An Attempt of an Objective Look at Scenario Balance in RoR

[GorkaBorka]

This is very impressive! Thank you very much, good sir!

[Cyrylius]

I would say I'm impressed if it wasn't already said by literally everyone posting in this thread. I like the data, if only for fun, but one thing I do not like is how you get rid of various outlier-y cases. For example, every class can join the scenario late and get low score, but cutting data you suspect to be tied to cases like this might also eliminate cases of genuinely bad performance (say, unguarded choppa rushing into slayer ball on cooldown doing negative damage is actually an important factor to consider, especially if he adds gtdc, but you ignore him altogether). On the same note, if a tank is snb and guards rdps who doesn't take any damage (and perhaps doesn't know how to use challenge) he might get more damage than protection while in defensive spec. In my opinion you should consider all the data, especially if you already did additional tables for top 50% of class users.
Edit: in game statistics show ranged and melee icons to indicate off spec, you could maybe use those to at least spot DPS healers? Not sure if they show in killboard data, but perhaps if they do?

[Cultist]

Very good work, really appreciated it!

Could it be possible to scrap oRvR kill counts within a period of time (let's say 1 month) at lakes?

I saw GamesBond on the latest RoR official streamed event pointing out Order had 10k kills over Destro during that streaming on Talabecland, which is quite impressive. I'd like to contrast kills on oRvR as well since there's always a faction that has the "feeling" of getting heavily farmed.

And again, thank you.

Edit: Apparently video got removed but it took place on 15th April 2022 Realm Event.

[tazdingo]

protection stats are unsurprising, SM can quite easily 'farm' protection with group bubble and raking talons if they wish which is actually a profitable thing to do as far as bonus crests go. bg bubble also lands at around 1k~ if you max the tree which with your dp means 2k protec every 10 seconds. this is miles from being a problem just interesting to see laid out, i didn't think bg would beat bo

[Haojin]

thanks for the effort; wish there is a way to see data from strict difference between premade / pug.

in general whoever has most premade or premade-ish groups in current scenarios tend to win.

[Sofong]

hmmmm this is interesting

Source: ./subdivided_scenario_statistics/scenario_statistics_t4_city.json

Total Scenarios: 34
Order Wins: 18 (52.9%) [total points: 9030]
Destro Wins: 10 (29.4%) [total points: 7280]
Draws: 6 (17.6%)

[Garamore]

That one is down to there being more organised groups that preferred to go order more than they went destro. A lot of the wins on either side will be the same people just xrealmed for a city.

Order have a better melee train if you take the right classes but most down to players.

[Scottx125]

Wow, that’s impressive.
I am a strong believer in balance being based around facts and numbers, not on looks, feels and my experience.

I agree with that sentiment partially, numbers don't lie. But a class has to have fun aspects to it otherwise what's the point in playing it. You want specs and classes to be fun and unique whilst being balanced. After looking at the data I have to agree that the sample size probably isn't big enough for an accurate representation and we need some way of logging 24v24 or ORvR. I also think you shouldn't remove outliers such as players who join late etc as that may skew the data covering poorly performing classes or players.

[Lion1986]

I recall someone said that class balance was NEVER taken in account for scenarios, because classes overall would be balanced for open rvr. This means that this topic is quite useless. If you balance scenarios you unbalance open rvr and vice versa.

[dalen]

This is interesting stuff. Was kind of hoping more community members would do more interesting stuff with the killboard data.

I will try to fix the skipping issue in the API so you can fetch all of them.

But regarding the scenario winners, try using the `winner` attribute, especially for city sieges that should be more accurate than looking at the points score.