WasylF's blog

By WasylF, history, 2 months ago, In English

Hello all!

I have some news to share about CF-Predictor (a service that allows you to view rating changes before official results). This extension was launched almost 6 years ago. Number of users has grwon from a couple hundreds to nearly 50k! I'm very proud that this service is used by so many people!

One of the cool features of mine extension is that it is friendly to codeforces API :) All the request to codeforces are sent from just one backend server. Therefore, users don't overload codeforces with extra API calls.

I was using Heroku as a cloud provider to run my servers. Unfortunately, they are closing their free accounts on Nov 28th. So, I decided to take down my servers on heroku. Thus, starting from 2022-11-27 cf-predictor-frontend.herokuapp.com and cf-predictor.herokuapp.com will not work.

On a bright side, I though that upsetting 50k people isn't a good thing. Therefore, I'm renting a VPS and you can continue to use CF-Predictor via cf-predictor.wasylf.xyz! I already updated extentions for Chrome, Mozilla and Opera to use the new backends. However only Mozilla has accepted the new version so far. I hope the changes will be reviewed quickly and you all will be able to use new version soon!

Good luck to all of you and have a high rating!

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +398
  • Vote: I do not like it

By WasylF, history, 11 months ago, In English

Hello everyone!

I'm the author and maintainer of CF-Predictor. Unfortunately, CF-Predictor didn't work well today during the beginning of round #773. Its backends were OOMing... This is very strange because it was using way less memory than the free limits permit.

As mitigation, I've restricted rating calculation to just round773, so sorry to everyone who participates in Kotlin Heroes: Practice 9. Predictions will start working again once round #773 is finished.

Thanks, everyone for using CF-Predictor! I'll look into the memory usage issue later this week, hope it will not be too complicated to fix :)

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +39
  • Vote: I do not like it

By WasylF, history, 5 years ago, translation, In English

Hi there! Rules of div 2 only contests were updated recently (allowing to participate purple in div 2 only rounds). I'm curious does this have an influence on the rating system. So. I've split all participants of this round by divisions and calculated rating change for them independently. You could find this results via following links: div1 only, div2 only. .

Let's define average rating change in the division as a sum of all rating changes of each contestant from this division divided by the number of participants in this division.

Division Average rating change
div. 1 only -8.09
div. 2 only -10.68

Let's calculate average rating change for the official result (rating calculates without separating by divisions)

Division Average rating change
div. 1 +3.86
div. 2 -12.21
both -10.81

As we can see, in combined round average purple contestants gain almost 12 additional rating points. In my opinion, this is quite unfair. Your thoughts?

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +43
  • Vote: I do not like it

By WasylF, history, 5 years ago, translation, In English

I was wondering if codeforces community might be able to give me some advice.

I'm doing the second year of my master's degree and need to write diploma project. I have created CF-Predictor for the previous course project. I would be really happy if I could exceed it for more scientific direction. Unfortunately, I have no idea what should it be. Could you please give me any advice for developing CF-Predictor in the more scientific way. May you know any other service (not too easy & not too hard for diploma project) that could make codeforces better? :)

Thank in advance

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +96
  • Vote: I do not like it

By WasylF, history, 6 years ago, In English
UPD 2 Sep 2018

Hi guys! As I mentioned on previous contest I'm working on tool that predicts rating changes. I'm happy to present it now!

A huge amount of your nerve cells die every time when you wait for a rating update on Codeforces. Stop this! From now you could use this service, it calculates approximate rating changes for every contestant.

The most interested thing for you is extension. It partly modifies the contest standings page and shows approximate rating changes for every contestant. It is available for three browsers:

Extension in work:

Also you could find more detailed information (seed, rank, expected delta, etc.) here.

A project still in beta, so predictions are not very accurate. Average mistake around 5 points, but for the contestants at the back of standings it could be greater up to a few hundreds.

Tech details

I'd like to thank Rubanenko and all other members of NBHEXT developers team for your shared sources and MikeMirzayanov for the great Codeforces platforms with shared API & ratings formulas.

AWESOME UPD

Prediction for todays contest (cf #399) is absolutely matching real rating changes! Thanks for riadwaw! He was close to find my bug in rating calculation:)

UPD

Thanking KieranHorgan now CF-Predictor has a new design. Check it out:)

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +187
  • Vote: I do not like it

By WasylF, history, 7 years ago, translation, In English

Hi guys!

First of all, sorry for my poor english and short translation:)

Now everyone could access rating formulas, and there is some not obvious mathematic formulas. So, I decided to check probability formula(probability that the i-th participant has better result than the j-th participant):

There was a lot of rating codeforces rounds, so it's easy to calculate this value by existing results. I choose cf rounds №200 — №350 (separately for each division). To solve this problem I wrote java program (sources). After getting results I copy/paste it to Excel and get plots:

for first division

for second division

Looks good, but lets try found something better:)

Unfortunately this formula doesn't look scientific, so I a little bit change it:

Now, lets build plots again

for first division

for second division

Now it's closer:)

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +269
  • Vote: I do not like it

By WasylF, 8 years ago, translation, In English

I've finally received t-shirt from Russian code cup 2014! It seems that this is impossible (because of political situation between Russian and Ukraine) but mail.ru did it!

Full text and comments »

 
 
 
 
  • Vote: I like it
  • +107
  • Vote: I do not like it

By WasylF, 8 years ago, translation, In English
 
 
 
 
  • Vote: I like it
  • +42
  • Vote: I do not like it