Improving the global ranking calculation

i see two ways to improve the global ranking calculation:

1.vary the maximum number of points for the levels depending on the number of players on this level. this would remove the need for limiting the maximum level rank to #99. but it would make some levels more important than others, which is what i would like to avoid.

2.gradually decrease the influence of the level rank. eg. each level rank position for #1-#50 counts as 1 pt, but #50-#100 gives only 0,5 pt per rank, #100-#150 is 0,25 per rank and so on. effectively, even with a huge number of players on certain levels we can ensure that each level influence can be limited to 100 pts and there is still a difference between #600 and #601 (in this case it would be 0,0002441406). i like this idea, but it is not very intuitive like the current "1 pt for each rank".

what do you think?

211 comments on 11 pages: <<<< 1 2 3 4 5 6 7 8 9 10 11 >>>>

and there is one more thing to fix in the new scoring system: "bottle-like" levels

of course these levels could be excluded from the competition, but this is not the best method, one person's "bottle" can be quite challenging for someone else

let's assume these results (current scoring)
#1.6,26 (100 pts)
#2.6,26 (99 pts)
#3.6,28 (97 pts)
#4.6,28 (96 pts)
#5.6,20 (95 pts)

i have 3 ideas:
1.give the equal number of points for people having the same time (but definitely not the medals

), so it is like we counted "group of players" instead of "players"
eg.:

#1.6,26 = 100 pts
#2.6,26 = 100 pts
#3.6,28 = 99 pts
#4.6,28 = 99 pts
#5.6,20 = 98 pts

advantage: you are not "punished" for entering the game after the world record was achieved.

2. similar to 1, but the score depends on the ranking position of the best player in group, not the group itself.

#1.6,26 = 100 pts
#2.6,26 = 100 pts
#3.6,28 = 98 pts
#4.6,28 = 98 pts
#5.6,20 = 96 pts

(a group of 10 players having 6,26 would all get 100 points but then the player with next best time 6,28 would only get 90 points)

3.calculate the score directly from time, not the sequence of players:

#1.6,26 = 100 pts
#2.6,26 = 100 pts
#3.6,28 = 100 pts
#4.6,28 = 100 pts
#5.6,20 = 98 pts
(this would be for the example formula: 0,05 sec = 1 point)

advantage: you don't have to play 1000 times to gain 0,02 sec on "stupid" level which does not allow for showing your skills and only depends on luck.

what is the best method?

Cirenco

SZ: Si hay que elegir me inclino por la "3" debido a que los otros métodos son desmotivantes para jugadores nuevos. Pero yo creo que un buen jugador debe ser capaz de superar esto y creo que se pueden mejorar los tiempos de los "bottle" , además es un mérito hacer uno , y creo que se debe premiar ésto también.

mike flips

It has to be #1.

Number 2 can not work fairly, for example, in Level 1 of Newbie Pack, "Shelf," 30 players have the same time of 7.74. If you get 7.76 you would only get 70 points instead of 100 which is extremely unfair.

Number 3 does not even work. For example, in Master Ko's Level 2, "Bottle," what amount of time would you choose to represent 1 point. If you choose 0.05 sec to be 1 point, then people at rank 100 would be getting too many points because their time is under half a second away from rank 1. However, their time of 6.64 for example is not a very good time for this level, and they are getting to many points that they dont deserve. If you try and decrease it to 0.02 or 0.03 sec = 1 point, then by that time you are defeating the purpose of not having to get that extra "0,02 sec on a 'stupid' level"...

# 1 works because it maintains the competitive nature of every "bottle" type level. In bottle, it takes a lot of work to get a fast time and if you end up at rank 30 you can still get 90+ points. And on levels such as shelf where you either get the trick or you cant get the trick, if you get it you get 100 points, and if you cant get it and still get a fast time using the normal method of beating the level, you can still get 90+ points for your fast time.

#1 all the way

as for #3: of course the formula would adapt itself to the best time for a given level, eg. 100 points for 6.26 (the best time), 50 points for 2*6.26 (twice the best time), 10 point for 3*6.26 etc.
i don't agree that #3 does not work. it works differently

if 100 or 200 people are able to complete a very simple level in almost equal time then they could get 90-100 points - all of them! this would make such levels unatractive for experienced players, we can assume that all serious bikers would get 100 points and that's the point: as a result such levels are "excluded" from the "premier league" competition, but they still can be played by beginners.
well... actually, after introducing the new rule for assigning points to ranks, player #600 will get slightly more points than #601 so maybe this feature of "method 3" is not so important. but the ability to "exclude" certain levels from professional players without hurting newbies is quite interesting.

mike flips

but SZ, for shelf... if you cant get the trick then you are automatically at least 1.5x7.74 which may mean only 75 points or whatever you deem appropriate. This is unfair, as some people (Me, Smalls, Jayce, Louis, and You Qi) all of whom are top 15 players are unable to get this trick and shouldnt have to suffer a loss of 25 points just because we cant get a glitch in a level to work for us.

Also, then you (Sz) would have to sit there and decide what is appropriate formula for each level that is considered a bottle,

it depends on how many people are able to do the trick. if there were 100 trick-able players then you would vote against ranking based method

the need for artificially invented formula is of course the drawback of "method 3", one can always argue that a different formula is better.

Ibanez

Wow Sz - fantastic on the distribution curve, now that's quick work. I think some of the levels/packs may need a smaller 'bucket' size to smooth the curve a bit and when there's someone who literally took forever to complete a level it screws it up a bit too, though the zoom feature is an excellent way to cover this. Just need to highlight 'your' position on the curve and my vision will be complete!

As for calc method, I actually like #2 - it's the Olympic way, if there's a three way tie they all get a gold medal, however if you're 4th you don't get a medal at all. Method #1 suggest you should get a silver in this circumstance which doesn't seem right. But will it still be for the first 100 places, or is this extended now?

I do like #3 too but i think the dynamic nature of the formula will probably get out of hand...

DavidKing

I think that "3" is the most fair method, perhaps someone has already obtained better time, just he has not submitted yet.

Kristopher

#1 20.00 (100 pts)
#2 21.00 (99 pts)
#3 22.00 (98 pts)

Assume the rankings for a particular level are like that (^) and #1 is put into effect. Then assume that the Silver Medalist gets a faster time and becomes tied with the Gold Medalist:

#1 20.00 (100 pts)
#2 20.00 (100 pts)
#3 22.00 (99 pts)

The Bronze Medalist, and everyone below him, would gain a point from this because of method #1. That's why method #2 is better.
Look at Nik's position in the Bottle level. He is currently #51, which means he gets 50 points for Bottle. But method #1 would deemphasive Bottle because of the ties, and he would get 95 points with method #1. If everyone in the Top 50 gets over 90 points for Bottle, most players could only lose 10 points for having a bad time. It would give Bottle less effect than the other levels. Maybe you want that, but I see it as an odd side-effect to method #1. If Bottle has three medalists like the other levels, shouldn't it be as important for the Time Trial?

Method #3: I don't prefer this over the current ranking. The current ranking gives you less points each time someone beats you on a level. That is how it should be, with your ranking based upon how many people can beat it. Method #3 is based upon the Gold Medalist's time. So when Roel found the secret to Underground, everyone else would have lost many points, possibly up to 90 points for some players. It might have been good if Roel had 90 points more than the Silver Medalist, because he was 10 seconds faster, but then everyone is forced to do the trick themselves. It would encourage Gold Medalists to get even faster times, but would put have awkward emphasis on Cave Dweller. _Alex_ is ranked as #10, but his time is closer to #35 (44 second difference) than to #1 (45 second difference). Also, you didn't mention how it would rank people who haven't completed this level. If a person knows they will get last play even when they complete it, then there would be no reason for them to complete the level. Although method #3 might be okay for Bottle.

Good work on the Distribution Histogram, BTW.

So I support method #2. But one change: instead of giving points for fast times, give points for slow times. Then the person with the fewest points wins. If a person doesn't complete a level, then they get one point more than the person who came in last place. It would remove the problem where #101 and #1001 get the same points. It would put much emphasis on the Standard Levels, especially the first five, where you could have an extra 1000 points for a bad time. But I think that is good; if a lot of people play the level, then it should be more important in the Time Trial. The only problem is for people such as Vega. With this new method I am proposing (method 4), a person who plays the first ten levels somewhat well would be ranked higher than Vega. A benefit to this though: players like Vega, Oliver, and Tabs might start playing all of the levels, instead of just trying to get medals. Method #4 would also rank people with the demo version even lower because they have only played 5 levels. It might be better this way.

BTW, I would have posted earlier, but I was removing a virus and replacing my WINNT\SYSTEM32\CONFIG\SYSTEM file.

See the new scoring system

Hall of Fame
all level points are calculated using the modified formula

by doing this, any number of players will always fit in 100..1 points range. this change does not influence the ranking for players having all levels in top 50 (because for first 50 players 1 rank = 1 point, as before). below #50 one rank is worth half point and so on. as a result, you will almost always earn much more than 1 point for completing a level (no matter what time) and you get rewarded for improving your position even if you not in top 100.

"bottle" and "shelf" are calculated using "method #1" in this experimental system (equal points for equal times) and "bottle" is not excluded from the total score.

kris: the idea of "giving points for slow times" is elegant (no artificial limit of 100 points/ranks) but it makes difficult levels "worthless" compared to bottles or standard levels 1-5.
and from the technical perspective i don't like the fact that each new player would change everyone's points

(the numer 99 in the picture is incorrect, should be 100)

Cirenco

Creo que al final nunca se conseguirá dejar contentos a todos, pero independientemente del método al fianl siempre el mejor se abrirá paso. Ojalá en el futuro evitemos los posibles "bottle" porque son un problema.

Kristopher

My method #4 puts emphasis on completing the hard levels and mastering the easy levels, instead of putting equal emphasis on all of the levels. It might be good; maybe not.

mike flips

its good for newbies but it ruins the whole point of the competition, which is to master every level to become #1

i think the experimental method now is perfect, it gives lower ranked players the points they deserve for completing the levels and it allows new players who tie old scores to get the points they deserve

Kristopher

Then use the new method, if the "EXperiment" is over.

Vega

I know there has been a lot of experimenting lately with the way ranking is scored, but I can't help but notice that there may be a missing element. Take Cirenco's ranking for example. It appears that his low ranking in "New Deal" is completely throwing off his overall ranking. He has more medals than anyone else (not to mention the fact that he completely annihilates everyone else in gold & silver) and yet his ranking doesn't represent that. Somehow he should be ranked #1 undoubtedly at this point. But players who submit decent times for all levels will always have a higher ranking than players who have truly excellent times in most levels. This seems to sort of penalize players who don?t keep up with every single level that comes out and allows mediocre players to be ranked above incredible players just because they submitted ok times for all levels. And since we don?t want mediocre players ranked as the Champions of BOD, to balance this out a little, I propose an idea that will put a little more weight on the medals but will not impede on the integrity of the overall ranking of all levels. What about making Bonus Points for Gold, Silver and Bronze. Since levels are currently valued at 100 points max (gold), what about awarding 20 BP for each gold, 15 BP for each silver, and 10 BP for each bronze. It doesn?t seem correct that someone could hypothetically be able to have no medals at all, be ranked #5 in every level, and be the champion when someone like Cirenco and DavidKing have 50 medals or more.

Something to keep in mind is that this type of idea will not lop side the standings. If a player earns 750 BP for medals, yet doesn?t submit any times for ?New Deal?, any other player can offset those 750 points by merely ranking #50 in all 15 levels of ?New Deal? (that?s simply 50 points x 15 levels, which of course is an approximation bcuz of ties but you get the idea). Take for example Tabs? ranking. His BP would total 490 at his current ranking. This would push his total points up from 6252 to 6742, which moves him from rank 39 to 36, an increase of only 3! In Cirenco?s case, he would qualify for 935 BP, which push him from 11,282 to 12,217. Even without figuring in everyone else?s BP, it still wouldn?t make him the top ranking player. Maybe this idea has already been kicked around, but I haven?t been involved at all in the previous discussions of how rankings work up until now. Maybe there is already something in place like this that I just don?t know about, but if there is, it doesn?t seem to be very effective. Anyway it?s worth some serious thought. Let me know what you think......

Vega

Just another quick point that I just thought of - - - if a new level pack with 20 levels comes out and one player gets gold in 18 levels and doesn't submit for the last 2 levels, he would earn 1800 points for that pack. If another player ranks as #10 in every single level, he also earns approx 1800 points (90 x 20) and the 2 players would be ranked the same for that pack. That's why that ranking system favors quantity of play more than quality a bit too much.

mike flips

To be the champion you need to be good at everything, vega.
If i am ranked 4 in every level then i deserve to be #1 overall, regardless of how many medals other people have. That is what the golden club is for, for people with the medals (like you and Tabs) who are good at lotsa levels but havent played them all.

However, you know i dont like the golden club

. I feel it should be the medal club and each medal should take into account some formula to do with active players or something... but nonetheless...

If you wanna be rank #1 submit all level times and be as good as possible at every level, thats why its call TIME TRIAL CHAMPION because they are the best in all time trial levels.

Cirenco should not be number 1 right now because he hasnt played a whole levelpack yet, even if his BP would take him to #1 he doesnt deserve it because he isnt good at 15 whole levels, when he submits he will go to #1 because then he will become the overall best...

to me it makes sense

Vega

sorry to say that I couldn't disagree more. There will always be new levels that come out (and quite a large amount I might add) It doesn't make sense to have the amount f levels someone plays rank over the excellence someone has already acheived but wants to retire from the game or take a break. This would literally mean that if DK, Cirenco retired now, their standings would tank immediately and never return, as every new level that comes out, they lose ground at an outrageous rate. One level not played wipes out a hundred points ahead of another player. Medals should be recognized! 20 golds and one zero should not equal 20 #5's and 1 gold! That's ridiculous.....if I may say so myself!

Kristopher

Vega, if everyone played just to get some Golds, they would take advantage of the scoring. I have not played the New Deal, much so Shane is ranked much better than I am. But I have more Golds in the new pack than he does. I am overall a better player (so far), with more Golds, but he should be ranked higher in the New Deal because he has played it more.