Is this truly wrong?

[Estelle9lives]

And like I said at the start. Take a 100th player put him in a 10th group and his performance will be nowhere close..

The way you guys are saying things imply comp/kill time is all that factors when it comes to percentiles, which...Is not true at all.

This is from a O5S clear party that I helped yesterday. DPS wasn't very flattering and we didn't have meta comp. As can be seen here 3 people were into the orange/purple zone (even more props to the healers since we had deaths). First and foremost, player skill is the biggest factor when it comes to percentile. Secondly, I'd say it's gear, and then finally group/overall group performance.

Inconsistent groups may lead to deaths, and if you die certainly your percentile is going to be affected - on that end, I can definitely agree that bad PUGs can cause FFLogs to not have an accurate representation of your performance. But if you're pugging and the run is relatively clean for YOU(i.e no deaths, damage down debuffs caused by other people etc), then most of your performance relies entirely on yourself.

A player that can generally perform at a decent level on average is NOT going to have their numbers crushed due to external circumstances or comp alone.

[KaivaC]

snip

Admittedly I don't use it to its full potential because I simply don't know how to do deep comparisons like that. I know how to look up how the timeline and stuff like that, but I don't know how to find, say, the number of casts of a certain skill versus the number of casts from another player. Unless I'm supposed to tally that up manually. In which case, I dunno if I can get dedicated enough to do that

[Ghishlain]

Admittedly I don't use it to its full potential because I simply don't know how to do deep comparisons like that. I know how to look up how the timeline and stuff like that, but I don't know how to find, say, the number of casts of a certain skill versus the number of casts from another player. Unless I'm supposed to tally that up manually. In which case, I dunno if I can get dedicated enough to do that

A little bit above the timeline graphic, there are options for things like "Summary", "Damage Done", etc.. At the far right of that row there is an option for "Casts" that will show the total number of casts each team member has completed for that log. You can then click on individual party members to get a more detailed breakdown on number of casts per spell / ability / weapon skills they've used.

[Trogvar]

The biggest problem of FFlogs compare to Warcraftlogs is that there is no split by item level. So if you have two players with 350 ilvl and 370 ilvl and same skill, guess who ll have higher percentile.
The other problem is food/ pots are required if you want to push for top percentile as well.
The raid composition matters as well, for example, BRD without DRG buff and with DRG buff have completely different percentile.
Overall, the correct rotation and the uptime contribute the most to the percentile, but there is no need to be obsessed with purple/ orange parces as long as you can clear the encounter.

[KaldeaSahaline]

Admittedly I don't use it to its full potential because I simply don't know how to do deep comparisons like that. I know how to look up how the timeline and stuff like that, but I don't know how to find, say, the number of casts of a certain skill versus the number of casts from another player. Unless I'm supposed to tally that up manually. In which case, I dunno if I can get dedicated enough to do that

Kaiva - see attached.


You can use FFlogs to directly compare against people by the URL. You can also ask FFLogs to directly compare you by setting parameters for it to find one.

If you are looking at a log in the top right there are buttons. Analyze is the default, but next to it, is compare. It gives pretty easy instructions on how to compare and then you can look at it like it was one combined log. For instance the image above shows you and I comparing casts.

Now granted, my static kills O6S about 2 minutes faster than you did (I grabbed a random old log to make a PLD to PLD comparison for illustration) so in order to make a more valid comparison I cut the last 2 minutes of your fight out by clicking on the graph and dragging right until the vertical dashed line (which signifies the end of my log).

As an exercise - strictly based off this data - what are some issues you see (on my side or yours)?

[KaivaC]

snip

Is that my most recent run? Oh, that was a really terrible run for me. I think I died once or twice. But comparison wise, I missed a lot of holy spirits, which I do know I tend to forget about in sword oath. Same with requiscast.

[ChaoticCrimson]

I know you're not focusing on SMN for the moment, but I was able to take a look at your gear on Lodestone Kaiva and I was wondering if it'd be okay to offer some suggestions on that. I think I might be able to help a little bit, if you're up for it.

[KaivaC]

I know you're not focusing on SMN for the moment, but I was able to take a look at your gear on Lodestone Kaiva and I was wondering if it'd be okay to offer some suggestions on that. I think I might be able to help a little bit, if you're up for it.

Sure. Though to be honest, I haven't done any melding or upgrades to my gear since starting this thread. Lodestone's only showing me as SMN atm because I'm in Eureka, otherwise I'm on a tank or heal job. But, fire away.

[Ghishlain]

I'd like to add, since we're on the topic of dissecting logs, that even a small increase in optimization can lead to rather substantial gains in percentile ranking.

For example, say two people were discussing a SAM rotation and discovered from their theory craft that if they made a small rotation change it would lead to an overall DPS gain of 2%. Assuming the SAM in question is a 40th percentile SAM, and using this SAM ranking here statistic, the 40th percentile SAM which has a DPS of 5,823.65 would go up to 5,940.12 DPS which is higher than a 50th percentile SAM DPS. A 2% gain resulted in a 10th percentile gain on the charts in this particular example.

Another example - let's take some numbers from the other thread regarding how a BRD loses something in the realm of 500DPS if a DRG isn't in their party. Using these statistic would push a 10th percentile BRD to close to 50th percentile and a 50th percentile BRD to close to 90th percentile. 500 DPS is a net 9-10% DPS gain but leads to close to a 40th percentile increase.

Small things don't feel like much but you can see how quickly they can change your rankings looking at the above. Understanding what you can do to get those small performance optimizations will help you shoot up in the percentile ranking. And something like FFLogs can help you dissect what you can do to continue improving. Not every circumstance will be the same but you can gleam enough information from the depths of the logs to get at least some idea on which direction you should be pushing towards.

This example above also makes me realize why my personal MCH numbers in a DRG-less static during the 3.4 raid tier were so terrible.. *coughs quietly*

[Bourne_Endeavor]

A player that can generally perform at a decent level on average is NOT going to have their numbers crushed due to external circumstances or comp alone.

To emphasise this, I generally average high 80s to mid 90s. With a death this week in o6s, I still reached 62% A 25% penalty for 90 seconds is far harsher than a random Ninja messing up Trick Attack. Will you dip in a less optimized group? Of course, but you aren't going from 95% to 50%. Bard and Machinist, unfortunately, stand as the sole exception due to piercing.