" or "player choke"

While it is impossible to determine exactly how much of these ban rate increases are due to one or the other, it has enabled a greater degree of strategy that much of the pro scene seems to be in favor of.  In any case, it is good to see both the professional and casual scenes responding favorably to this additional ban, and I think its overall impact has been significantly positive.  Let's just hope that we don't need to wait 3 more years for a fourth ban!

Looking for more of my work?  I previously did a piece much like this one, showcasing Raynor's use in the HGC.  That article is located here.  Feel free to let me know your thoughts to this article in the comments section below, or on Twitter at my handle @CriticKitten.  I look forward to your replies!

Hope to see you again next time!
There are no comments for this post.
Thanks for putting this together. It's about what I expected, high impact on draft strategy, low impact on what heroes we actually see played.