Wednesday, January 26, 2011

Raid Tools: CompareBot

Over the years, I've gotten a lot of emails asking for help. People will send me their parses and ask why Moonkin X is doing better DPS then they are even though Moonkin X is similarly or lesser geared. I want to help these people, but the truth is most of the time it was difficult to find anything significant to point out.

I think most people will agree with me that WoL parses are not easy to interpret. It's not hard to navigate your way around a log, but it can be difficult to pull all of the relevant pieces together to get a fully developed picture of the situation. Tearing a parse a part to find the detail can take an hour or more. As a result most people only go for the low hanging fruit and as a result begin and end their analysis on the Damage Done tab. In the end they get very few answers because they are comparing apples to oranges or don't go deep enough to see what is driving those numbers.

In this post I want to introduce you to a tool that can help you analyze your parses and hopefully give you more information on how to improve your performance. That tool is CompareBot.

How to use CompareBot?

CompareBot is a tool that will summarize up to three parses and present the information in a way that is easy to compare and find discrepancies. The web site provides fairly simple and straight forward instructions on how to make the page work, but there are several different strategies you could use to analyze you performance.

Week to Week Analysis: You can compare yourself to yourself from week to week or from attempt to attempt. This could allow you to evaluate several different DPS strategies, or possibly identify why one attempt went really well while others did not.

The week to week analysis is nice because it limits or eliminates the impact of gear and strategy differences. The downside to this type of analysis is that it's limited by your own ability and imagination. If there is a fundamental flaw in your performance, you are unlikely to identify it by comparing yourself to just yourself.

Internal Peer Analysis: If your guild raids with multiple moonkin comparing yourself to the other moonkin in the raid can be very helpful, especially if they are performing better then you. Internal Peer Analysis is great because it limits the impact of strategy differences and other variables like fight length that can skew the results when comparing different attempts.

Internal Peer Analysis is limited by the ability and dedication of your guild mates and depends on your guild running multiple moonkin. If you're the best moonkin in your guild it could help you identify what they are doing wrong, but could do very little to help you.

External Peer Analysis: The great thing about External Peer Analysis is that the possibilities for review are endless. As long as the moonkin has a parse on WoL you can see how you stack up to them. You can compare yourself to a rival in another guild, or a well known moonkin like myself or Murmurs. Maybe you want to see what makes Lappè so good at Paragon, or find out why a particular moonkin has the number one ranking on WoL for Cho'gall.

This can be very helpful in identifying why the greats are great and where you are different, but you also have to be a little weary of the results. There are a lot of variables that can impact performance that are not easy to identify in a WoL parse. Strategy differences between guilds can have a big impact that won't show up in WoL gear differences can also be very difficult to quantify. The quality of other players in a raid you can have a significant impact on a player's
performance.

What should a Moonkin look for:

In this Link I compare myself to another moonkin in the raid. At first glance Rootcanal appears to be awesome, and I appear to suck. How else would he beat me by 3k DPS? Honestly, that was my first reaction when I saw this comparison. So I dove into it to see where the differences where.

  1. Hits per Minute: The number one rule of DPS is to Always Be Casting, because the number one driver of DPS is how much you cast. Therefore, the first thing I would look at is the Hits per Minute of each spell. Looking at the total number of casts can be misleading because it's skewed by fight length, but Hits per Minute is a pretty good indication if someone has excessive amounts of down time between casts.

    Ticks per Minute is another good thing to look at for DoTs. Though you should be careful when comparing Ticks per Minute for a moonkin. How you utilize Nature's Grace could have an impact on the numbers you see.

    If you look at the example, Rootcanal and I compare fairly well. He beats me slightly on Wrath and Insect Swarm. There are also very slight differences on Starfire and Starsurge. However, I do a little better with the Moonfire/Sunfire combo spell. I am a little bit behind here, but the difference isn't huge and I doubt it's responsible for the 3k DPS difference.

  2. Eclipse: As we all know Eclipse is drives moonkin DPS, therefore it is the next stop on my comparison. First off, ignore the uptime stat. Having a higher uptime isn't necessarily good, because you could have the buff up but casting a spell that it does not affect. What I would look at is the number of procs per minute. This number could be a little skewed by Euphoria procs, but it's the best indication we have of how well we transition from one Eclipse buff to the other.

    On the Buff tab of the comparison you can see that Rootcanal procced eclipse 9 times while I procced it just 8. This suggests that he may have been a little more efficient in his Eclipse Transitions then I was, but it is ever close.

  3. Other Buffs: There are other buffs that can give you a hint as to what happened in a fight as well. Nature's Grace is another good indication of how well you transitioned from one Eclipse to another. Shooting Stars is RNG based, but it can also help to explain a DPS difference. I also like to look at it to compare the uptimes of Shooting Stars. This is a case of where a lower uptime is better because it shows that you are good at keeping Starsurge on cooldown. There also may be fight specific buffs that need to be managed

    In this example Rootcanal had Owlkin Frenzy and I didn't . He was also a little better at using Shooting Stars then I was, but for the most part we were about the same.
So why did Rootcanal beat me in DPS by 3k DPS? He did cast a tiny bit faster then I did and got an additional Eclipse proc, but I have a hard time believing that those differences resulted in a 3K dps difference.

There are three other reasons why he how performed me on this particular fight.
  1. RNG: Look at the crit rates of Wrath and Starfire. My Crit rates were below average. His crit rates were above average. RNG is rarely the problem people like ot make it out to be, but every so often it does raise it's ugly head and bite someone. This is clearly one of those cases, and there is nothing I could have done about it.

  2. Power Generator: He had twice as much uptime on the debuff as I did. It's hard to judge special buffs some times because it may or may not be your fault for not getting them. In this case, we had one spawn during a spread out phase of a fight. He was near it and I was not.

  3. Gear: Honestly, I hate to list gear as a reason, because most of the time gear is used as an excuse to ignore the other reasons for poor performance. That said, it is a factor, and needs to be taken into account.
A Word of Caution:

I think you can see how powerful a tool CompareBot can be when used correctly. However, you have to be careful how you use it. Make sure you pick a relatively equal player for comparison. If your guild is 5/12 on normal modes it's may not be a good idea to go and compare yourself to Lappè of Paragon who is 13/13 on hardmodes. In this situation it could be hard to tell if the differences are gear based or performance based. I would also suggest that you use multiple people to compare yourself against.

6 comments:

Anonymous said...

Hi,

Trying to find where you see casts per minute?

Thanks.

Graylo said...

@Anon

It's actually Hit's per Minute on the analysis under the Damage tab. I will change the post.

Dave said...

:D Glad you like the tool. Let me know if there's anything else I can add to make the process easier / more informative.

Anonymous said...

Thanks for the quick response on the first Q... looked a little closer and figured it out myself.

I'm running through the logs for tonight's 25M Blackwing and just want to thank you for enlightening me on this and for the author of the tool. This is great. Really helps when you can compare same class/spec on the same raid. You know what happened etc. I'm seeing where I'm leaving some DPS on the table! Not keeping my dots uptime as high, and I wasn't using Lifeblood.

The report is located here. Maybe you can find some more areas where I can improve.

http://www.stratfu.com/kamigami/comparebot/4d425bd38ead0e882d000000#energy

Cheers.

Ambermist said...

This is an awesome tool; just a week ago I was sifting through logs trying to see how the other moonkin in our raid and I compare. This is just what I needed. Thanks for highlighting this and explaining exactly what to look for.

Anonymous said...

I'd go up against Lappe only being 5/12. That guy's scrub.