Bayesian skill rating FAQ

Anwers to common questions about the Bayesian skill rating implemented in Legacy mod.


What is the Skill rating system?
How does this compare to the popular Elo?
What does the rating value represent?
New players have a rating of zero. Does this mean the system believe they are noob?
Why not using kills or in-game actions to rate players?
Why is TrueSkill adapted to public servers?
What is the map bias?
What is taken into account while calculating SR?

What if...

What if players change team to be on the winning team instead?
What happens if I am playing only with noobs, which make me lose evertime. I cannot get a strong rating, despite being better than them!
What happens if players switch team to the winning team before the end of the map?
What happens if players disconnect from the current match?
What happens if players disconnect and reconnect to the same match?
How does one cheat the system?

Tell me more...

The win percentage indicated team A would win, but team actually B won. Explain!
I’m playing alone on an empty server, and the win percentage indicate 50%. Explain!
How does one make advantage of the system to boost his rating?
The Scoreboard has most strong players in one team, yet the scoreboard set the winning percentage to the weak team. Explain!
How long does it take for the rating to converge toward an accurate value?
Why use shuffle by SR instead of XP shuffle?
The maps are always favoring a side. What about this?
How is the map bias computed?
What influence do bots have on map bias (is this good for bias calculation)?

Advanced questions

Why using TrueSKill and not using the system implemented in ETPub?
What are the differences between TrueSkill and the system implemented in Legacy mod?
What are the actual limitation of the system?
Why aren’t you calling this system "True Skill"?
Will this system be used for match making?
Who is Josh Menke?
Can you give me resource to learn more about the system?