Dr.Bombay - there's no one value that works for all variants to "get them in range." huffy plays mostly 10x6 and has an average win time of under a minute so their already pretty high Elo gets a 4x boost.
I looked at the actual average play time regardless of variant and it looks like it's more like 2.71 minutes so I changed that calculation. It's still pretty fictitious though. I just thought I'd give a taste of what a time-dependent Elo might look like.
MrFixit
back to the topic.... would this save penalty also apply to network timeouts?
"would this save penalty also apply to network timeouts"
Yes. I lost a lot of chess games because of network disconnection where I was actually winning.
And my average time here in this site has a lot of 15+ minute games in it, which I would win in 1 to 3 minutes.
but what about the elo penalty... your game reloads after a loss of connection regardless of how long you wait to reconnect
If saving through UI only is not taken into account and expected performance part is not implemented it would still be as is.
so you're saying that a pause due to a connection loss is not penalized
player rights is an issue... what i do while my hands are not on the keyboard and/or mouse cannot and should not be interpreted by freecell.net as anything other than that
maybe calling it a 'save' surcharge would be a better approach
100% loss of win elo points is unfair
maybe meet in the middle at 50%
I like simple things. If you click "Save", you will not get any Elo points if you win.
If Denny were to give the real elapsed time in daily data files, I can implement it myself in my site.
"I like simple things. If you click "Save", you will not get any Elo points if you win."
50% is just as simple
And yes i would like to see it to be able to be recalculated from the logs
"50% is just as simple"
I believe setting something to zero is simpler.
What is the frequency of saves? I myself do it sometimes. But I could adjust myself to not using it at all. And if I were to use it, I would have no problem with getting no points for my win.
"I believe setting something to zero is simpler."
surely you can handle a divide by 2
saves and restores are logged, but not tracked
"surely you can handle a divide by 2"
It's simple because zero times something is zero. That something is anything.
To divide by 2 you need to have something and which is not anyting.
Saved games which are lost, absolutely *must* count towards ELO; otherwise player behavior will become: "I think I've lost this game, I'll hit save come back in and then lose it and it won't affect my ELO"
Likewise a saved game which is won shouldn't count towards ELO, saving destroys the time basis which should have been a parameter requirement for ELO calculation from the get-go.
Hop, that makes no sense... but programmers can deal with that
When a game is saved, is it possible to keep the clock running?
ELO used to show up on the daily scores page , Can we get it back there again ?
Keep the clock running? When closing your tab/browser - assuming everything, including the server, is behaving, the game should be in the same state when you return(?).
When the server is misbehaving, and the client is disconnected, does the clock stop & resume when reconnected? I've never monitored that scenario, but I imagine in that event, time elapsed would be saved along with the other data & timing would recommence in sync with the game restarting.
Help me out here, omnescient...
"Hop, that makes no sense... but programmers can deal with that"
Of course I know it makes no difference practically. I am trying to explain it conceptually.
What I mean is before dividing something by two, you first need to calculate something. You don't need any calculation in "setting to zero" case. So which one is simpler? Conceptually.
That said. This 50% thing would be better to do in all cases. Player Elos do not reflect the reality. Player Elos should not inflate that much. I have an Elo above 2700 and so a 2000-Elo game should be a piece of cake for me. Not the case at all.
So my suggestion is that games get 100% and players get 50% when an Elo exchange is calculated.
"So which one is simpler? Conceptually"
ok. i missed that transition away from reality
"So my suggestion is that games get 100% and players get 50% when an Elo exchange is calculated"
That could work. But to be Elo-ish, if K is 8 then the 50% taken away from the player needs to be given to the game... a customized exchange calculation
"But to be Elo-ish, if K is 8 then the 50% taken away from the player needs to be given to the game... a customized exchange calculation"
But that's just making K equal to 4. I saw chess sites doing it the way I said. For example, in a game, I gained 4 Elo for a win but my opponent lost half of it.
no... if the exchange is player 4 and game 4 ,
using the 50% adjustment, player would get 2 and game would get 6
"using the 50% adjustment, player would get 2 and game would get 6"
How is that? I can't follow you. Currently, when you calculate an exchange, winner takes it and loser gives it.
What I suggest is; say 4 is calculated, if the winner is player, they get 50% of it (2), but the game gives 100% (4) and vice versa.
that doesn't balance. the game gives 4? the player only takes 2. where do the other 2 points go to?
Yes, but are we sure the total is correct and the rating are correct? So be it. An equilibrium would be reached somehow.
ya... i guess.
altho i'd do it a bit different. set K to 4 so the player elo is done normally. then a 200% on the game side. maybe even more so that it approaches its true elo faster on the lesser played variants (then there's already that elo boost thing)
I'm rapidly forming the view that ELO is a nonsense.
I certainly think that the new "Time Adjusted Elo" is nonsense!
But as an intellectual exercise, it certainly is entertaining.
Tony doesn't "do" intellectual exercises. He uses cards and takes notes, so your observation is lost on him, sadly 😕.
Elo ratings are not nonsense, they're a predictive tool based on sound math. There's an abundance of nonsense in this thread that people are calling Elo though. Let 'em have their fun.
I guess this is not strictly Elo, but a variation on that theme. The player is 'active' and the deal is 'passive'. Numbered games do not employ strategies that can be adapted depending on how a particular opponent is playing them.
'Predictive' in this scenario involves a retrospective element - the player doesn't know the probability of winning or losing until the final move (triumphant or capitulatory) releases the adversary's rating.
I have tried psyching out the more vindictive deals, but they are persistently impervious to my attempts at provocation, intimidation, appeasement and pleas for mercy.
My fourth husband used that last line in court describing me. Is your name Quentin?
It may have been once, before I took it too seriously. Is your name Blunt? (No point in asking.)