its that time again...top 100 raidbots data

10 Blood Elf Paladin
10
http://raidbots.com/dpsbot/Overall_DPS/25N/100/14/60/default/

interesting things of note:

-fire mages have caught up to assassination rogues

-destro outperforming affliction, guess nerfing sac into the ground did something

-melee: rogues/warriors/dks doing well, feral/ret/ww/enhance doing poorly

-best dps: 170k, worst: 120k...or about a 30% gap

this seems to suggest a few things. 1) scaling is way off, as some specs are doing far better than others with 522+ gear. 2) the gap between specs is much larger than 5%, and 3) dps shaman, hunters (bm/mm), rets, monks, and ferals need some love.

last tier, affliction/arcane were well above everyone else. it seems like the gap is starting to widen even more, with regard to dps overall.

is the rapid increase in ilvl to blame? or specifically, is scaling out of control? some specs seem to improve rather rapidly (ie: fire), while others cant make nearly as big a jump. I can only imagine how large the gaps will be in t16 when we hit 600 ilvl.
Edited by Sanctifìed on 4/4/2013 12:31 AM PDT
Reply Quote
90 Undead Warlock
13160
Okay, I am going Marks on my Hunter now since it is so underrepresented.
Reply Quote
10 Blood Elf Paladin
10
Why would you use top100?

No education when it comes to statistics I take it?


all parses isnt an ideal set of data: top 100 basically represents each spec pushed to its limit, or max potential. these are cases where individual player skill isnt in question. on all parses, a ret paladin might be doing better than a fire mage, but that data isnt really accurate, as the mage should be doing more dps assuming equal player ability.

overall averages include the best and the worst of each spec. the top 100 parses are much more telling.
Reply Quote
90 Blood Elf Death Knight
5445
Why would you use top100?

No education when it comes to statistics I take it?


all parses isnt an ideal set of data: top 100 basically represents each spec pushed to its limit, or max potential. these are cases where individual player skill isnt in question. on all parses, a ret paladin might be doing better than a fire mage, but that data isnt really accurate, as the mage should be doing more dps assuming equal player ability.

overall averages include the best and the worst of each spec. the top 100 parses are much more telling.


Um no the fudge it doesn't. Top 100 basically tells you who does the most when they get lucky procs. hell the ppl in top 100 cannot even recreate those numbers. There is a reason the devs discredit top 100.

P.S. you know he's actually right when people do statistics they tale all the data not just the top 100.
Edited by Fayte on 4/4/2013 1:00 AM PDT
Reply Quote
10 Blood Elf Paladin
10
on all parses, enhance is 3k dps behind fury. ret is 5k behind. (and the worst melee, /cry)

this is why that data is flawed. we know that isnt the case. a geared fury warrior will stomp both into the ground on the meters.
Edited by Sanctifìed on 4/4/2013 1:08 AM PDT
Reply Quote
6 Night Elf Hunter
0
http://raidbots.com/dpsbot/Overall_DPS/25N/all/14/30/p90/

More accurate, statistically speaking.
Reply Quote
90 Blood Elf Death Knight
5445
on all parses, enhance is 3k dps behind fury. ret is 5k behind. (and the worst melee, /cry)

this is why that data is flawed. we know that isnt the case. a geared fury warrior will stomp both into the ground on the meters.


top 100 is the flawed one for reasons i already mentioned. in order for your statement to be true you assume that more than 50% of the dps that are posting their results on WoL are bad and you are also assuming RNG plays no part in dps.

heres the funniest thing about top 100, click and enable all tanks. Why is a prot pally pretty high and stopping ret in dps while in All parses the tanks are at the bottom.
Edited by Fayte on 4/4/2013 1:33 AM PDT
Reply Quote
90 Dwarf Hunter
16860
Why is it that every time I go to that site looking for hunters I have to scroll my eyes down so far.
Reply Quote
90 Worgen Warrior
11370
Why is it that every time I go to that site looking for hunters I have to scroll my eyes down so far.


Because you click yourself at night.
Edited by Bladeletters on 4/4/2013 2:09 AM PDT
Reply Quote
90 Blood Elf Mage
6075
Why would you use top100?

No education when it comes to statistics I take it?


all parses isnt an ideal set of data: top 100 basically represents each spec pushed to its limit, or max potential. these are cases where individual player skill isnt in question. on all parses, a ret paladin might be doing better than a fire mage, but that data isnt really accurate, as the mage should be doing more dps assuming equal player ability.

overall averages include the best and the worst of each spec. the top 100 parses are much more telling.


False.
Top 100 tends to favor specs with higher potential variance.
Top 100 is also biased by the number of parses used, as 100 is a raw number having less of a spec generally leads to less accurate top 100 data.
See 5.0 arcane on ToES fights, where there were under 100 parses on normal modes.
Reply Quote
90 Undead Warlock
13160
False.
Top 100 tends to favor specs with higher potential variance.


this makes sense, as the anomalies can sneak up there even if 99% of the spec is performing below.

Like Demo simming Godly on Counsel fights where 100% Crit Doom via trinket is spread on all targets with Glpyh of Everlasting Affliction.

I would think looking at the top 3000 or so would be more accurate.
Reply Quote
MVP - World of Warcraft
90 Human Mage
10015

I would think looking at the top 3000 or so would be more accurate.


Top anything is inaccurate. If you picked top 3000, you'd be comparing the top 10% of a spec with 30,000 parses (including all its top outliers, but excluding all its bottom outliers) against the top 50% of a spec with 6000 parses (same outliers problem) against all of a spec with 2500 parses (including all outliers at both ends).

How do you factor out gear differentials? You can't assume there's an equal distribution of gear levels across three such different samples. You have no was of knowing what the distributions are. Ditto with skill.

It's possible, but not simple, to extract useful information from Raidbots about some, but not all, specs. It isn't really possible to extract useful information from a Top X sample.
Reply Quote
90 Orc Warlock
13715
Why would you use top100?

No education when it comes to statistics I take it?


Neither top 100 or all parses are good samples.

Top 100 includes RNG, as well as things like people going out of their way to pad meters, or their guilds feeding them high DPS.

All parses include people with unenchanted/ungemmed gear and those playing horribly wrong, as well as those who died during the fight. This results in easier to play specs appearing higher than they actually are.
Reply Quote
90 Blood Elf Rogue
10525
i find it funny, everyone has mentioned how top 100 is bad. yet the OP is still defending it. they're right, top 100 is terrible. i once did 180k on Darkmaster Ganding, another time i did 140k. does that mean i did worse the second time? nope, you know what it was? the fight lasted longer the second time so my dps ramped down. the first fight was quick, so my burst was still high.....a factor which you also didnt take into consideration. please dont post top 100 parses thinking it shows what the best is capable of, these people just got lucky. lucky != best
Reply Quote
90 Human Warlock
17000
this is why that data is flawed. we know that isnt the case. a geared fury warrior will stomp both into the ground on the meters.

One of the key rules of statistics, and science in general, is to look at the data and then form a conclusion. You don't start with a conclusion and look for data that supports it.
Reply Quote
04/04/2013 07:41 AMPosted by Serinicas
this is why that data is flawed. we know that isnt the case. a geared fury warrior will stomp both into the ground on the meters.

One of the key rules of statistics, and science in general, is to look at the data and then form a conclusion. You don't start with a conclusion and look for data that supports it.


This rule is never followed by people on here. Their method is that they know that 1. Mages should be nerfed, and 2. Their class should be buffed, then they find the data to support that and then justify why all the other data is false and the one link they posted is the one that counts.
Reply Quote
10 Blood Elf Paladin
10


One of the key rules of statistics, and science in general, is to look at the data and then form a conclusion. You don't start with a conclusion and look for data that supports it.


This rule is never followed by people on here. Their method is that they know that 1. Mages should be nerfed, and 2. Their class should be buffed, then they find the data to support that and then justify why all the other data is false and the one link they posted is the one that counts.


dont get me wrong, mages shouldnt be nerfed, in fact I expected fire to become really strong, as it always has been, at higher gear levels (like in ICC for example). specs like fire/fury scale really well with gear. that doesnt mean mages/warriors need a nerf (they dont), but it appears that a lot of specs arent scaling nearly as well.

I know devs dont care much for the "scaling" word, but it does seem to be an issue. who would have guessed destruction would outpace affliction in 1 raid tier, or fire over arcane, for example?

then again, ilvl has inflated rather rapidly, so I guess some of this data isnt surprising, even if it doesnt tell the whole story.
Edited by Sanctifìed on 4/4/2013 8:17 AM PDT
Reply Quote

Please report any Code of Conduct violations, including:

Threats of violence. We take these seriously and will alert the proper authorities.

Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.

Harassing or discriminatory language. This will not be tolerated.

Forums Code of Conduct

Report Post # written by

Reason
Explain (256 characters max)
Submit Cancel

Reported!

[Close]