so you think you like stats ?

Another nerd idea, please ignore.

I noticed that i can get some very good (or very bad) ratio against some players within my rating range. Or an unlikely 50% win against some player 500 points above me and vice versa.

Instead of the 1 0 1 0 1 0 1 ... table down the game with the total, it'd be interesting to have also a 63% / 37% column and the theorical p% / (100-p)% .

The very glicko system is all about the winning probability you have. Let's say you are 100 point below an opponent, you should have a 40% /60% ratio but you are only 35% / 65 % so something is wrong with your opening, your style, your connection, or whatever ...


As said before, i'm not 20 yet, i don't shave and wear nerdy fat lens glasses. I'm probably not allowed to vote in most states so my my opinion doesn't count except for the high school poetry contest.

I don't understand this feedback. Consider that (ideally) software is developed in this manner:
http://en.wikipedia.org/wiki/Waterfall_model#Model

1. Identify requirements
2. Analysis (data modeling)
3. Functional design
4. Coding
5. Testing
6. Deployment & support

Often those providing feedback skip straight to design or coding without explaining new requirements or even doing proper data modeling. This ambiguity can lead to XY problems:
http://xyproblem.info/

So please spend less time self-deprecating and more time clearly expressing your ideas. Thanks.

I don't think that glicko-2 involves the probabilities that you're talking about

Even if it did, this wouldn't be anywhere near useful until you start playing 30+ games against each opponent

I'm not depreciating myself. In real life, i'm a 190 pounds, 37 years with a manly beard and a low voice.
Stop please stop implying i have a xy chromosom problem, my treatment worked fine, until very recently. And i have no clue how you found it out before my BFs and my chihuahua did.

Just saying my idea is not a priority.

Yes, it would take some games before being relevant, i agree, and it would also need to take in account the history of players.
And yes you 're right, glicko has nothing to do with this, my bad.

But wouldn't it be nice to put in perspective a result. You beat me 70% of the time while it should really be 75%.

I agree it might be interesting. But honestly I'm not sure we would really use such information. Right now I have positive records against people much higher-rated than me, and even then I don't especially look into it to understand why. And that's an extreme case, most of the discrepancies reported by your indicator would be smaller, so we would be even less interested in investigating them.

And please try to be fair about this poetry contest.

#4 deprecate, not depreciate:
dep·re·cate (v.): express disapproval of

as in "Another nerd idea, please ignore." where you express disapproval of your own idea while passive-aggressively sharing it...

What confuses me is your comment "it should really be 75%." Is that 75% number based on the players' ratings, and/or based on something more complicated (such as Bayesian inference -- that is, looking at past results between the players)?

A bad guess from : "Déprécier" en français.
Pardon my french . Who is being passive-aggressive Mr Webster?

No, i'm not talking Bayesian approach. It's totally about the players ratings.

If you look at their history, all you do is to take their win / loss history and divide it by the total. And that is the first column i suggest.

I'll make up the next numbers for the sake of example.

I start a game against a player 200 pts above me.
If i win, i get 15 pts.
If he wins, he gets 5 pts.
The rating system suggests that the odds are of 25 to 75%. (My second column)That what i meant by "it should really be 75%."

A difference of n points implies a theoretical win ratio of p / 100-p %. That is what elo is all about.

That's based on the assumption that ratings are not too volatile, which makes my case even more frail.

And some players do have a long love history together.
fr.lichess.org/4UEdaEkp#7

Anyway, i'll just vote for Mary Summers, she has the most flamboyant bustier. Who cares how teen-agers trancend their inner turmoil ?

My answer on the off-topic things here en.lichess.org/forum/off-topic-discussion/passive-aggression-and-self-depreciation#1

"If i win, i get 15 pts.
If he wins, he gets 5 pts.
The rating system suggests that the odds are of 25 to 75%"

This is wrong.