Life is short, I don't want to spend years to get back to that beautiful yellow :D
# | User | Rating |
---|---|---|
1 | tourist | 4009 |
2 | jiangly | 3823 |
3 | Benq | 3738 |
4 | Radewoosh | 3633 |
5 | jqdai0815 | 3620 |
6 | orzdevinwang | 3529 |
7 | ecnerwala | 3446 |
8 | Um_nik | 3396 |
9 | ksun48 | 3390 |
10 | gamegame | 3386 |
# | User | Contrib. |
---|---|---|
1 | cry | 167 |
2 | Um_nik | 163 |
3 | maomao90 | 162 |
3 | atcoder_official | 162 |
5 | adamant | 159 |
6 | -is-this-fft- | 158 |
7 | awoo | 157 |
8 | TheScrasse | 154 |
9 | Dominater069 | 153 |
9 | nor | 153 |
Life is short, I don't want to spend years to get back to that beautiful yellow :D
Name |
---|
I agree, I feel that new rating system is just multiplying the rating change in old system by 1/2 (so if you would gain +100 in old rating you will gain +50 in new one, same if you would lose 100 points then in new one you will lose 50)
furthermore ,the new system makes div1 contest has very low number of contestants which requires from you to get better rank in order to have your rating increased
I really don't mind raising the limit for div1 but the change rates really bother me.
what I meant that raising the limit for div1 indirectly affect the change rates of ratings so it should bother you too :P
"furthermore ,the new system makes div1 contest has very low number of contestants which requires from you to get better rank in order to have your rating increased" — such a bullshit
such a good argument! you completely convinced me :)
I just encouraged you to think a bit more on your statement ;).
If you will be more familiar with concept of calculating rating you will know that there is kinda monotonic bijection between your rating and your expected performance. If you perform better then your expected performance, you go up, if not, then you go down. Taking expected value, this is irrelevant to who competes with you.
I am aware of everything you said except "Taking expected value, this is irrelevant to who competes with you." (I'm assuming you mean "expected rank" by "expected value"
how can it possible? if contestants who competes with me were all red is the same if they were all purple?
No, I didn't mean expected rank by any means. I meant expected value of your new rating. Your questions prove that you really don't understand how rating works. Losing with tourist has negligible influence on your rating while losing with worse should have a notable influence.
I know that too , my questions are canceled since my assumption was incorrect
OK, yeah, if you thought that I meant expected rank then your questions are justified :P. However on second hand, you should be aware that talking about expected rank has little relevance to what we discussed earlier :).
User wyxourlove before last contest have 1900 points. It's last place for expected performance. 321 place from 446. Therefore better than 125 other participants, but -11 as result.
"It's last place for expected performance." — no, it's not
By what you are suggesting, tourist should always have expected rank 1. So how does his rating ever increase?
Expected rank doesn't work that way. Depending of difference between the rating of any 2 users, the system decides the "probability of one user getting better rank than the other". Hence wyxourlove's expected rank was towards the lower half of the ranklist, but not the last.
Also comparing (sum of 3 highest rating gains)/(sum of 3 highest rating losses) for last 3 rounds (new system) and 3 rounds before that we get 1243/775 ~ 1.60 now, but 2180/1080 ~ 2.02 before. I think that also means getting inconsistent results (some very good, some very bad) will have worse average rating than getting consistent mediocre results compared to before. Which also means safer strategies (solve AB or ABC in that order, as fast as possible) got buffed compared to not-so-safe strategies (like reading and thinking about all problems after solving A, trying harder ones looking for something you've seen before). Not sure if that's a step in the right direction. What are your opinions about that?
Agreed. Look at the contest standings around my rank. Everyone on page(except 4-5) are blue. Yet, because I was green before, I only get +50, and not even get cyan.
A little unfair.
Totally agree. Suppose someone lost 150 points in last two contests before the 'revolution'. How long will it take him to grab the points back now? Man, we don't have so much life left..
It's possible to get 70+ even if you're orange (I got) so I think 2 or 3 contests are enough pretty much for everyone (maybe except top reds)
I lost about as much in the last 3 contests before the Revolution (net loss). I gained it back in the first 3 contests after it. Simply put: if you're good enough to have some rating, you'll get it. And failing once is less of an issue here as well.
I had 5 consecutive bad contest and them revolution occurred . :/
I don't get why so little changes. Getting out of purple seems very hard.
This new rating calculation is discouraging. It's need to work hard to get some points, it almost makes no difference.
It depends on the person. I think that the new system is more motivating.
Just before revolution there was a contest where I got 2708 rating before it, took 75th place and lost 75 points. In today's contest ecnerwal got 101st place and fell from 2751 by 62 points. So if you fell by a lot you really had to deserve it, I don't see a problem in "if I fell a lot I had to regain a lot" given that falling a lot is hard. It looks like it will make everything more stable and we won't have events like one I will describe. In #270 round (which was joint with Div2) A-D were very easy, E-G were significantly harder. I did A-D, but failed D because of a slight TLE (right complexity, just a bit too high constant factor, but it passed pretests) and because of that I fell to place 700-800 and in terms of rating it meant for me -130 instead of ~+50, which is 180 in total because of a really not that important issue. Variance was sometimes too high, now probably your rating will not be based in 80% on just 3 last contests or something.
don't solve problems for rating
I hope new rating have something like accelerating, if you are constantly doing good, your rating changes will get better and better (similar to Topcoder, I guess)
I dreamed of becoming yellow. Now I dream of becoming blue.
This new color is cool though.
0ld yellow is now purple, so you should dream of becoming purple :P
It became harder to win Div.2 contests after 0ld purple became blue.
.
.
But yeah no argument, you're right...
Hate that color...
And it's kind of TOO hard to increase your rating to Div 1 lvl.
Why? Realy Why? Why does Mike change the formula?
Some of my friends and me are planing to get candidate master(Div1) and observe a contest even we work on question but now the formula changed and we must to try more. beacause getting raise with this formula is harder than past.
The problem pre-revolution was that you could get into Div1 without really being able to solve Div1 problems. Someone like myself in the 1700 range ended up bouncing between divisions and participating less. The bar needed to be higher.
Exactly! I feel like the same way. I also think that in the new system it's less probable for someone to get lucky and have a higher rating than their actual skill or vice versa.
Agree with you man. I've been to that place where you aren't exactly a part of any of the divisions. You give a div2 contest and you're a div1. But you perform terrible in a div 1 contest and come straight back to div2. Becomes frustrating after some time.
Mybe Mike wanted to raise the programming ability of every body and think that changing formula and color is the best way to do this. I hope to it's not right. ;)
In my opinion it would have been better if they kept everything as it is, only made the yellow range bigger (raised the red boundary). Because the yellow is now very hard to get to, and its range is really small.