Elo Rating System: Explaining Statistic Details

  • Context: Undergrad 
  • Thread starter Thread starter luxxio
  • Start date Start date
  • Tags Tags
    Rating System
Click For Summary

Discussion Overview

The discussion centers around the statistical details of the Elo rating system, specifically focusing on the rationale behind comparing the expected value with a player's actual score in games. Participants explore the underlying assumptions and implications of this comparison.

Discussion Character

  • Exploratory, Technical explanation, Debate/contested

Main Points Raised

  • One participant seeks clarification on why the expected value represents the score of a game within the Elo rating system.
  • Another participant notes that Elo's assumptions are based on normal distribution, referencing external material for further details.
  • A participant reiterates their familiarity with the reference but emphasizes confusion regarding the logic of comparing expected scores with actual scores.
  • Some participants suggest that a score exceeding the expected performance indicates improvement, which should lead to a higher rank.
  • There is a repeated emphasis on the distinction between expected scores as probabilities of winning and drawing, questioning the rationale for this comparison with actual scores.
  • One participant simply states "heuristic," possibly indicating a method or approach related to the discussion.

Areas of Agreement / Disagreement

Participants express differing levels of understanding regarding the comparison of expected and actual scores, indicating that the discussion remains unresolved with multiple viewpoints presented.

Contextual Notes

Participants have not fully clarified the assumptions behind the expected value calculations or the implications of comparing these values with actual scores.

luxxio
Messages
44
Reaction score
0
who can explain me the statistical details under the elo rating system? in particular why the expeted value rapresent the score of the game. thanx
 
Physics news on Phys.org
zli034 said:
Good to know, Elo's assumption is based on normal distribution as well.
Details here:
http://en.wikipedia.org/wiki/Elo_rating_system

i know this reference. the question is why the expected value is compared with the score of a player. really i can't understand the logic behind this.
 
if I have a score higher than the expectation of my performance, you can infer I'm doing better than my early plays. then I should gain higher rank.
 
zli034 said:
if I have a score higher than the expectation of my performance, you can infer I'm doing better than my early plays. then I should gain higher rank.

yes i know this tale. the question is a little different. the expected score is the probability of win and half the probability of draw. and so question is why this probability is compared with the actual score?
 
heuristic
 

Similar threads

High School Elo, Glicko, and Truskill rating systems are most probably wrong
  • · Replies 21 ·
Replies
21
Views
4K
Undergrad Help creating a ranking system
  • · Replies 2 ·
Replies
2
Views
2K
Undergrad Please Explain (actually explain) The Monty Hall Problem
  • · Replies 131 ·
5
Replies
131
Views
10K
Graduate How Does Scoring Simplify Chess Rating Systems?
  • · Replies 2 ·
Replies
2
Views
2K
[Math/Philosophy]: Does ELO hell exist?
  • · Replies 1 ·
Replies
1
Views
3K
Graduate ELO chess ranking system applied incorrectly in video games
  • · Replies 28 ·
Replies
28
Views
13K
Graduate Proof of thermodynamic stability condition
  • · Replies 4 ·
Replies
4
Views
3K
Graduate Understand Beta-Binomial Model for Win/Loss Rating Systems
  • · Replies 1 ·
Replies
1
Views
3K
Undergrad Meaning of average pressure in statistical mechanics
  • · Replies 1 ·
Replies
1
Views
2K
Graduate Statistical theory/distribution for tuples/vectors of LLR values
  • · Replies 15 ·
Replies
15
Views
4K