RR Calculation Wrong?

Post a reply

Confirmation code
Enter the code exactly as it appears. All letters are case insensitive.
Smilies
:points: :-D :eyeroll: :neutral: :nmr: :razz: :raging: :-) ;) :( :sick: :o :? 8-) :x :shock: :lol: :cry: :evil: :?: :smirk: :!:
View more smilies

BBCode is ON
[img] is ON
[flash] is OFF
[url] is OFF
Smilies are ON

Topic review
   

If you wish to attach one or more files enter the details below.

Expand view Topic review: RR Calculation Wrong?

Re: RR Calculation Wrong?

by Mercy » Mon Jan 14, 2019 8:23 pm

A_Tin_Can wrote:
Mon Jan 14, 2019 11:46 am
@cspieker - is the reliability calculation the same on vDip?
I presume this question was meant for me? Yes, the calculation is the same on vDip.

Re: RR Calculation Wrong?

by A_Tin_Can » Mon Jan 14, 2019 11:46 am

@cspieker - is the reliability calculation the same on vDip?

Re: RR Calculation Wrong?

by cspieker » Sun Jan 13, 2019 9:30 pm

Yeah, I figured the cube was to spread out the distribution at the high end. But I don't understand why that would be necessary. The utility of this RR is so that one can judge whether someone is reliable. If someone is 99% or 100% both those seem basically the same to me.

Not sure why one needs to break ties or spread people out. This is a metric that is useful for basically seeing if someone is at or above a threshold that one considers acceptable.

Re: RR Calculation Wrong?

by Mercy » Sun Jan 13, 2019 8:56 am

The probable reason for the cube in the formula is so that it is easier to discriminate between high but different Reliability Ratings. If there was no cube, all players who currently have a Reliability Rating of 99 or 100, would then all receive a Reliability Rating of 100.

I agree with cspieker that the fact that the Reliability Rating using a % sign is very misleading, as it is not a percentage at all. The fact that we, with some frequency, get topics on the forum from players who think the RR calculation might be wrong, is proof of this.

As usual, a solution already exists and it can be found on vDiplomacy. On vDiplomacy, if you look at someones Reliability Rating on their profile, no percentage is shown. For example, on webDip it says "Reliability rating: 94%" when on vDip it says "Reliability: R94". On top of that, if you hover with your mouse on top of the Reliability Rating, the formula that is used in calculating it is shown. This makes it a lot more clear. Needless to say, on vDip we never get questions on the forum about Reliability Rating.

Re: RR Calculation Wrong?

by jmo1121109 » Sun Jan 13, 2019 6:00 am

cspieker wrote:
Sat Jan 12, 2019 7:47 pm
My main issue with that stupid formula is that is presents itself as a percent but it is not.

The average of the non-NMR% and the non-CD% would actually be a percentage in some sense.

I guess you guys must have discussed why that simple intuitive formula didn't suffice and so you had to cube it for some reason.
This is an open source website. If you'd like to contribute a better formula or report an issue with an any part of the site then feel free https://github.com/kestasjk/webDiplomacy/issues

Or, explain here in detail what you would do instead and why. But fair warning if the ideas are wrapped within snarky replies I won't waste my time.

As for why the formula is the way it is, no idea, not a math guy. I just coded the SQL for the formula I was given, so have at it.

Re: RR Calculation Wrong?

by cspieker » Sat Jan 12, 2019 7:47 pm

My main issue with that stupid formula is that is presents itself as a percent but it is not.

The average of the non-NMR% and the non-CD% would actually be a percentage in some sense.

I guess you guys must have discussed why that simple intuitive formula didn't suffice and so you had to cube it for some reason.

Re: RR Calculation Wrong?

by PRINCE WILLIAM » Sat Jan 12, 2019 7:25 pm

Why don't we use the simple analogy of NMR to overall movements? Is it to make CDs more punishable?

Re: RR Calculation Wrong?

by jmo1121109 » Sat Jan 12, 2019 3:22 pm

cspieker wrote:
Sat Jan 12, 2019 3:05 pm
That is a seriously stupid formula.
Feel free to offer a better one, otherwise that comment isn't really useful.

Re: RR Calculation Wrong?

by cspieker » Sat Jan 12, 2019 3:05 pm

That is a seriously stupid formula.

Re: RR Calculation Wrong?

by fadgyasb » Fri Jan 11, 2019 7:52 am

Okay, thanks, so both the factors are on the cube... okay, thats strict. Fair enaugh. Thanks a lot

Re: RR Calculation Wrong?

by ziran » Fri Jan 04, 2019 3:22 pm

it's not a prediction, it's a history.

Re: RR Calculation Wrong?

by thdfrance » Fri Jan 04, 2019 2:02 pm

I believe I'm more reliable than any equation can predict

Re: RR Calculation Wrong?

by A_Tin_Can » Fri Jan 04, 2019 5:10 am

Your RR is correct:

(((95+92.1)/2)^3)/10000 = 81.87.

Re: RR Calculation Wrong?

by fadgyasb » Thu Jan 03, 2019 2:48 pm

I have a very similar issue and been searching for the correct formula that you suggest above from Jmo. Would you maybe have a chance to share the formula?
Since I started playing again, I had a few NMRs, but generally improved my RR, however its been hardly moving since?

Just to illustrate:
My NMR currently is 59/1121, where noNMR% becomes 95%
By finished games of 38, my CD is 3, which makes me noCD% becomes 92.1%.
This should give 87,5% of my RR (which is still far from ideal), however my profile stays at 83%. Where am I taking this wrong?

Thanks

Re: RR Calculation Wrong?

by Squigs44 » Sat Aug 11, 2018 11:35 pm

The calculation is correct. You have to factor in your CDs as well. Your noNMR % is 98.9, but your noCD % is 95.3. You have to average those and cube them to get your RR which is 91.7.

Re: RR Calculation Wrong?

by Claesar » Sat Aug 11, 2018 7:59 pm

Did you use the calculation presentedby Jmo in the other topic?

You seem to be suggesting RR is calculated by (number of moves- ((number of NMRs)/(number of moves))%, but that is not the formula we use.

RR Calculation Wrong?

by Omphalos » Sat Aug 11, 2018 5:30 pm

According to the stats given on my page I have 39 NMRs in 3698 moves (as I write this). This should be a RR of 99%, but it is given as 92%.

Top