Making a universal HPM reader for everyone

Hello there, I've been playing tera for a long time and, as an old tera player People has been asking me what HPM (hit per minute) they should have with their classes.

To answer this question I have made a program who reads a json file extracted from shinra meter.

For each player you have : global remarks (how many death if any, for how long etc)

then the DPS on the picture below 22 325 552dps (22,3m)

then a table with the following :

the skill name

what are your hpm compared to what it's expected to be(❌if you're below expected hpm, 🔵 if you're in the expected hpm range, ✅if you're above expected hpm range)

a comment if the hpm are above or below expected,

the hpm of the player, and the min expected and the max expected.

Below there's general remarks, here one arcan barage diden't went through the boss before exploding and some hits of tb and bombardment diden't hit

below that in green the unchained skill (technically most classes has chainning mecanics just invisible)

And here is where i need you, Please help me define what are acceptable HPM foreach class and what extra tiny things I can look at foreach class.

Added a grade base on the hpm compared to the expected hpm and their contribution to dps (for exemple lancer sheild bash dosen't contribute a lot since it deals almost no dmg, wheras assault contribute for a big amount since it deals considerable dmg)

you get a 100% grade if you match every expected hpm for important spells but you can go over if you're above expected max hpm for important spells. you also get a grade from F to Lolimagie (my sorc in seren server) based on the accuracy% of you playing the class

I've also added icons for the spells

I have made a rough translation for most of the program output, (the spells' names are still in french but there's an icon with it and for spells that appears more than once they're special mention like hit1 or hit2 when the shinra differentiate them)

To make the final note easier to understand I've added two colums, the contribution to the rank and how much a spell in the expected range contribute to it.

To get a reasonable contribution if in range for most classes I've taken the dmg % of spells in normal runs, removed the ones who contributes for 0% or less and keept the % contribution of each spells (here's below an exemple of a sample sorc run) (meteor is separated in 2 hits in the shinra contributing each separatly in the final rank but shown, as one spell)

foreach spells :

if you're higher than expected hpm you get (contribution expected * the ratio of how much you were above the expected max)

if you're below you get (contribution expected * the ratio of how much you were below the expected min), else you get the contribution

Here for exemple we've got a played who overused but neglected , as such he gets more contribution for than expected but he looses some on

(Screen of a sorc run in seren patch 101)

Of course gameplay might vary for players but if someone does great I expect the rank to be high even if the player neglected what I expected as high in dps spells but did amazingly enough on other spells to get more dps

For classes like sorc who gets a special fusion mecanic that makes 3 spells on one key the two elements fusion share the same total contribution and makes sorc ranked on 104%

as you're expected to get in the ranges for 2 fusion (look at elemental fusion's mecanics for more detail)

Here is for everyclass the sample run used for each spells' dps contribution

Archer

Zerk

Warrior

Slayer

Valk

Reaper

Ninja

If you've got better sources or reliably better hpms for at least a 2minutes fights against any 455+ bosses on official server or equivalent (AKA not menma or servers with significant changes on spells) feel free to post your hpm here and I'll change the grade contribution or expected hpm accordingly.

Using the above program a lot I've just realized now three things,

1st) some classes are a lot stronger than others by a huge margin even more than in old tera and this no matter the fight duration, I think it's due to the crystals changing the fundamental balance of the game.

2d) I've seen way less archer/ninja/reaper/slayer than valk/sorc.

3rd) I now have enough data to give a grade to many players but I guess That would be wrong to do it without their consent.

However if you're a dps or a tank (and not a brawler since their hpm are somewhat irrelevant overall with parry not beeing differentiated from main hit via shinra) and you'd want to improve, feel free to dm me on discord and send me the a '.json' of fights extracted with shinra where you feel it represent you well. You should only send fights in a 5 player dungeon and at least a 1min 30 long fight since shorter fights tends to makes hpm way higher than they should be and healers and tank buff affect your hpm.

PS

Currently I can't auto rank healer since it's not only he uptime on buffs that matters but, sometimes it's better to delay buffs and have a lower uptime to give the group a better dps.

(A good exemple of that would be waiting 3 secs for eredict of judgement to buff as enrage starts rather than cast on cd with high burst classes.)

After a lot of testing and trolling around with the hpm reader I've made a rough estimation of classes dps compared to accuracy, and from that data I made extrapolation and estimated for BIS the dmg compared to accuraccy of classes on this server.

As a reminder you should be able to reach around 100% accuraccy on anyboss that dosen't require you to stop dps (IE not lachelitas 2d boss, slayer etc) with a lancer in your group (unless you're a tank).

Healers dps values comes from estimations compared to lancer's dps at bis and aren't really meaningful

I made the calculation for 4 keypoints, F rank (10% accuraccy), C rank (50% accuraccy) ,A rank (80% accuracy), SS rank ( 100% accuraccy) making the accuraccy variation from what I've observed in players (Ie I've watched where C rankers lost their accuracy and used that to calculate a 50% accuraccy rather than just loosing 50% on the main spell as nobody plays like that).

We can see that classes that has great out of burst dps benefits a lot from crystals as well as classes that has native high crit rate on main spells.

As brawler hpms aren't good estimation I've calculated their position based on what I see on this server and what I saw on official server

On official server on patch 100 class ranks in kr where (with BIS gear)

( Valk == war ) >> (ninja == zerk == reaper ) >= (sorc == gunner == archer) >> slayer ( slayer sucked back then xD )

While I find it cool that slayer got a little buff and valk went from busted to 3rd place, wardps got stronger and we have now a cluster of classes that are just way below the main 3, the difference is bigger now between strongest classes and average classes than it was back on official servers

When we consider that the top 3 classes also are classes able to dodge easely as they all have good mobility and survivability it feel a little weird that classes with less survivability are also just below dps wise when perfectly geared and played perfectly.

I know balencing is hard but making the bad classes more viable could be a great way to bring more people as valk and warrior have been top tier in everyserver that used the leaked patch for a long time now. (Keep the buff on slayer tho, they need it)

So that pretty much locks the issue, I wont share the source code and that's the reason why I made an .exe out of it (since with the source you could easily analyze the runs of everyone it's one foreach loop away from doing it and people could get toxic over it). I don't really see a way to solve the issue.

I get why you would be concerned about security issues, tho since it's hosted on github you woulden't be the entity that have to answer from it.And since it only does local calculations with very restricted use input the security risk is low, yet you shoulden't take my words for it ^.^). As you know, I could be a evil hacker in disguise or something like that...

But I get your point, it's too bad for the players who would have benefited from it. I guess that's life.