All site revenue goes to charity
Subject: more Elo history
Date: Wed Jan 26 09:45:05 2022
User: MrFixit
Message:
Thanks to cellmate for the latest Elo history and Hop for the previous history, we now have several months of Elo history for every player. Go to Scores --> individual stats [yourname] Go and then click Elo History on the far right. It should show you a decent graph of your Elo over the last few months. See link below for mine. Now I'm very depressed. MrFixit aka Denny
Link: my Elo history
User: MrFixit
Message:
Thanks to cellmate for the latest Elo history and Hop for the previous history, we now have several months of Elo history for every player. Go to Scores --> individual stats [yourname] Go and then click Elo History on the far right. It should show you a decent graph of your Elo over the last few months. See link below for mine. Now I'm very depressed. MrFixit aka Denny
Link: my Elo history
Date: Wed Jan 26 10:48:54 2022
User: Bertie
Message:
Re depression: I found it a real relief to drop my ELO from 1730 to 1200. High up you waste endless hours being rewarded with 0.005 or thereabouts per game and then one overhasty click gets punished with almost the full 8 points. That's dpressing !
User: Bertie
Message:
Re depression: I found it a real relief to drop my ELO from 1730 to 1200. High up you waste endless hours being rewarded with 0.005 or thereabouts per game and then one overhasty click gets punished with almost the full 8 points. That's dpressing !
Date: Wed Jan 26 11:16:29 2022
User: redberet
Message:
That's the thing about ELO, if you play the way you normally play, then that's what your rating is.
User: redberet
Message:
That's the thing about ELO, if you play the way you normally play, then that's what your rating is.
Date: Wed Jan 26 12:32:30 2022
User: ElGuapo
Message:
Well said, redberet. And to Uberman's point, Bertie, please don't make a habit of deliberately losing games. It skews the game stats and potentially messes up the ratings for the whole game level you're playing. (And for anyone looking at my own history, yes, there are a bunch of deliberate losses there from my work fixing ratings. Please know I never deliberately lose winnable games.)
User: ElGuapo
Message:
Well said, redberet. And to Uberman's point, Bertie, please don't make a habit of deliberately losing games. It skews the game stats and potentially messes up the ratings for the whole game level you're playing. (And for anyone looking at my own history, yes, there are a bunch of deliberate losses there from my work fixing ratings. Please know I never deliberately lose winnable games.)
Date: Wed Jan 26 12:51:08 2022
User: msm98lw
Message:
Elo does mark the hard level 5 games that often trip me up. My elo is about 1730 (95%). Any win usually has a small upward effect on my score but a lose makes a big change. Sigh.
User: msm98lw
Message:
Elo does mark the hard level 5 games that often trip me up. My elo is about 1730 (95%). Any win usually has a small upward effect on my score but a lose makes a big change. Sigh.
Date: Wed Jan 26 12:57:20 2022
User: Crunch
Message:
I normally play streaks on the 10x2 variant. I normally see gains of .01 and occasionally .1, and rarely >.1, for streaks of 1-50. Then I really rake in the ELO points on the level 10 games of .1 +. When I have a really difficult game, I can get .5 or so. In my opinion, the plus side of the levels is very good. On the loss of games, the variation is always a -7.xxx. When the winning percentage of the game is high, the loss is more punitive which is expected. However, for losing low percentage of wins games, the loss is -7.2+. I think that the % wins should be factored into the calculation a little more than just an addition of a range of 7.xxx. If I lose a really difficult game, with a win/loss of 20%, I would expect less of an ELO loss. After all the rating is supposed to reflect the comparative skill of the players. Meanwhile, I have dropped 100 points in ELO ratings in just a week or so, from about 2000 + to closer to 1913 in 12 days. If Bobby Fischer had a bad 2 weeks of chess, I doubt he would drop down in the ratings by 50 players or so, and he would still crunch me. I would like to see some discussion about loss calculation. I think this system can be improved in this area. Perhaps a loss of a 50% game would be -5, a 40%, -4., etc as a rough approximation. I think the -7 is a good number for a loss of a high % win game. Any comments?
User: Crunch
Message:
I normally play streaks on the 10x2 variant. I normally see gains of .01 and occasionally .1, and rarely >.1, for streaks of 1-50. Then I really rake in the ELO points on the level 10 games of .1 +. When I have a really difficult game, I can get .5 or so. In my opinion, the plus side of the levels is very good. On the loss of games, the variation is always a -7.xxx. When the winning percentage of the game is high, the loss is more punitive which is expected. However, for losing low percentage of wins games, the loss is -7.2+. I think that the % wins should be factored into the calculation a little more than just an addition of a range of 7.xxx. If I lose a really difficult game, with a win/loss of 20%, I would expect less of an ELO loss. After all the rating is supposed to reflect the comparative skill of the players. Meanwhile, I have dropped 100 points in ELO ratings in just a week or so, from about 2000 + to closer to 1913 in 12 days. If Bobby Fischer had a bad 2 weeks of chess, I doubt he would drop down in the ratings by 50 players or so, and he would still crunch me. I would like to see some discussion about loss calculation. I think this system can be improved in this area. Perhaps a loss of a 50% game would be -5, a 40%, -4., etc as a rough approximation. I think the -7 is a good number for a loss of a high % win game. Any comments?
Date: Wed Jan 26 15:15:13 2022
User: BuzzClik
Message:
Date: Wed Jan 26 11:16:29 2022 User: redberet Message: "That's the thing about ELO, if you play the way you normally play, then that's what your rating is." That's a great comment. I look at my paltry Elo (Bertie territory) with no particular interest, but it is curious to read the discussion that Elo is supposed to be a weighted evaluation to get a handle on how good a player one might be. redberet's comment suggests that unless you are "gaming the system" to inflate your Elo, it will come to some level based solely on your playing habits. I totally agree; my Elo seems to get a bigger bump for losing a game that's been played but never won than for winning a game that's never been played. Seems backwards. (Please don't explain -- I really don't care why it's true.) I have no illusions: I cannot compete with 99% of the folks who post on this board; so, my sucky Elo seems totally appropriate. But, a fantastic freeceller playing the same games I play would have an Elo similar to mine and nowhere near 2500 or 2000 or 1700 or ....
User: BuzzClik
Message:
Date: Wed Jan 26 11:16:29 2022 User: redberet Message: "That's the thing about ELO, if you play the way you normally play, then that's what your rating is." That's a great comment. I look at my paltry Elo (Bertie territory) with no particular interest, but it is curious to read the discussion that Elo is supposed to be a weighted evaluation to get a handle on how good a player one might be. redberet's comment suggests that unless you are "gaming the system" to inflate your Elo, it will come to some level based solely on your playing habits. I totally agree; my Elo seems to get a bigger bump for losing a game that's been played but never won than for winning a game that's never been played. Seems backwards. (Please don't explain -- I really don't care why it's true.) I have no illusions: I cannot compete with 99% of the folks who post on this board; so, my sucky Elo seems totally appropriate. But, a fantastic freeceller playing the same games I play would have an Elo similar to mine and nowhere near 2500 or 2000 or 1700 or ....
Date: Wed Jan 26 23:07:32 2022
User: sprucegoose
Message:
I deliberately lose games; that's the way I play. I've played that way for more than 20 years. FWIW, the ELO is a biased rating that doesn't really measure a player's skill.
User: sprucegoose
Message:
I deliberately lose games; that's the way I play. I've played that way for more than 20 years. FWIW, the ELO is a biased rating that doesn't really measure a player's skill.
Date: Thu Jan 27 05:34:09 2022
User: TitanicTony
Message:
I used to deliberately lose games, for the WWC & SSC competitions, in order to start out at level 5! Not any more!!!
User: TitanicTony
Message:
I used to deliberately lose games, for the WWC & SSC competitions, in order to start out at level 5! Not any more!!!
Date: Thu Jan 27 09:02:25 2022
User: msm98lw
Message:
sprucegoose, can you explain how ELO is biased? Thanks
User: msm98lw
Message:
sprucegoose, can you explain how ELO is biased? Thanks
Date: Thu Jan 27 09:12:13 2022
User: redberet
Message:
To me streaks are made to be broken. The quickest way there is not sitting on ball. That's also why I don't have many ongoing streaks. When it comes to SSC/WWC, that attitude helps when one's streak comes tumbling down. ATW!
User: redberet
Message:
To me streaks are made to be broken. The quickest way there is not sitting on ball. That's also why I don't have many ongoing streaks. When it comes to SSC/WWC, that attitude helps when one's streak comes tumbling down. ATW!
Date: Thu Jan 27 09:56:20 2022
User: MrFixit
Message:
I dunno spruce, you look pretty good to me. I think it might be biased *towards* you!
User: MrFixit
Message:
I dunno spruce, you look pretty good to me. I think it might be biased *towards* you!
Date: Thu Jan 27 11:26:06 2022
User: bcmoore87
Message:
>I would like to see some discussion about loss calculation. I think this system can be improved in this area. Perhaps a loss of a 50% game would be -5, a 40%, -4., etc as a rough approximation. I think the -7 is a good number for a loss of a high % win game. Any comments? Your comments makes sense if all players are the same rating. observations: Playing against entry-level games: If a person intrinsically win 80% (4 out of 5) of games against a level 1200 game, then each win should raise my score by X and each loss should lower by score by 4X. This is why you generally see big losses and small gains. Playing again very hard games: You will find the opposite. Gains are bigger than losses. (Because your intrinsic win probability is under 50%) The ELO system is really cleverly implements on this site. Very well done!
User: bcmoore87
Message:
>I would like to see some discussion about loss calculation. I think this system can be improved in this area. Perhaps a loss of a 50% game would be -5, a 40%, -4., etc as a rough approximation. I think the -7 is a good number for a loss of a high % win game. Any comments? Your comments makes sense if all players are the same rating. observations: Playing against entry-level games: If a person intrinsically win 80% (4 out of 5) of games against a level 1200 game, then each win should raise my score by X and each loss should lower by score by 4X. This is why you generally see big losses and small gains. Playing again very hard games: You will find the opposite. Gains are bigger than losses. (Because your intrinsic win probability is under 50%) The ELO system is really cleverly implements on this site. Very well done!
Date: Thu Jan 27 12:02:41 2022
User: Kumquat-of-Conciliation
Message:
The Cleverlys are a more-than-competent, semi-novelty, but real-talent pseudo-bluegrass band out of the Ozarks, mostly. Never seen them, but from all accounts they're a great show. Here's a couple more: https://www.youtube.com/watch?v=UoIB7uwDs80 https://www.youtube.com/watch?v=HVTD-FTovRU
Link: The Cleverlys being clever
User: Kumquat-of-Conciliation
Message:
The Cleverlys are a more-than-competent, semi-novelty, but real-talent pseudo-bluegrass band out of the Ozarks, mostly. Never seen them, but from all accounts they're a great show. Here's a couple more: https://www.youtube.com/watch?v=UoIB7uwDs80 https://www.youtube.com/watch?v=HVTD-FTovRU
Link: The Cleverlys being clever
Date: Thu Jan 27 13:23:55 2022
User: BuzzClik
Message:
That would be great discussion material in the proper context. Ko'C, I think this landed in the incorrect thread.
User: BuzzClik
Message:
That would be great discussion material in the proper context. Ko'C, I think this landed in the incorrect thread.
Date: Thu Jan 27 14:23:59 2022
User: outskirts
Message:
Let's see how this works Thanks Denny and crew
Link: https://www.freecell.net/f/c/ps_elo.html?u=outskirts
User: outskirts
Message:
Let's see how this works Thanks Denny and crew
Link: https://www.freecell.net/f/c/ps_elo.html?u=outskirts
Date: Thu Jan 27 16:19:54 2022
User: Kumquat-of-Conciliation
Message:
No, actually, Buzz, it's where I intended it to be. I was tagging along to the awkwardly-constructed use of phrase containing "cleverly" in the previous post. It just made my headbone work that way. ;) Additionally/alternatively, I was tempted to put about 4-5 of theirs in the "really different covers" thread, but couldn't resist the play on that one 'trigger' word here.
User: Kumquat-of-Conciliation
Message:
No, actually, Buzz, it's where I intended it to be. I was tagging along to the awkwardly-constructed use of phrase containing "cleverly" in the previous post. It just made my headbone work that way. ;) Additionally/alternatively, I was tempted to put about 4-5 of theirs in the "really different covers" thread, but couldn't resist the play on that one 'trigger' word here.
Date: Thu Jan 27 16:27:11 2022
User: BuzzClik
Message:
Well, I think the following is an appropriate response to my missing your point:
Link: https://www.freecell.net/f/c/onepic.html?code=6403
User: BuzzClik
Message:
Well, I think the following is an appropriate response to my missing your point:
Link: https://www.freecell.net/f/c/onepic.html?code=6403
Date: Thu Jan 27 16:28:40 2022
User: BuzzClik
Message:
I do wish they would report more digits past the decimal. I'd like to put a fine point on my progress
User: BuzzClik
Message:
I do wish they would report more digits past the decimal. I'd like to put a fine point on my progress
Date: Thu Jan 27 16:41:14 2022
User: Kumquat-of-Conciliation
Message:
Well......................it wasn't *totally* obvious, I guess, and it *surely* was out of place except for that one (key) train-of-thought connection. On rare occasions...........I can be "implementally" clever like that? Or maybe to some, too clever by one? Or something.......
User: Kumquat-of-Conciliation
Message:
Well......................it wasn't *totally* obvious, I guess, and it *surely* was out of place except for that one (key) train-of-thought connection. On rare occasions...........I can be "implementally" clever like that? Or maybe to some, too clever by one? Or something.......
Date: Thu Jan 27 16:42:38 2022
User: Kumquat-of-Conciliation
Message:
........and I share your concern about "significant figures". ROFLOL.
User: Kumquat-of-Conciliation
Message:
........and I share your concern about "significant figures". ROFLOL.
Date: Thu Jan 27 18:37:51 2022
User: roo
Message:
The line needs to be thinner to reflect the accuracy of the science. Oh, hang on ...
User: roo
Message:
The line needs to be thinner to reflect the accuracy of the science. Oh, hang on ...
Date: Tue Feb 1 19:24:55 2022
User: Crunch
Message:
About the ELO Loss Calculation: I've been keeping track of the stats on the games I lose. I play my favorite variant of 10x2. I go for streaks of the day and max streaks. Since ELO, my game has gotten stronger. Thanks! I tend to lose when my eyes get tired and I go past "just one more game". :-) Whether I lose a difficultly level of 958 thru 1622, I lose 7 ELO points plus some number from 0 thru 9. The range I saw was -7.906 to -7.167. with an ELO 0f 1958 to 1980. The range of difficulty calculation only effects the decimal place after the -7. , regardless of ELO or difficulty rating. So if I lose a level 5 game, or a level 10 game, the loss is -7.x. I think "bcmoore87", in_ the thread above, had the right idea of making the -7.x to a -5,-6, or -7. When I win ordinary games, I can gain only about .01 or .001 ELO points. In the streak playing, there is an occasional difficult game with only 37% win where, I can pick up maybe a max of 1.5 points. I am now clawing my way back to the top 50 after dropping to 102 after a few bad days. That's on me. But losing 7.x points whether my ELO is 2000 or 1000, or the difficulty level is 1622 or 958 is not right. Hope this helps Crunch
User: Crunch
Message:
About the ELO Loss Calculation: I've been keeping track of the stats on the games I lose. I play my favorite variant of 10x2. I go for streaks of the day and max streaks. Since ELO, my game has gotten stronger. Thanks! I tend to lose when my eyes get tired and I go past "just one more game". :-) Whether I lose a difficultly level of 958 thru 1622, I lose 7 ELO points plus some number from 0 thru 9. The range I saw was -7.906 to -7.167. with an ELO 0f 1958 to 1980. The range of difficulty calculation only effects the decimal place after the -7. , regardless of ELO or difficulty rating. So if I lose a level 5 game, or a level 10 game, the loss is -7.x. I think "bcmoore87", in_ the thread above, had the right idea of making the -7.x to a -5,-6, or -7. When I win ordinary games, I can gain only about .01 or .001 ELO points. In the streak playing, there is an occasional difficult game with only 37% win where, I can pick up maybe a max of 1.5 points. I am now clawing my way back to the top 50 after dropping to 102 after a few bad days. That's on me. But losing 7.x points whether my ELO is 2000 or 1000, or the difficulty level is 1622 or 958 is not right. Hope this helps Crunch
Date: Wed Feb 2 15:26:56 2022
User: ElGuapo
Message:
Thanks for the feedback, Crunch, it's good to hear how players are experiencing this real-time Elo updates. (And some of us have been around long enough to remember your own scoring system for tournaments, "Crunch points"!) You're seeing all or nothing point exchanges because you're rated so far above even the hardest 10x2s you're playing. Try 9x2 and you'll see more of those in-between rating changes you're looking for.
User: ElGuapo
Message:
Thanks for the feedback, Crunch, it's good to hear how players are experiencing this real-time Elo updates. (And some of us have been around long enough to remember your own scoring system for tournaments, "Crunch points"!) You're seeing all or nothing point exchanges because you're rated so far above even the hardest 10x2s you're playing. Try 9x2 and you'll see more of those in-between rating changes you're looking for.
Date: Wed Feb 2 23:10:39 2022
User: rbw--3
Message:
Back a few days sprucegoose said ELO is biased. I'm pretty sure that has to do with no adjustment for time. Spruce's times are super fast and his real skill isn't reflected in ELO. Same with all the speed players.
User: rbw--3
Message:
Back a few days sprucegoose said ELO is biased. I'm pretty sure that has to do with no adjustment for time. Spruce's times are super fast and his real skill isn't reflected in ELO. Same with all the speed players.
Date: Wed Feb 2 23:58:02 2022
User: Crunch
Message:
Re: ElGuapo You have quite a memory. Back in the day, I was trying to figure out a method for a tournament players rating. I gave up. The ELO system we have today, seems to be quite good based on my daily streak playing and seems to need only a slight tweak for the basic deduction of -7.x for all lost games. Thanks for your suggestion on the 9x2's. I will try them when my current 10x2 streak ends. As far as tournament ratings, I can say that if my daughter plays, (she's in school now and is smartly studying instead of playing), she will always beat me. But, I am still close behind and not the biggest behind! :-) I still enjoy playing against speedsters like yourself and others and hope for the situation that there are 3 or 4 extremely difficult games with lots of rabbit holes, and I get lucky and pick the right one, while you and the other speedsters miss it. I like the exclusion of tournament games from the ELO rating. Also, I like the loss of 0 points for a game that cannot be won. I used to play chess about 40 years ago and quit after making a blunder in a tournament. Even after that long pause of not playing, I was able to play against my grandkids and beat them because of the discipline and challenge of thinking ahead, that freecell requires. No shade on them because they all seem to be smarter and definitely faster. Cheers Crunch
User: Crunch
Message:
Re: ElGuapo You have quite a memory. Back in the day, I was trying to figure out a method for a tournament players rating. I gave up. The ELO system we have today, seems to be quite good based on my daily streak playing and seems to need only a slight tweak for the basic deduction of -7.x for all lost games. Thanks for your suggestion on the 9x2's. I will try them when my current 10x2 streak ends. As far as tournament ratings, I can say that if my daughter plays, (she's in school now and is smartly studying instead of playing), she will always beat me. But, I am still close behind and not the biggest behind! :-) I still enjoy playing against speedsters like yourself and others and hope for the situation that there are 3 or 4 extremely difficult games with lots of rabbit holes, and I get lucky and pick the right one, while you and the other speedsters miss it. I like the exclusion of tournament games from the ELO rating. Also, I like the loss of 0 points for a game that cannot be won. I used to play chess about 40 years ago and quit after making a blunder in a tournament. Even after that long pause of not playing, I was able to play against my grandkids and beat them because of the discipline and challenge of thinking ahead, that freecell requires. No shade on them because they all seem to be smarter and definitely faster. Cheers Crunch
Date: Thu Feb 3 04:49:01 2022
User: cellmate
Message:
i added a bar graph for players elo distribution ie bell curve
Link: cellmate's page
User: cellmate
Message:
i added a bar graph for players elo distribution ie bell curve
Link: cellmate's page
Date: Thu Feb 3 08:25:32 2022
User: MrFixit
Message:
Beautiful graph, cellmate. Confirms Elo is working as intended altho it appears we're shading a bit high. rbw - we've had different ways of measuring folks for years. While most play for the "streak" stat, some play for time or win % and now we've got Elo. Who's the best jumper, the high jumper or the long jumper?
User: MrFixit
Message:
Beautiful graph, cellmate. Confirms Elo is working as intended altho it appears we're shading a bit high. rbw - we've had different ways of measuring folks for years. While most play for the "streak" stat, some play for time or win % and now we've got Elo. Who's the best jumper, the high jumper or the long jumper?
Date: Thu Feb 3 10:24:09 2022
User: Kumquat-of-Conciliation
Message:
cellmate, on the histogram "Players Elo Distribution", you've got each bar labeled with a number (with 1500 being the largest/highest). Obviously there aren't that many players with that specific number. So is that bar 'centered' on 1500 (meaning the range would be 1475-1525), or is it the upper value of that segment, or lower value of that segment? Just curious. TIA.
User: Kumquat-of-Conciliation
Message:
cellmate, on the histogram "Players Elo Distribution", you've got each bar labeled with a number (with 1500 being the largest/highest). Obviously there aren't that many players with that specific number. So is that bar 'centered' on 1500 (meaning the range would be 1475-1525), or is it the upper value of that segment, or lower value of that segment? Just curious. TIA.
Date: Thu Feb 3 11:04:23 2022
User: cellmate
Message:
KoC, Close. i used range of 50 on the 50's. ie 1400-1449 1450-1499 etc. If you hover your mouse over a bar it will show the values. the chart tool does all the work to make it look nice i noticed that for charts with a legend that clicking on legend item toggles including and excluding that data from the chart
User: cellmate
Message:
KoC, Close. i used range of 50 on the 50's. ie 1400-1449 1450-1499 etc. If you hover your mouse over a bar it will show the values. the chart tool does all the work to make it look nice i noticed that for charts with a legend that clicking on legend item toggles including and excluding that data from the chart
Date: Thu Feb 3 11:07:10 2022
User: Kumquat-of-Conciliation
Message:
Well, I tried that, and just did again. It still only shows "1500", and below that a green box with "438". Doesn't show a range.
User: Kumquat-of-Conciliation
Message:
Well, I tried that, and just did again. It still only shows "1500", and below that a green box with "438". Doesn't show a range.
Date: Thu Feb 3 11:31:37 2022
User: cellmate
Message:
i'm reluctant to clutter the label. i think if we see the values for adjacent bars we can figure it out i added another for 8x4-5 Elos
User: cellmate
Message:
i'm reluctant to clutter the label. i think if we see the values for adjacent bars we can figure it out i added another for 8x4-5 Elos
Date: Thu Feb 3 11:34:25 2022
User: Kumquat-of-Conciliation
Message:
Well, I understand about not cluttering, but there's simply no way to tell what the range of those bars is. The adjacent bars also are only one specific value. Is everybody else seeing the same thing I am?
User: Kumquat-of-Conciliation
Message:
Well, I understand about not cluttering, but there's simply no way to tell what the range of those bars is. The adjacent bars also are only one specific value. Is everybody else seeing the same thing I am?
Date: Thu Feb 3 11:35:03 2022
User: cellmate
Message:
i'm a bit surprised by the 8x5-5. It gets the most plays. i expected the curve to be flatter
User: cellmate
Message:
i'm a bit surprised by the 8x5-5. It gets the most plays. i expected the curve to be flatter
Date: Thu Feb 3 11:42:04 2022
User: Kumquat-of-Conciliation
Message:
Hmmm...... Now I see you've changed it from histogram bars to points. I think the bars were better, but neither is showing the range. To be clear, I'm not at all complaining - was only asking for a clarification on grouping.
User: Kumquat-of-Conciliation
Message:
Hmmm...... Now I see you've changed it from histogram bars to points. I think the bars were better, but neither is showing the range. To be clear, I'm not at all complaining - was only asking for a clarification on grouping.
Date: Thu Feb 3 11:47:46 2022
User: Kumquat-of-Conciliation
Message:
Ok, better. But does that mean 1500-1550? Or 1475-1525? I presume the former?
User: Kumquat-of-Conciliation
Message:
Ok, better. But does that mean 1500-1550? Or 1475-1525? I presume the former?
Date: Thu Feb 3 11:56:40 2022
User: BuzzClik
Message:
The reason for the confusion, cellmate, is that the bars are centered over multiples of 50 -- the bar labeled 1500 is centered over 1500. So, the extent of that bar is approximately 1475-1525. BUT, your ranges are 1450-1499 and 1500-1549. Therefore, each bar straddles two ranges. I'm sure this traces back to the graphics program. (Considering where I am "on the spectrum", I'm not worried about this at all.)
User: BuzzClik
Message:
The reason for the confusion, cellmate, is that the bars are centered over multiples of 50 -- the bar labeled 1500 is centered over 1500. So, the extent of that bar is approximately 1475-1525. BUT, your ranges are 1450-1499 and 1500-1549. Therefore, each bar straddles two ranges. I'm sure this traces back to the graphics program. (Considering where I am "on the spectrum", I'm not worried about this at all.)
Date: Thu Feb 3 12:05:21 2022
User: Kumquat-of-Conciliation
Message:
Like always, I'm just looking out for Titanic Tony. We couldn't let his excellence be muddied, could we?
User: Kumquat-of-Conciliation
Message:
Like always, I'm just looking out for Titanic Tony. We couldn't let his excellence be muddied, could we?
Date: Thu Feb 3 12:10:54 2022
User: BuzzClik
Message:
The label of 1500+ on the bars helps a ton. The 1500 bar is 1500-1549, right?
User: BuzzClik
Message:
The label of 1500+ on the bars helps a ton. The 1500 bar is 1500-1549, right?
Date: Thu Feb 3 12:49:48 2022
User: MrFixit
Message:
I swore when I looked this morning it was a bar chart. MrGuapo - some good insights there. Regarding the curve for the game Elo I guess I hadn't imagine what it might look like but that's probably about what I'd expect. cellmate - can you add dropdowns to select the variant and difficulty real quick? :-)
User: MrFixit
Message:
I swore when I looked this morning it was a bar chart. MrGuapo - some good insights there. Regarding the curve for the game Elo I guess I hadn't imagine what it might look like but that's probably about what I'd expect. cellmate - can you add dropdowns to select the variant and difficulty real quick? :-)
Date: Thu Feb 3 12:59:41 2022
User: cellmate
Message:
KoC, Buzz for this chart, the labels describe the bar, not the x value the bars are numbered from left to right so here, the x grid line actually represents 'bar number', not elo value hiding the x grid (vertical lines) should help... when i figure out how
User: cellmate
Message:
KoC, Buzz for this chart, the labels describe the bar, not the x value the bars are numbered from left to right so here, the x grid line actually represents 'bar number', not elo value hiding the x grid (vertical lines) should help... when i figure out how
Date: Thu Feb 3 13:51:44 2022
User: cellmate
Message:
MrFixit, sorry no. its far easier to do server-side. i'll email
User: cellmate
Message:
MrFixit, sorry no. its far easier to do server-side. i'll email
Date: Thu Feb 3 13:56:25 2022
User: MrFixit
Message:
I was mostly joking. But I bet ElGuapo would be a big consumer of such graphs. I suspect variants which are less winnable have flatter curves but the 8x4 variant is so inherently winnable we see the pronounced bell. MrGuapo probably has thinking on that topic.
User: MrFixit
Message:
I was mostly joking. But I bet ElGuapo would be a big consumer of such graphs. I suspect variants which are less winnable have flatter curves but the 8x4 variant is so inherently winnable we see the pronounced bell. MrGuapo probably has thinking on that topic.
Date: Thu Feb 3 14:16:34 2022
User: Kumquat-of-Conciliation
Message:
I still can't see any difference, cellmate. As I and Buzz said, the "+" definitely helps, and I think we'll just assume "1500+" means 1500-1550. Quite interesting stuff.
User: Kumquat-of-Conciliation
Message:
I still can't see any difference, cellmate. As I and Buzz said, the "+" definitely helps, and I think we'll just assume "1500+" means 1500-1550. Quite interesting stuff.
Date: Thu Feb 3 15:45:56 2022
User: ElGuapo
Message:
Yep the Elo distribution for games is interesting but in any variant outside of 8x4 it's too early for it to tell us anything meaningful. We're reaching the point with 8x4-5 where individual game ratings are a better indicator of difficulty than the game stats. So for instance games 8x4 1891-5 and 8x4 23772-5 are now both identically rated at 1595 even though one has a 39% win rate and the other 28%. Other variants just don't get enough coverage for the individual games to all be played and therefore adjust based on who played them and how they fared. The average for the whole variant will move a lot with our boost algorithm, but a distribution chart will still look more or less like the day we started ratings. Cool charts, cellmate!
User: ElGuapo
Message:
Yep the Elo distribution for games is interesting but in any variant outside of 8x4 it's too early for it to tell us anything meaningful. We're reaching the point with 8x4-5 where individual game ratings are a better indicator of difficulty than the game stats. So for instance games 8x4 1891-5 and 8x4 23772-5 are now both identically rated at 1595 even though one has a 39% win rate and the other 28%. Other variants just don't get enough coverage for the individual games to all be played and therefore adjust based on who played them and how they fared. The average for the whole variant will move a lot with our boost algorithm, but a distribution chart will still look more or less like the day we started ratings. Cool charts, cellmate!
Post follow-up