Raw Data needed for HK -> RP approximation

[Murtrude]

Like many people on this server, I have been trying to determine a method for calculating (or at least estimating) how much RP will be gained at the end of the week. It would be helpful to get raw data from other players to verify and refine my work. Here is what I have so far: The data is just the RP value at the start (week 0), and then the RP and Weekly Honor gained for each subsequent week.



Now it is apparent that the Rating Points you have at the start of the week will decay, and if you do not pvp you will lose rating and ranks. Some people differ as to whether or not the decay of rating points is done before or after the weekly RP gain is added to the total RP. I have therefore calculated
both outcomes (the difference being a single multiplicative constant).
It is worth noting that I have assumed the decay rate to be 0.20 (20%). The reason being is that I have seen many other people use this same number, but it may be different in actuality, only more data will confirm.



After plotting the data pairs and curve fitting, these are the best approximations I found.



Using the equations of the lines of best fit, we have:



And so the final Macro looks like:

/script P=(math.floor(GetPVPRankProgress(target)*10000))/100 W=UnitPVPRank("player") N=(W-6)*5000+50*P Q=((W-5)*5000)/0.8-N H=(Q/574.29)^4.016 SendChatMessage("Rank "..(W-4).." "..P.."% RP Total: "..N.." Honor to rank up: "..math.floor(H).."", "emote")

EDIT: Added this macro which tells you how much honor is needed to remain at your current rank progress

/script P=(math.floor(GetPVPRankProgress(target)*10000))/100 W=UnitPVPRank("player") N=(W-6)*5000+50*P H=((N/4)/574.29)^4.016 SendChatMessage("Rank "..(W-4).." "..P.."% RP Total: "..N.." Honor needed to float: "..math.floor(H).."", "emote")

I will update this as I collect more data from my own characters, but if anyone has data from trying the same thing, I would appreciate sharing information to get a better picture.

What I need from players is just their Current RP from whenever they start taking data. Then the New RP values paired with the amount of honor earned in that week (as seen in the first picture).

EDIT: Updated macro and graph to incorporate new data from Henkebenke

[snawfu]

Yeah that macro isn't right for sure.

[Murtrude]

The macro is correct, it now takes into account the decay for the RP gained during the week and gives an approximate honor value

[snawfu]

Well it's telling me that I need 15k rp for my next rank and that is 100% not true.

[Murtrude]

Either one of two things is happening:

1) Your rank is either 1 or 2, the macro doesn't work for these ranks because it counts in increments of 5000RP and the first two ranks are not in multiples of 5000

2) You actually do require 15K RP to progress to the next rank

This happens at high RP because you have to not only get 5000RP to rank up, but more RP to make up for the decay of your current holdings,

ex:

Rank 10 (40000 RP)

If we gain 5000RP our new total is 45000RP which decays to 36000 and we actually lose progress and go to rank 9

In order to get to rank 11 we need 16250 RP which gives us a total of 56250 RP, which then decays down into 45000 RP, which is exactly making rank 11

So you need 16,250 RP to go from 0% rank 10, to 0% rank 11 in a single week

[Arts]

It doesn't work %100 and no my rank is not 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 12 or 13 or 14

[Jarek]

Yeah that macro isn't right for sure.

EXTRA WEEK HYPE!!!!!!!!! AV PRIMAD?

[Murtrude]

bump for more data please

[arOo]

For anyone considering using this macro don't expect any relevance by using it.

OP let me save you some time by saying this:
Your effort is completly irrelevant since these graphs need to be readjusted every week.

However, here's a couple of things that you could do to make your data somewhat useful:

- Incorperate Graphs for the individual BG- Weekends (big differences in honor caps on brackets)
- Incorperate Bracket Breakpoints - RP Gain is only (somewhat/heuristicly) linear within a bracket.
---> RP=(FloorRP(Bracket)+ [(Your Honor/Bracketleader Honor)*1000])
ThIs is an uncosty approximation for anyone that isn't in dire need for exact data.
>>And with the large set of data on this server, this should be good enough for almost anyone.

Hence the information you need to determine a somwhat working function is on a variant 3-week cycle in which you have to know:
- The amount of players on your faction eligable for Rank Calculation (to determine how many people are within bracket 1, bracket 2... --> consequently getting the standings)
- The Honor of each bracket leader.

Any function that disregards these points, won't ever stand up (let alone to next week).
Consequently you will notice that the effort is pretty much in vain - the only effictive way of gathering this data is to have access to the database itself.

So what can or should you do?:
- Ask Nost to implement such a feature if you absolutely want it. (Not a fan myself - since it would provoke player behaviour that would be detrimental to Rankers (dynamic adjustment of bracket caps))
- Work with the Bracket Caps that will be determined each week by Rankers.
(Only avaiable for Bracket 1&2 / but why should anyone care before that)

[Murtrude]

My goal is to get an approximation that will provide at least a good estimate for the amount of RP that will be gained from a given amount of honor. Yes I could track the standing as well, but since you can't know your standing before the week ends, there is no way to incorporate it into an estimation. Also without knowing the internal workings of the game, it is impossible to determine the number of brackets, and the width of a single bracket. This macro is a simple curve fit that will improve with more data, and it uses raw data that is 100% accurate (total RP and honor gained).

So far my efforts have yielded a 98.7% correlation with the lowest estimate being 95% of true value. It works well enough to give a real estimation that is useful when setting weekly goals. All I really need are people who can post their current RP and Honor gained for more data.

If you actually try the macro and run the numbers yourself, you will see that it is indeed accurate and relevant.