Subject: hop question
Date: Thu Feb 4 14:22:08 2021
User: ix
Message:in the joey competitions, it used to be that clicking on a player's name took you to their stats, now it gives some if won/if lost stats. would it be easy to make the player's name in that box clickable to go to their stats?
Date: Thu Feb 4 14:31:37 2021
User: free@last
Message:hop,
either skip this question or jump on it. ;)
Date: Thu Feb 4 23:27:24 2021
User: ix
Message:that is so cool, thank you!
Date: Thu Feb 4 23:47:51 2021
User: WRAC
Message:Thanks, hop, and great suggestion, ix. This is very helpful!
Date: Fri Feb 5 00:01:43 2021
User: Oded789
Message:I haven't played these for a while, but I gotta chime in and say once more that the data display on Hop's pages is truly excellent, and even when you think it can't be improved anymore, somehow it does.
Date: Fri Feb 5 00:34:09 2021
User: cellmate
Message:
While you're at it Hop...
I'm looking at a "...Alltime best daily Standings" page. I'm waaay down the list so paging thru. The table width changes with Player length so the arrow buttons keep moving so my clicks oftn miss. Kind of annoying. Can that be fixed? Or make the left/right arrow keys work for that?
Date: Fri Feb 5 00:39:53 2021
User: cellmate
Message:
should mention i use Chrome. i just gave IE 11 a try for comparison... uh... needs work!
Date: Fri Feb 5 01:04:38 2021
User: HopDiriDiriDattiriDittiriDom
Message:cellmate,
It's not clear to me which page you are referring to. Can you give me the address?
IE? I only test on FF (not even Chrome but it seems like it's OK on it though)
I don't think I will bother myself with IE.
Date: Fri Feb 5 01:17:20 2021
User: cellmate
Message:
https://hoppi.neocities.org/freecell/variant.html?variant=8x4&column=5
don't blame you for not bothering. but a couple of times i learned something useful figuring out why it did't work. just saying.
Date: Fri Feb 5 01:23:32 2021
User: cellmate
Message:
case in point https://www.freecell.net/f/c/ps.html?u=cellmate
in Chrome i see a green +. FF not. Some setting needs setting
Date: Fri Feb 5 01:44:25 2021
User: HopDiriDiriDattiriDittiriDom
Message:How did you end up in that page? That page is archaic. Redirected from somewhere?
Because I now link (from joeys competitions page) to:
https://hoppi.neocities.org/freecell/scores-a-la-hop.html?variant=8x4&column=2
And you can click any column in that page. And there is no arrow in that page.
And your last post:
"case in point https://www.freecell.net/f/c/ps.html?u=cellmate
in Chrome i see a green +. FF not. Some setting needs setting"
It's not clear to me either. That's Denny's.
Date: Fri Feb 5 01:57:03 2021
User: cellmate
Message:
various competitions (live), shows 12-sum Current Standings, click on 8x4 in column headings
Date: Fri Feb 5 02:13:43 2021
User: cellmate
Message:
oh... using the link you provided... its the same problem trying to click '+1'... it keeps moving
Date: Fri Feb 5 02:18:37 2021
User: HopDiriDiriDattiriDittiriDom
Message:I don't know what I can do for it.
Would you be happy with arrow keys (keyboard)? Maybe I can do it.
Date: Fri Feb 5 02:31:04 2021
User: cellmate
Message:
i would think there's a way to control the width of the table
but yes... i'd be ok with arrow keys... if it remembers that my last click was '+1' or '+5' or whatever
Date: Fri Feb 5 03:01:47 2021
User: HopDiriDiriDattiriDittiriDom
Message:"i would think there's a way to control the width of the table"
That's beyond my current knowledge.
why not +1 always? You click first and then press the keys? Can be done but I don't see the point. Always +1 makes sense to me with the arrow keys.
Date: Fri Feb 5 03:16:36 2021
User: cellmate
Message:
Ok. arrow keys. +1
Date: Fri Feb 5 08:06:45 2021
User: HopDiriDiriDattiriDittiriDom
Message:The right arrow key is +1 (if not the last) and the left one -1 (if not the first).
Date: Fri Feb 5 11:10:31 2021
User: cellmate
Message:
Thanks Hop!
Noticed that going too fast with left arrow sometimes gets twice as many rows
(i'll slow down)
Date: Fri Feb 5 12:17:40 2021
User: cellmate
Message:
Hop, its worth learning how to control width... altho it can lead to endless tweaking
i usually use %'s. just remember its a percentage of the parents width
you have 3 div's
1. for -1 -5 etc
2. for the table
3. for the +1+5 etc
i added some width attributes
for div 1, i added style="width:10%;...
for div 2, i added style="width:80%;...
for the table itself in div 2 i added style="width:100%;...
now the +1 stays in place, and the table stays the same size, and i can click away on +1 without having to move the mouse
Date: Fri Feb 5 12:38:03 2021
User: HopDiriDiriDattiriDittiriDom
Message:I don't like %'s. Sometimes I use them but not that much. So I will not change it. Anyway sounds like you have already copied and done it for yourself.
Personally it's not problem for me if it (+1) does not stay at the same place.
Date: Fri Feb 5 12:50:44 2021
User: cellmate
Message:
no, i used the debugger to patch it in
i'm not interested enough to maintain my own version
arrow keys work so i will use that
Date: Fri May 14 00:16:27 2021
User: ix
Message:is there a simpleton friendly way of explaining expected values? right now in 12 sum i have:
Expected value -17,738 46,610 67,299 3,865 4,004 3,665 3,889 -6,285 -161,345
i'd be pleased to know what they mean
Date: Fri May 14 02:49:58 2021
User: HopDiriDiriDattiriDittiriDom
Message:ix,
We had discussed it before (see the link, especially the one about the similarity to roulette wheel).
In summary it's about your chances based on your current score in any variant. It's gives you a lead which to play next. The higher it is the more chances you have to increase your current score.
Link: Expected value for Joey's scores
Date: Fri May 14 06:57:52 2021
User: ix
Message:thank you
Date: Fri May 14 09:01:49 2021
User: ix
Message:i crunched my 11x1 numbers and got the same as you, so i actually get it!
Date: Fri May 14 12:22:02 2021
User: TNmountainman
Message:Just as a curiosity..........for us non-cognoscenti..........where does one find this "expected value"? I looked at the link Hop provided, going back to the thread starting Dec. 6, but couldn't find anything there.
Date: Fri May 14 12:49:07 2021
User: cellmate
Message:
click link to Hop Pages
click Joey's various competitions
on the left click the -sum you're interested in
on the right click a username
Link: Hop's Freecell Pages
Date: Fri May 14 13:20:29 2021
User: TNmountainman
Message:Ah, thanks. Never seen that 'page'. So.........a negative expected value means one is more likely to lose the next game of that variant, and vice versa? No......maybe that's not it.......
Date: Fri May 14 13:43:44 2021
User: ix
Message:i think it's for comparison (not sure if that's what you were saying), if i have things right:
using the roulette example, a dollar bet to win 35 has an expected value of
(1/38)*35 + ((37/38)*(-1)) = 0.0526
a dollar bet to win 10
(1/38)*10 + ((37/38)*(-1)) = -.7105
a two dollar bet to win 35
(1/38)*35 + ((37/38)*(-2)) = -1.026
shows you are better off betting a dollar to win 10 than 2 to win 35 and best is a dollar to win 35
Date: Fri May 14 13:44:17 2021
User: HopDiriDiriDattiriDittiriDom
Message:I would say it means "if you play that variant many times in the same conditions you would lose those points on average."
Date: Fri May 14 13:50:00 2021
User: ix
Message:i missed a minus sign on -.0526, but hop's answer is way better, like way better
Date: Fri May 14 14:31:21 2021
User: TNmountainman
Message:Ok, so taking Hop's answer literally, if one "breaks" that down to 1 game, is it not the same as saying "more likely to lose than win"?
And ix..........I see your first example, but it's non-intuitive (to me). If .0526 represents the "return", then it would seem to make sense to do that bet, and over 20 individual bets, the cumulative total would be above 1, thus making it pay off. Yet we know that's not true, so I'm somehow missing something there. (Not surprisingly.) And I'm likely not ever going to really delve into this, so don't waste much time humoring me. :)
(Oh wait............if the expected outcome is *negative* .0526, then it makes sense. Just now read the last post.)
Date: Fri May 14 14:48:38 2021
User: ix
Message:if you look at hop's answer, the words "in the same conditions" are very important, it's a weighted thing. in 12x0, the odds of hitting an unwinnable are about 10 percent, but since you stand to lose so much more than gain, even though you will likely win the next one, if you play a bunch you will lose points on average
Date: Sat May 15 03:20:32 2021
User: TNmountainman
Message:So my buddy smelly kindly gave me permission to take a gander at his stats and expected values. For 4x6, it's 1. Just 1. "New score if won" is 9, and "New score if lost" is 6. And with an expected losing percentage of 73.2%, I can guesstimate that that means a DELTA of +9 or -6. I think. Is that right? And thus "expected value" means the expected delta, really, I think?
(Sorry I was absent those class days this must have been discussed earlier; and I think it was too much in computer-ese, anyway.)
Date: Sat May 15 03:54:24 2021
User: ix
Message:your current score is 6, so it's a delta of +3 or -0, this is a case of rounding. ev would be (points to gain)(chance to gain)-(points to lose)(chance to lose) or just 3*.268=.804
hop's statement of "if you play that variant many times in the same conditions you would lose those points on average" was very enlightening for me. i means it's a long term concept with little application to the next game played.
Date: Sat May 15 04:17:20 2021
User: HopDiriDiriDattiriDittiriDom
Message:Looking at his stats, his 4x6 has no score. So if he plays and loses 4x6 his current score (6) will not change but if he wins by the chance (100-73.2)% his new score will be 9 So:
gp > (9-6)*(1-0.73209)+(6-6)*0.73209
%51 = 0.80373 // expected value
In other words if he plays 100 games in the same conditions (as if he has the same stats when plays them), roughly he will lose 73 and win 100-73=27 of them. So we might as well write it as follows:
gp > (3*(100-73)+0*73)/100. (that is 27 times he wins 3 points and 73 times gains nothing out of 100 attempts)
%54 = 0.81 // roughly the same as above
So expected value is an average based on the player's current stats and the variant's winning percentages.
Date: Sat May 15 04:54:20 2021
User: TNmountainman
Message:I think I got it with ix's help, but would it be fair to say that the "1" could/should be "+1"? Or is that redundant because "by definition" no scores will be below 0?
In your example, Hop, is "gp" "games played", "game probability", or something else?
I (think) I understood Hop's sentence, ix, but how is different from reductively saying the same thing, proportionally, for each game? A player's score will change with each game played, likely, so that's the important part of "in the same conditions" is that it would be assuming the player's score had not changed, I believe/hope.
Maybe some of this goes back to how the "score" is calculated - hopelessly far back in time for most of us to know how that's done, without the formula(s) on the scores page. (Just my opinion.) Is smelly's "score" 6, as in 6.00, or some fractionated variation?
And if one looks at smelly's 10x0, which has fairly close to 50% winning percentage, one would *think* that a win or loss would impact the "score" equally, yet the expected value doesn't show that - unless it's just a rounding phenomenon - which I'm sort of guessing it is.
(And again, my apologies for missing class the days this stuff was worked out. So don't spend much time on this for my sake. Chances I'll actually use it to make decisions about what games to play is close to nil. I'm just curious about it.)
Date: Sat May 15 05:06:22 2021
User: HopDiriDiriDattiriDittiriDom
Message:TN,
As for that "gp>" thing; it's the command prompt of PARI/GP calculator (a freeware program to do math calculations). It displays the following at the startup:
GP/PARI CALCULATOR Version 2.13.0 (released)
amd64 running mingw (x86-64/GMP-6.1.2 kernel) 64-bit version
compiled: Oct 19 2020, gcc version 8.3-posix 20190406 (GCC)
threading engine: single
(readline v8.0 enabled, extended help enabled)
Copyright (C) 2000-2020 The PARI Group
PARI/GP is free software, covered by the GNU General Public License, and comes WITHOUT ANY WARRANTY WHATSOEVER.
Type ? for help, q to quit.
Type ?17 for how to get moral (and possibly technical) support.
parisize = 8000000, primelimit = 500000
(10:10) gp > eulerphi(43)
...
Date: Sat May 15 05:08:51 2021
User: ix
Message:definitely some rounding going on, i think you are at 5.87 or something like that
Date: Sat May 15 05:11:10 2021
User: ix
Message:5.68
Date: Sat May 15 05:30:42 2021
User: TNmountainman
Message:Hop, thank you for your explanation, but that assumes I have recent computer skills and/or knowledge, which I don't. And I suspect 95+% of the other players here don't, either. Not in any way a criticism - just making sure you appreciate the lack of such knowledge among "the general populace".
But anyway. .....I understand better now, so thanks to you both.
Date: Sat May 15 05:55:03 2021
User: ix
Message:>> And if one looks at smelly's 10x0, which has fairly close to 50% winning percentage, one would *think* that a win or loss would impact the "score" equally
not true, if you are playing a variant with a 60% win percent, but stand to gain 100 points with a win and lose a 1000 points with a loss, you are more likely to win games yet lose points
Date: Sat May 15 14:12:46 2021
User: TNmountainman
Message:Let's say one has an ELG of 1500. And plays a 10x0-5 game. Further assume that the winning percentage of 10x0-5 games is 50% - which it's not, exactly. One would expect an equal impact in either direction with a win or loss. Would *that* be correct, IF said player played perfectly and won all winnable games. Is that *theoretically* accurate? And thus the divergence from that is because of the 'real' success rates on 10x0s?
Date: Sat May 15 15:27:50 2021
User: Oded789
Message:No. Roughly put, average ELO (why is it you call it ELG? It's not an abbreviation) of variant and level is determined by its actual win rate, not the theoretical one.
Meaning that your question is more fitting for 11x0-5 in which the actual win rate is very close to 50%.
However, the average game ELO of that is way higher (currently 1669) due to some (should be all, but not yet) unwinnables being marked at 3000+.
We'd do better to check something like 6x5-5 (over 99% winnable, about 55% actual win rate), in which the average game is indeed just over 1500 (currently 1534).
So if the win rate was 50% in a level which is 100% winnable, the answer to your question about equal impact would be yes.
Date: Sat May 15 15:42:37 2021
User: TNmountainman
Message:I term it "ELG" as a tip o' the cap to EL Guapo. And yes, I know it's not an abbreviation, as I've noted before. But I do think it's a hybrid, so trying to give EL Guapo some credit, whilst retaining the flavor of the original.
Will respond to the rest later....
Date: Sat May 15 16:10:48 2021
User: TNmountainman
Message:As to ELO/ELG, your first paragraph above corresponds to my last sentence above it, I believe. "Hypothetical", as we know, doesn't help much with the difficult variants. I was only trying to understand the "first principles", from which the practical application derives.
So your comparison to 11x0 is interesting. I can see where (finally) the unwinnables, rated above 3000, would raise the average of the whole - but isn't that true for 10x0 as well?
So with that said......IF IF IF all players played 10x0-5s perfectly, AND it was winnable/won 50% of the time......would the gain or loss delta be the same in both directions - beginning from 1500? But now, with you pointing out that the known/apparent unwinnables have a higher rating, I can see where that could create an upward bias. Hope all that makes sense.
Date: Sat May 15 16:30:54 2021
User: firenzes_mother
Message:One of my ex-husbands had an upward bias. Made him walk oddly.
Post follow-up