Dimsas's blog

On what criteria is the rating of a problem calculated?
By Dimsas, history, 8 hours ago, In English

The difficulty level of some problems seems to be well below or above what the rating indicates, so I was curious as to what criteria problem ratings are calculated upon.

  • Vote: I like it
  • +9
  • Vote: I do not like it

»
8 hours ago, # |
  Vote: I like it +1 Vote: I do not like it

I think it has something to do with the amount of submissions the problem records within a contest,you'd notice a lot of div3 D's and E's are rated 1400-1500 but are considerably easier than a div2C or a div1A of the same rating.

→ Reply
  • »
    »
    6 hours ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Yea that does make sense

    → Reply
»
8 hours ago, # |
  Vote: I like it 0 Vote: I do not like it

Actual true ratings nowadays cannot be evaluated due to massive cheating on contests. Since problem ratings are calculated based on the number of correct submission during the contest.

→ Reply
»
8 hours ago, # |
← Rev. 4   Vote: I like it 0 Vote: I do not like it

i think it has to do with the amount and the rating of people solved it during the contest, also div2 problems that get a rating of 1400 — 1500, sometimes are harder than div3 problems getting rating of 1500 — 1700

→ Reply
»
7 hours ago, # |
  Vote: I like it +1 Vote: I do not like it

A problem gets a rating of X in a contest if 50% of participants of rating X solved it during the contest

→ Reply
  • »
    »
    6 hours ago, # ^ |
      Vote: I like it +2 Vote: I do not like it

    Ah I see Do you have a source for this??

    → Reply
    • »
      »
      »
      5 hours ago, # ^ |
        Vote: I like it +1 Vote: I do not like it

      As far as i know Codeforces uses the ELO rating system

      To calculate the performance of a problem we need to see how many participants of which rating it defeated during a contest(participants that failed to solve it), then you use the ELO performance formula to calculate its rating.

      Now we can arrive to a Winning probability table based on ELO difference where you can see that if the rating difference gets close to 0, the winning probability gets close to 50%, thus if a problem defeated 50% of participants of rating X the problem has a rating of X.

      → Reply