Statistics don’t work like that, though.
20% error much? You can reach that by just “slight” error in picking your sample. Large disparity can and will happen in statistics when your method of gathering is incorrect.
Heroesprofile is not a statistics company. It generates stats through compiling replay data. This is fine but not a professional approach. Hotslogs reached rough parity through sheer volume of data but to reach this critical number you need tons of replays, almost approaching the complete count of matches for a set time period. Proper statistics works by analysis to pick a set amount of memebers to represent the entirety of the playerbase. This is done carefully and approaches decent numbers by this approach. If you don’t do this you can only achieve accurate numbers if you literally gather data from all matches.
Henceforth, heroesprofile’s statistics have to be wildly inaccurate. In addition you ignore the fact that answers regarding to said stat differences were already provided. So your argument is shot dead on this matter.
This matter is not about whether Deathwing is OP or not. It’s about the validity of the source.
2 Likes
Actually, they do. The sample is more than large enough for general trends. Especially when you got a hero that was played that much EVEN when he got banned that much, you still had a ton of data relative to other heroes. If you’re looking at just master tier data where there are too few games, you might have a stronger point.
That’s actually not how statistics or sampling work. You can always get a sample, how good the sample is, will always be subject to accuracy, thus measurements of margin of error, because that’s the whole point of a sample. Which is why the arguments against the purity of stats is rather shallow. It’s this all or nothing approach. Some data might not be complete, so that means all of it is worthless. It’s just a silly way to look at it, especially when the data skewed that bad. Your claim that heroesprofile was going to get a 20% margin of error is just being pointlessly argumentative.
Huh? No.
They’ve hatched, you were wrong, the stats corroborated the prediction as they have multiple times in the past. But feel free to make those points.
4 Likes
Ah good, then that means that Deathwing has already been adjusted and you can stop posting about it.
No. That’s not how it works. Before you can come to any conclusion you need to address these questions:
- How is the result calculated?
- What are the sources?
- Who are the participants?
- How does it account for bias?
- Have the results been duplicated by a third party?
- What is the context in which the data was collected?
- Have the terms been defined?
1 Like
And we are back to arguing two separate arguments.
Let me make 2 things clear:
- Deathwing is indeed overpowered and getting more nerfs. (Also slight buff to worldbreaker).
- HeroesProfile is not perfectly reliable.
There things don’t really have that much to do with each other. Because what’s that phrase again? There’s lies, damned lies, and then there’s statistics. Sure heroes profile grants a wide picture, but it is not perfectly accurate.
1 Like
Well, this needs a context here right? DW is getting buffs on the thing everyone said he needed, his ranged form. And he is getting nerfs on everything people said he needed, his melee talents. So yeah, people were right.
2 Likes
Yay worldbreaker form potentially viable!
1 Like
I couldn’t be more happy! Melee is fun to play, but I wish ranged was somewhat usable.
And when DW was so fat and couldn’t even fit in the picture, people still had the audacity to say, well you know, maybe he isn’t because it’s your fault.
1 Like
Feel free to answer those questions.
Zenas is not arguing about deathwings power level. Initial post was sarcasm. This has been gone over. Zenas is trying to point out flaws in arguments and has never said Deathwing is fine. He’s arguing about how people rely on heroes profile too much and ignore common sense and the facts that prove them wrong.
No I get that, but he only ever really uses it in service of whenever people try to point to the data. The problem, he’s almost always wrong about it lol, which is why it’s kind of sad.
2 Likes
To be fair, heroesprofile and the old hotslogs always had enough games uploaded to it to represent something close to what blizzard has =p
2 Likes
Agree. Just people have this wild tendency to only post the stats that support their argument and ignore all others. HeroesProfile might be accurate and unbiased, but people aren’t. And not all statistics are relevant at all times. Basically people’s statistical analysis is not great.
I guess I’m just tired of people saying things like “Deathwing can’t 1v1, he’s fine,” or “Just draft % damage, easy win”. Things aren’t quite that simple. 
2 Likes
Oh sure there is a lot of people that only base their opinion on hotslogs/heroes profile. That’s why I look at everything possible before having mine with third part websites being just 1 of multiple evidences.
2 Likes
Deathwing’s win rate is going down because people still seem to think that World Breaker and Q are worth using.
I cringe every time I see someone switch forms. All you have to do to win with Deathwing is basically run it down mid with Destroyer build. I’m like 20-1 with that.