|
|
#32
08-08-2005, 10:13 PM
|
|||
|
|||
Re: Longest feasible downswing for solid player
This is a fantastic graph of an average winning online low limit player.
1. Notice the win rate. People say they win at 3bb/100 blah blah blah but this is a very very very rare exception, and most people are full of [censored]. 2. Notice the semi-large sample size. Not even close to a true sample, but most graphs I have seen hold less than 30k hands. 3. Notice how many breakeven streaks, even little downstreaks, our hero has over 10-15k sample sizes. 4. Notice how many huge 300bb upswings our hero has over similar sample sizes. Many questions by newer players ask "whats a good win rate", "is 30k hands a good sample?", "I lost over 15k hands...WHY!" and "I win 4bb/100 over 20k hands...am I god?" This graph answers many of those questions. I would love to see three or four 100k hand graphs by some of our full ring game pros, such as a microbob type player thats has over a years worth of data, and see the differences over these 100k hand samples 
|
|
#33
08-08-2005, 10:45 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
[ QUOTE ] 10k hands is nothing more than static in LHE. Your win/loss result over that amount of hands has almost nothing to do with your actual edge in the games. [/ QUOTE ] This really isn't true for limits less than 15/30 IMO where winrates are commonly over 2.5BB/100 and Std. Dev less than 18. -SmileyEH [/ QUOTE ] I find it hard to believe that many 10/20 players or even 5/10 players have true winrates as high as 2.5BB/100, especially if they usually four-table. But anyway, I will take your WR and SD and do some math and we can look at the results.  Turned out to be a pretty interesting exercise. I hope we can pull something from the results. Hopefully some of the more probability versed people can tell me if I made a major mistake in my calculations. The one assumption I made that I wasn't sure of, was that if a player with given stats has a WR between +N and -N for a confidence interval of X%, that that player would have a WR > N (1-X/2)% of the time, and a WR < N (1-X/2)% of the time. It makes sense intuitively but it might not be correct. You can see that even after a 400BB downswing over 50k hands, the player will have a true WR of >=2.5BB/100 0.08% of the time, or about one time in 1250. This seems farfetched, however the player will have a true WR of >=1.5BB/100 about one time in 74! This is a very possible occurence. Additionally, the player will be breakeven or better more than one time in five. So even a tremendous downswing like that won't really tell you that you're a losing player. Another interesting thing to note is that a player that experiences a 400BB upswing over 10k hands still has a greater than 1% chance of being a losing player! And he is only going to be a 1.5BB/100 or greater player 92% of the time. I'll have to do this again at some different SDs too.
|
|
#34
08-08-2005, 10:59 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
I would love to see three or four 100k hand graphs by some of our full ring game pros, such as a microbob type player thats has over a years worth of data, and see the differences over these 100k hand samples [/ QUOTE ] Well this isn't 100k hands but I never use the same database for that long, still it is starting to get into the range of "statistically viable sample sizes".  Over this sample which lasted ~67,400 hands, my WR was 1.52BB/100 hands and my SD was 15.7BB/100 hands. This means: 0.6% chance I'm actually a losing player over this stretch 19.5% my winrate is less than 1BB/100 5% chance I'm greater than a 2.5BB/100 winner And if you took any individual 10k hand sample from here you would get plenty of negative WR samples. 10k hands = static.
|
|
#35
08-08-2005, 11:05 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
I think if it goes past 200-250BB down you should start to worry. [/ QUOTE ] I think that's a bit low, depending on what he's playing. I agree if he's playing 2/4 (easy) or 3/6 (low variance), but 5/10 and up, a 200 BB downswing will probably happen every 50k hands or so. I think more worrisome than a 200 BB downswing would be an extended break-even stretch of like 25k hands
|
|
#36
08-08-2005, 11:06 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
someone with that sd who has lost 200 BB over a 10k hand sample has about a 7-8% chance of actually being a winning player. it happens. [/ QUOTE ] I'm just trying to say, if you're playing 2/4 and you breakeven for 20k hands you have a fair bit you can work on. Sure it's going to happen to good players, I know - I've run bad, and then worse than I ever thought possible. My point is that it's very unlikely to breakeven for such a long time at low limits where winrates are high and standard deviations are low. FWIW an increased winrate goes along with an increased std dev IMO (play more aggressively/more hands increases both) so I think the two factors cancel eachother out (this summer my std dev. is 16.5 FWIW). But again, assuming a 2BB/100 winner with a 14/100 SD - they only have about a 7% chance of breaking even or worse for 10k hands. When people usually come asking whether they are losing players/just running bad etc. it is usually their first x hands at a limit. If they're breakeven or losing for their first 10k of 2/4 or whatever and we can say that with 93% certainty their winrate is less than 2BB/100, well that is saying something. Of course there will always be freaks of variance but they are the exception. My biggest peeve is that posters with 10's or 100's of thousands of hands talk about 20k breakeven streaks as if they are nothing - well if you play 300k hands a year they aren't because all those hands make freaks of variance bound to happen - but if you move up a limit and suddenly you find yourself down after 100 hours of 4tabling: you'd better sit up and analyze your game because odds are you've got significant room for improvement. (notice 4BB/100 didn't appear once in this post [img]/images/graemlins/smile.gif[/img]). -SmileyEH
|
|
#37
08-08-2005, 11:08 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
(notice 4BB/100 didn't appear once in this post [img]/images/graemlins/smile.gif[/img]). [/ QUOTE ] [img]/images/graemlins/confused.gif[/img]
|
|
#38
08-08-2005, 11:17 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
I can't argue with any of this. I think you are especially correct to say that it is a much bigger concern when it is someone's first x number of hands where they are losing.
I just think people tend to underestimate just how much variance plays a role in this game, and putting some numbers to show that is a good thing. [ QUOTE ] (notice 4BB/100 didn't appear once in this post ). [/ QUOTE ] [img]/images/graemlins/grin.gif[/img]
|
|
#40
08-08-2005, 11:37 PM
|
|||
|
|||
|
Re: Longest feasible downswing for solid player
[ QUOTE ]
[ QUOTE ] (notice 4BB/100 didn't appear once in this post [img]/images/graemlins/smile.gif[/img]). [/ QUOTE ] [img]/images/graemlins/confused.gif[/img] [/ QUOTE ] I am confused ... are you guys saying the people on this forum don't have winrates above 4 typically?
|
 |
|
|
Linear Mode
