[Official] Request for Feedback on Potential Scoring Changes

[Twill]

New Scoring Proposals - Feedback Needed

As many of you hopefully know, we announced here that there would be an update the the scoring system to ensure that it becomes "stdb04-proof scoring". Seeing as any changes made will have a significant impact on the way the ranksing currently stand, change and move forward, we would like to ask for your feedback on two possible itterations of the future system.

These are not the only possible solutions, but the ones we currently think best, please suggest actual alternatives or give specific criticism on them so that we can ensure the best possible scoring system is put in place.

The following proposals are based on suggestions which have been made throughout the site which range from a loss cap (i.e. a player cannot gain/lose more than 50 points per game) a floating max (i.e. you cannot gain/lose more than 10% of your score) and complete overhauls.
You will notice that we have opted for a mix of these suggestions leaning towards a fixed level of gain/loss based on rank.

Please go through these proposals with a fine toothed comb and pick out any and all possible problems. Lack and I are only 2 brains, we miss a lot of stuff. We know each proposal has specific implications for how easy or hard it is to gain a rank or lose one, which are easier and what combinations of ranks will screw you over faster, we just cant figure out what they all are.

And without further ado:

Option 1
Matrix scoring

The following scoring matrix is based upon the current rank requirements and points equation.
To read: your current rank is on the top, the person you beat is on the left, your score is where the two meet.

(i.e. a new recruit beating a captain would get 28 points, a captain beating a new recruit would get 14.3)

What this mean is that as a private you will always get 40 point for beating a colonel and 20 points for a sergent. Once you become a captain you will always get 14.3 points for a private and 57.1 for a general (not yet rounded to whole numbers). This happens regadless of current score, only rank.

Option 2
Rank Differential Scoring
This proposal is based on the above matrix but depends rather on the difference in rank number rather than having a matrix.


Note that in this one, a private beating a major (a difference of 4 ranks) would get 28 points, in the previous matrix they would get 32


As you can tell, there are distinct differences to both.

What, if any, are the advantages/disadvantages of each.

Thank you for your help, hopefully we can avoid another Grievances 1.0 fallout

Twill

[Haydena]

I see a disadvantage by doing scoring through the matrix. What if a lower rank has enough points to be very high rank, but not enough games played? He can collect huge amounts of points for players a few points ahead of him or even lower than him on the scoreboard.

For example rocksolid (sorry about this ), is so good he's managed to accumulate 1600 points and has only played 28 games, thus he is still a lieutenant, if he beats a major with 1600 points as well, then he will get more points, instead of the 20 he would get now, he gets 26.7 using the matrix scoring. (I think)

I'm too tired to see any problems with the rank differential scoring at the moment.

[Marvaddin]

This is a good point, Haydena. Maybe you should consider only the number of points, not really ranking to do this. But the matrix idea is the best one in my opinion. The second one gives, at first sight, less points per game, huh? Same level, 15 points? But, that idea of "difference" between rankings, well, private - major = 4, but major - private too. So, Im now a Colonel, if Im defeated by a private, the difference is 5, so I lose 32 points... But if I win, I will gain 32, is this right (wow, now I can be General! )?

Seriously, I believe the difference idea doesnt work. We need something like the matrix, maybe a bit refined. For example, I agree that recruits and privates gains / losses should be the same, but not for sergeants.

I (and someone else) suggested already a bank of points taken from deadbeats... They could be distributed to all players. Did you think about that?

[Fieryo]

What if we keep the matrix idea in #1 but instead of ranks its point groups? So people with scores of say 1000-1200 would all be in one point group instead of rank, and they would all lose/win the same amount of points if the beat/lost to someone with say 1201-1400....or something

[max is gr8]

We should make it so you can only play with people who are within 100 points of you so it stops a newb playing the top rank which I can't remember

[Jota]

Will using discrete categories like this exacerbate the issue where the order you finish games in affects how many points you gain/lose from them? (Since a win or loss could propel you right into a new rank/scoring category.)

[lackattack]

I think Rank Differential needs to be expanded: if a private beats a major 28 points are exchanged, but what if the major beat the private?

[Marvaddin]

I think Rank Differential needs to be expanded: if a private beats a major 28 points are exchanged, but what if the major beat the private?

A major beating a private? At least 40 points, since that seldom happens

[mrdexter]

I think Rank Differential needs to be expanded: if a private beats a major 28 points are exchanged, but what if the major beat the private?

Are you saying if a Private beat a major he would deserve more of a points win than if the major beat the private?

I'm not sure I fully understand what you are asking here, however i do agree with the sentiment that it needs expanding on

[Banana Stomper]

What if we keep the matrix idea in #1 but instead of ranks its point groups? So people with scores of say 1000-1200 would all be in one point group instead of rank, and they would all lose/win the same amount of points if the beat/lost to someone with say 1201-1400....or something

Isn't that what a system based on ranks does?

[Jota]

Not if someone has a high score but not enough finished games to make the next rank, like Haydena described.

[Haydena]

Yeah.

What I believe should be put in place is something like 0 > 800 (stdb04 proof as it will be just like he has 800 points, there is no point him dropping so low, and no massive losses for generals or majors... then going up in 50s or 100s until it reaches 2000 or so, then it goes up in 200s, till 4000, then its simply 400 +...

I don't know if that's a good idea or not...

[qeee1]

The second one I don't think has been explained well. Or maybe it just wasn't thought through. What seems to be happening is that a private beating a major gets 28, but similarly a major beating a private gets 28pts. I think you still need to do a section in that scoring system for
-1 rank
-2 ranks
etc...


Of the two, I think the matrix works best, the exact figures may need some fine tuning, but I think that overall it works better than the other one, simply because it can be more finely tuned to each rank.


The disadvantage in both the scoring systems is the grouping format.

2 problems that I can see stemming from the grouping system:

1. Score altering tactics

As Jota said:

Will using discrete categories like this exacerbate the issue where the order you finish games in affects how many points you gain/lose from them? (Since a win or loss could propel you right into a new rank/scoring category.)

Players may try to stay in a lower rank until they can get a big win to propel them into the higher rank to affect their score more, or if you see a loss coming to have that first and delay winning another game so you'll gain more/lose less, and other things like that.

2. It's a bit vague.

Groups can span anywhere from 200 pts, to 2000pts, to (technically) infinity. There's probably a big difference between beating someone with 1600 pts and 1900pts for example, but both are considered Majors. I think to keep the system fair, groups would have to be pretty small (maybe 50 pts). Keeping the groups small would also help address the first problem. However very many groups would be difficult to calculate and explain.

What I'd propose would be simply to cap the amount of pts a player can lose in a game to 100. Which also theoretically means a max gain of 500. And just keep the old scoring system. It would only come into play in a few rare situations and would allow us to preserve our existing scoring system, which is in my opinion the fairest, as it takes players based on the smallest group possible of 1pt. I like the system as it is and would like as few changes as possible, so I'm opposed to these complete overhauls.

[Jota]

The 100 point cap qeee1 describes seems like a nice, simple solution that would solve the current problem without bringing in any more complications than we have to.

[Twill]

doky oky....

Apparently I had a brain fart the day I drew up the rank differentials one, I'll add negative rank diff (the major beating private) just for good measure later this week...or are we just scrapping that idea outright?

Would it be worth tweaking the matrix to smaller groups or is that a moot point and everyone just wants a cap on the lost points?

[alster]

We should make it so you can only play with people who are within 100 points of you

Oh... How did you figure that the top 10 players will be able to play? As the scoreboard is today the no. 1 would only be able to play with no. 2 and 3. Then, after the smoke has settled it's not unlikely that the then no. 1 guy are 100+ points ahead of the no. 2. So then what?

[Marvaddin]

100 points lost in a game is a bit high. I lost 65 in a game today, argh!! Since the game is not really based only in skill, but also in luck, 50 points is a good maximum. And I still like the matrix.

[qeee1]

There is a problem with the 100 pts thing.

If a player is on 90 and someone on very low points beat them they could go into minuses, so you'd probably have to combine it with a minimum point level of 1.

Alternatively you could make the maximum points loss score specific, to say 5% of your score. At 2000 points that works out at a loss of 100, at 1000, 50 etc. This would also mean a player couldn't drop below either 19 or 9 pts (depending on how points are rounded off, I'm not sure what the method is).

They are the two approaches to the problem I can see if you're using a pts loss cap.

As to the level it should be set to, I don't think 100 is too much to lose in a game, I think it's about right, but maybe others disagree.

[zorba_ca]

I believe we may be overanalyzing the problem.

The current system is only a problem because one player may have 1/1000 of another player's total. To minimalize the exponential aspect of the current ranking system:

We should just make a minimum score of 500!

This is simple, minimally alters the current point structure and works in just about every scenario. Plus, other than Std (who is reducing his score on purpose), this seems to be as low as anyone can go.

Even if one player makes General and reaches 4000 points (a situation we are far from), and another player is at the minimum of 500, the most points that player would receive is 160. Hardly worth the effort!

That's my vote.

[Darklord001]

I like zorba's idea - seems simple, fair, take into account qeee1's point about the benefit of the existing system working in 1 pt increments and all of that.

but, as an aside, does anyone else have a desire to see stdb04's "cheat" work - at least long enough to get a snapshot for posting in the forum?

I'd love to see the game window show

stdb04 awarded 20,000 points
darklord loses 5600. . .
etc.

as long as the scores were restored and then the fix put in place afterward!

he might be gaming the system, but it was fairly cunning!

[lackattack]

So instead of the matrices, it seems the leading solutions are:

A) minimum score of 500

B) maximum point transfer of 100

I wonder which would be better?

[Jota]

I'd go with B (so long as scores weren't allowed to drop below 1).

[AK_iceman]

i would like the minimum score of 500 better, that way the scoring stays the same

[Fieryo]

but then what happens if I have 500 points and lose to someone who has say, 1000. would they still get points but i just wouldnt lose any? where would those points come from? Or would it be that no one could get points from beating someone with 500 points?

[zorba_ca]

If a player with 500 points loses to a player with 1000, the player with 1000 would get 10 points (as it stands now). The player with 500 points would lose 10 points but his score would automatically get bumped to 500.

Perhaps, if someone reaches this Mendoza line of Conquer Club Incompetence we can come up with a new scarlet letter type ranking.

Some suggestions: Min., Reserve, Guard Duty, K.P., Discharged

For fun, perhaps Lack can track the player's further ineptitude with a not so subtle negative beside the ranking. For example:

stdb04 -- Discharged -- 500 (-364)

As an added layer of penalty/complexity/enjoyment (if Lack could swing it), perhaps the player would have to work his way out of the minuses before any victory points count to working his way above 500 and being "Reinstated" to the Conquer Club army.

I don't think there would be much opposition to this suggestion's immediate implementation!