In a previous article on quantiyfing spec strength in 2v2, we defined a metric that compared the specs based on how they do on the leaderboard compared to how they do on the general ladder. As a quick recap, if a spec accounts for 5% of all specs on the ladder, but accounts for 10% of all specs on the leaderboard, this spec is seen twice as often on the leaderboard as it is on the ladder. The relative representation of this spec would therefore be 2. If the spec only accounted for 2.5% of the leaderboard (while still accounting for 5% of the ladder), the relative representation would be 0.5, i.e. we see the spec half as often on the leaderboard as we do on the ladder.
For the 3v3 data, I have increased the sample size by 50% to get more power to do this analysis (to just over 60,000 characters, pseudorandomly sampled from EU and US realms), and I have also added errors bars which show 90% confidence intervals on the bars. The error bars capture the uncertainty that arises due to the fact that we're sampling from the ladder. If we were to repeat this analysis with a different set of randomly sampled characters, each sample would give slightly different results and the error bars capture this variation.
If we look at the relative representation of melee in 3v3, we find a slightly surprising spec at the top. Enhancement Shamans, while being a pretty poor performer in 2v2, are played far more frequently at high rating than we would expect from their overall representation alone. They are the least popular Shaman spec by some margin, which probably reflects that Enhancement has been a greatly underperforming spec for a long time. One possibility is that there may be an emerging trend of Enhancement Shamans in 3v3, which is starting with top players. If this were the case, then the relative representation could be temporarily inflated and exaggerating how good Enhancement Shamans are at the moment. Moving down the list, we find Subtlety Rogues in second place. Sub does not have a huge number of viable comps at the minute, having been quite heavily nerfed earlier in the season. Nevertheless, the primary comp they do have available to them - RMP(ala) - is arguably one of the stronger comps of all, which is presumably the reason we see rogues doing really well at the top of the ladder. Following on from this, we have Windwalker Monks, Feral Druids, and Arms Warriors. All are very strong specs across the board, and seem to be performing similarly. Windwalker is arguably the strongest melee in 2v2, and they are also doing very well in 3v3. Feral Druids may perhaps be the most surprising spec to make the top tier, but this performance is consistent across both 2v2 and 3v3. There is no evidence from either of these brackets that Feral Druids are particularly weak at the moment; instead they appear to just be rather unpopular.
Next on the list are Survival Hunters. A word of caution here, Survival Hunters have very low overall representation, so the confidence intervals on the relative representation are very large. The data we do have suggests that Survival Hunters may not be doing too badly, despite being a hugely unpopular spec. Retribution Paladins, Frost DKs, and Havoc DHs are all roughly performing similarly. Retribution Paladins are enjoying a high representation on the ladder, perhaps indicating that a lot of people are finding it enjoyable to play Ret. Frost DKs have low representation, but are presumably benefitting from the additional cleave potential in 3v3 and so are coming out in a decent shape in 3v3. Havoc DHs, while being in a pretty miserable state in 2v2, are not looking too terrible in 3v3. However, overall representation is very low for the only viable arena spec for Demon Hunters.
There are no surprises at the bottom of the melee list. Outlaw and Assassination Rogues, Unholy Death Knights, and Fury Warriors are all performing very poorly indeed. Unholy DKs are played the most out of all of these specs, but their representation on the leaderboard is exceedingly low given their high representation on the ladder.
Topping the ranged DPS, we have Fire Mages, Elemental Shamans, and Shadow Priests. These three specs are also the top performers in 2v2, and have a wide range of comps available to them in 3v3. They all also enjoy relatively high overall representation on the ladder, meaning they are not only popular, but also strong specs. Elemental Shaman is the least played spec, accounting for only 2.7% of the ladder and 3.8% of the leaderboard.
Balance Druids are also performing very well. While seemingly not quite as strong as the three top specs, they are far more popular than Elemental Shamans. A cynical person might suppose that Convoke one-shots have something to do with it. Both Destruction and Affliction are doing okay in 3v3, with Affliction being the far more popular spec. Out of the bottom four specs - Arcane Mage, MM Hunter, Demonology Warlock, and BM Hunter - only MM Hunters are played at any significant level, accounting for 5.5% of the ladder. There was not enough data on Frost Mages to say anything about their relative representation, so they have been omitted from the graph to avoid confusion.
The healer data for 3v3 show some differences from 2v2. Holy Paladins and Discipline Priests continue to reign supreme, accounting 9% and 11% of the leaderboard, respectively, while only accounting for 6% and 7% of the ladder, respectively. Restoration Shamans appear to be firmly a tier below, being both less popular (4% of the ladder) and relatively less well represented at the top of the ladder (4.4% of the leaderboard).
Note here that there is very little data on Holy Priests and Mistweaver Monks: they are both only played by about 0.8% of the ladder, but I have included them since the uncertainty about their relative representation isn't enormous. But do interpret their relative representation with appropriate caution; not a lot of people play these classes, so even small changes on the leaderboard can give large changes in the relative representation. At the bottom of the healer list, we find Restoration Druids. While we can't definitively say that Resto Druids are performing worse than Mistweaver Monks with this data, it is clear that both are doing very poorly in a bursty meta. Restoration Druids are a fairly popular spec (3% of characters on the ladder), but are seen incredibly rarely on the leaderboard (1.4% of characters) given how often they are played. Given that PvP is said to be balanced around 3v3, it may be high time to give Mistweaver Monks and Restoration Druids some sorely needed attention.