I think there are several coders who are very good at TopCoder, but are very poor at Codeforces, or someone else who are very good at Codeforces, but are very poor at TopCoder.
In my case, I'm good at TopCoder (I like some kind of combinatorics or enumlation in TC), so now I am a redcoder in it. But I can't solve Codeforces problems! Sometimes I can't solve even problem-A!!
I'm interested about coders who are very good at the one contest and are very poor at the other contest.
My Distance between TopCoder rating and Codeforces rating ( = abs(_TCRating_ — CFRating) ) is 513 now(TC:2273 CF:1760). If your DBTRACR is higher than mine, please show me how do you feel! And who has the most DBTRACR ?? I am wondering!!
Ask Petr — his difference is 863 (3619 on TC, 2756 on CF).
Difference between user:rng_58 and "[[rng_58]]"(TopCoder, Admin) is 2815! I never see the difference more than him.
this is his profile as a contestant. His rating is 3468. Now the difference won't be that much.
897, few contests ago it was 1026.
CF(1904) — TC(1356) => 548, my two ratings are equivalent respect to the distance from start of Division 1 in both sites
I guess TopCoder requires a lot of math skills, but CodeForces requires more programming skills, that's why sometimes we have this difference.
U mean viceversa?
Nope:) I mean exactly what I've said. Don't you agree?
CF 1702-TC 1445
Codeforces seems more generalist, it includes graph theory, geometric, maths, combinatorics problems... Topcoder has often mathematics and dynamic programming only I think.
Topcoder — 1710 Codeforces — 1929
Not a very big difference.
wow!! Your TC rating graph is impressive, from Green to RED .. :) inspiring.. :)
I think percentile comparisons are better. The rating formulas are different, so 1900 in Codeforces may not be equivalent to 1900 in TC. Codeforces percentile is hard to find though.
If we go by percentiles. I have a better ranking than 96% coders in Codeforces and 91% coders in TopCoder. Although most of the difference might be because I just had a terrible match this morning in TopCoder.
Edit: For example Petr has 99.97 percentile in TopCoder and 99.98 in Codeforces, hardly a difference. Even though the ratings are quite different.
Percentile comparisons may not be that great either, perhaps the overall competitiveness is better in Codeforces than in TopCoder or vice versa.
Agree! Another example, rank 100 in Topcoder differs about 250 ratings from that in Codeforces (2575 versus 2324).
Still, a small factor would affect percentile comparisons is that Topcoder doesn't take recently inactive participants (no participation for the last 6 months) into account. At a result, fluctuations in ranks often occur when veterans come back for TCO.
One can not just simply calculate rating difference in order to get difference in performance. But I've already done least squares analysis on common handles in TC and CF and posted in my blog http://sharpc.livejournal.com/72884.html (in Russian): cf = 0.39491 * (tc − 1229.2) + 1567.5
And how much correlation coefficient is?
I calculated my expected CF rating by SharpC 's way. My expected rating on Codeforces is 1780, but actual rating is 2084. They are quite different, I think.
And I increased TC's rating in last SRM and decreased CF's rating in last round, even the difference is small to me.
My difference is 1816 (CF) — 1231 (TC) = 585. It is really impressing that our both differences are so big but you have better topcoder and I have better CF.
My Codeforces (Expert) — Topcoder (Grey) = 651
As for myself...
Current ratings difference: 461
Using percentiles to measure cause some problem on Codeforces: a lot of users have 1 match and ~1750 rating :D
Maybe we should only count those who has participated in at least 1 Div.1 contest, or regular Div.2 participants.
Rating doesn't matter