Aftermath of the January 20 balance patch

[Ryme]

Disclaimer: All data taken from all MMR levels to generate a baseline, then higher ranks examined to fill in data and only data sent to hotslogs is used.

So, the new aftermath thread - this one's a little late, but I've been outrageously busy at work and, hey, let's just claim I did it for data integrity reasons.

Largest Movers



Data:

Movers	Global	Diamond	Master
Trande	-4.8	-6.5	-8.1
Raynor	-2.7	-4.3	-5.5
Lunara	+6.6	+8.1	+3.2
Brightwing	+5.9	+8.1	+20.3
Greymane	+5.1	+4.8	+3.1
Malfurion	+4.1	+3.9	+7.1
Stitches	+3.6	+2.2	+6.5
Cho'Gall	+2.9	+2.4	+2.2

Highest/Lowest win rate

The winners:

Diablo is still the king of hero league, sitting at 56.5% global win rate (D: 58.6%, M: 59.1%)
The Lost Vikings haven't lost their touch, sitting at 55.4% global win rate (D: 59.8%, M: 60.4%)
Kael'thas continues to be strong, sitting at 54.7% global win rate (D: 56.2%, M: 55.7%)

All three are at that nasty 55% threshold. We already know there are KT changes coming, I expect dibbles nerfs on the horizon as well.

On the side of the spectrum:

Abathur is consistently the weakest hero league character, sitting at 42.2% global win rate (D: 45.1%, M: 47%)
Illidan is consistently the second weakest, 42.9% global win rate (D: 46.4%, M: 50.9%)
Zeratul still hasn't recovered from his nerfs, 44.1% global win rate (D: 45.9%, M: 44.8%)

Abathur and Illidan are probably ok, their high skill cap is reflected in them doing a lot better in higher MMR brackets. Zeratul is bad in all levels of play, I hope he gets some love soon.

Talent changes for the top movers

Tyrande (popularity in hero league down 15.9%)

• 1 - Popularity for the nerfed Ranger's Mark is down 16.5% with all other talents soaring up ~30-40%. Despite this, all talents except Seasoned Marksmen are down 9-10%. All talents this tier are now around 53% WR.
• 4 - Pierce has soared in popularity by nearly 20%, focused attack and protective shield are down 15% and 11% respectively. Healing Ward and Protective Shield still jostle for position of highest WR, both sitting happy at around 54%.
• 7 - Dramatic uptake in Quickening Blessing by nearly 22%, Mule is up 15% too. No significant changes to WR, Battle Momentum continues to be top dog.
• 10 - Starfall up 37% as people begin to see if it has a home again. Sadly, the gulf between heroic WRs is largely the same falling from 10% to 9%.
• 13 - No significant changes, Overflowing Light is still the most popular and highest winning talent, other talents are still largely ignored.
• 16 - Ranger is up 28%, suggesting with the pierce uptake that many more players are going for the epic Owl build. Trueshot Aura is still the highest WR talent here, oddly Mark of Mending is down significantly but this might be due to data error since it's rarely ever taken.
• 20 - Celestial Wrath is up 38% to match the level 10, however no other significant changes have occured for Tyrande here. All talents feel balanced to each other at around 60-65% WR.

Raynor (popularity in hero league down 5%)

• 1 - No diversity, popularity is almost identical here to last week, Raiders' "Recruitment" is down in WR by about 8.6% but this is likely just variance.
• 4 - Same issue as above, with Confident Aim the only mover going down 5.9%.
• 7 - Nothing significant has changed, Revolution Overdrive is still the most picked and won with.
• 10 - Nothing significant changed. Hyperion is still the most picked and won with.
• 13 - Double-Barreled is down 11% following the nerf to Raynor's 16 stun talent. Steel Resolve climbs to be very nearly on par with Giant Killer for top talent this tier.
• 16 - Both Berserk and Executioner climb in popularity dramatically, 21% and 18.5% respectively, Bullseye is down is 7.5% following the slight nerf on the stun. Interestingly, very little has changed here for win rates and executioner remains the top talent, it just seems people are now realising how good it was in the CC meta.
• 20 - Everyone still takes Nexus Frenzy and Nexus Frenzy is still the top winning talent.

Honestly, Raynor is desperate for talent diversity, this is the most unchanging list of any meta hero. Most of the loss is likely attributable to player perception.

Lunara (popularity in hero league up 3.5%)

• 1 - Photosynthesis is down nearly 22% in popularity and gains no significant WR changes. Other talents this tier are all up about 20% in win rate with Cruel Spores remaining at the top.
• 4 - Nimble Wisp continues to climb in popularity despite being statistically inferior to Timelost Wisp which is the only breakaway talent, holding nearly 1% over the other choices this tier.
• 7 - It seems people have finally caught on to Nature's Culling. It boasts a 33% improved popularity, and 33% improved WR putting it as the clear top talent this tier. (Seriously, try it guys, it's so broken now)
• 10 - Thornwood climbs 20% in popularity over Leaping Strike. Both ults have climbed dramatically and currently Thornwood holds slightly more than 1% improved WR over Leaping.
• 13 - All talents popularity are down except Unfair Advantage which appears to be soaking it all up. All win rates are up this tier, Pestering Blossom and Giant Killer have climbed the most at 30% and 23% respectively with Giant Killer currently being the top talent this tier.
• 16 - All talents popularity are down except Star Wood Spear which is up 21%. Choking Pollen has gained the most in WR with 26% improvement. However, Let Them Wither and Invigorating Spores continue to jostle for position as top talent this tier.
• 20 - Abolish Magic is the only talent to improve in popularity with a 9% increase, no significant changes to win rate on any talent.

Brightwing (popularity in hero league up 1.9%)

• 1 - Bribe is up 29% in popularity and 20% in WR. Unfurling Spray and Hyper Shift continue to jostle for top position, however.
• 4 - Arcane Barrage is up 242% in popularity and boasts a near 30% increase to WR pushing into an intensely balanced tier that has a total range of less than 1%.
• 7 - Dream Shot and Cleanse are up 214% and 137% in popularity respectively. However, Mistified climbs 19% in WR and is currently the best choice here statistically.
• 10 - No significant changes here, Emerald Wind remains niche but not terrible.
• 13 - Sticky Flare is the interesting one here, it's climbed 56% in popularity and 22% in WR. It's now a mere 1% behind the top choice of Shield Dust.
• 16 - Hardened Focus is the only real change here, it's up 8.5% in popularity and an impressive 21% in WR. It's now 1.1% behind the top choice Bouncy Dust.
• 20 - Storm Shield is the only thing to gain any ground here, though is still statistically inferior to Double Wyrmhole/Rewind.

[Deleted]

I really like data instead of anecdote based balance discussion; big thank you for this post. One caveat worth mentioning is that the Masters level data for Anubarak, Azmodan, Li-Li (all above 55% win rate at the moment of this writing) and Brightwing (top mover on your list) is too thin to be reliable yet. Even TLV being number one with Masters data based on less than 500 games comes with a big margin of error.

Overall I feel the balancing patch was a huge step in the right direction, looking both at the hero win rates and at the talent choice diversity of the heroes who underwent the changes. Another hero worth a look currently is probably Malf, who got a nice boost ofterall and for whom an offensive talent choice seems to be suprisingly viable. Strangling Vines @lv7 and Tenacious Roots @lv16 are the choices with the highest WR and Full Moonfire @lv13 is in the second spot very close to Life Seed.

The heroes to watch in the near future will most likely be Nova and Rehgar with extensive revamps and Greymane with additional buffs and a playerbase still learning too use him.

[apepi]

So people remember what was actually changed, here you go.

How did certain hero nerfs hurt those who combo well with other healers? Were they hurt more or worst than the ones that recieved the nerf? Like Tyr and Diablo.

[Mormolyce]

I tend to think that global win rates should be taken with a grain of salt. Also this is based only on data from people who use the app or people in games with them?

[Nerraw]

I tend to think that global win rates should be taken with a grain of salt. Also this is based only on data from people who use the app or people in games with them?

It includes everyone who has been in games with people uploading replays.

[Ryme]

I tend to think that global win rates should be taken with a grain of salt.

I agree, I did actually mean to put a 'get your salt shakers ready' disclaimer for the talent section but I must have forgotten. I would love to analyse the data for higher specific ranks and I do try to do this for the other sections. However, the talent section is a fairly long write up and I couldn't really justify expanding it for other ranks like I did the former.

In the end, I had to make a choice and I went with global. This is because Master league players rarely deviate from accepted norms, when they do the sample size is often too small to draw many conclusions from, we have a similar if less pronounced situation for Diamond.

Thankfully, the owner of hotslogs has said that things like Diamond + (i.e. Diamond and master) will be coming along shortly so I hope to be able to make use of this going forward for other write ups.

[Mormolyce]

Obviously, since Blizzard doesn't make complete match data public this is the best data we have available, just saying we should bear in mind potential biases caused by the sample. And as you say, the lower the skill level the more data you have, but on the other hand the less valuable the data because it's really the high levels of play where people get the most out of heroes. But there you're looking at a small highly abnormal group of people. So it's give and take.

Unfortunately on forums some people just grab the headline and run screaming with it without properly considering all the caveats :P

[ellieg]

One thing ill say about aba and illi is people get upset and don't want to play if you pick them before the game even starts. I don't remember my exact team set up but it had an ETC, aba, and me on a falstad. Someone picks Aba and the ETC starts spergin the fuck out about how its going to be a loss no matter what because its "4v5" every time now. He just keeps bitchin and refuses to help. Hes soloing lanes instead of team fighting and says "it doesn't matter, we are going to lose anyway." Of course we lost and I know hes gonna be sitting there smug as fuck "I told you so." So all im saying is I like these stats, but I think people rely on them too much and use it as an excuse to not play well.

[Tharkkun]

Obviously, since Blizzard doesn't make complete match data public this is the best data we have available, just saying we should bear in mind potential biases caused by the sample. And as you say, the lower the skill level the more data you have, but on the other hand the less valuable the data because it's really the high levels of play where people get the most out of heroes. But there you're looking at a small highly abnormal group of people. So it's give and take.

Unfortunately on forums some people just grab the headline and run screaming with it without properly considering all the caveats :P

Nailed it.

This is good stuff even if it's not 100% accurate. It's kinda neat to see people's tendencies.

[BananaHandsB]

I tend to think that global win rates should be taken with a grain of salt. Also this is based only on data from people who use the app or people in games with them?

Of course, but since it's essentially the only data available, you gotta work with what you got. It's certainly more meaningful than wild speculation.