![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
peterbirksCoding is a dangerous thing.
I played today for about 90 minutes with Auto HotKey rather than using the mouse, and, sad to say, the strain on the hand seemed about the same. However, I can swap from mouse to keypad and back to mouse, which seems to use slightly different muscle movements.
Ninety minutes in and Party Poker spotted that I was running a script. A popup appeared asking me to type within 90 seconds the five letters displayed. This failsafe anti-bot system is fairly pathetic, to be honest. I could quite easily have had a bot_script (Pokerpad merely exchanges mouse movements and mouse clicks for keystrokes) running and be doing other more interesting things like watching TV or reading a book. But I guess it filters out the "fire it up and go out to work" botters.
The trouble is, the guy who wrote PokerPad is no longer updating it, which is causing the donks on 2_+_2 no end of irritation. This seemed a clear challenge, so I downloaded WindowSpy and started looking through the PokerPad code. It seems fairly obvious where the numbers are that have caused IPoker to stop working (or, rather allegedly stop working; I haven't actually tried AHK on IPoker yet) and it seems to be to be a matter of finding the right screen co-ordinates for the various boxes.
This should also solve the PartyPoker "won't bet a proportion of the pot" problem. I just have to find the right co-ordinate on the screen that displays the size of the pot. I suspect at the moment that AHK is just looking in the wrong place, and therefore is trying to calculate 75% of _blank_.
++++++
I made some progress with Pokertracker 3 reports as well, and I managed to generate a list of players with a VPIP of less than 22% and an attempt to steal of more than 35% (min sample size, 200 hands, which might be a bit low). I leave it to the reader's imagination to what use I might put this.
My various att_to_steal stats are 29%, 27% (NoIQ and Betfred), 25% (Party) and 25% (PokerStars). The average of all qualifying players was 19.5%. I might put in another filter to have a minimum pre-flop raise percentage, to see how that changes things.
(Just threw in a minimum PFR of 9% as an additional filter. Average att_to_steal went up to 27.5%.)
Of the 20 big winners per hundred (once again, in very small samples of hands -- you really need about 3,000 hands as a minimum, I reckon) 12 raised less than the 19.51% and eight raised more than the average 19.5%. Which seems to indicate that it's not how often to attempt to steal (or, rather, how tight your range is), as how much you raise and how you play it when you get called and/or reraised.
Fold to 3 bets among the big winners ranged from 50% to 100%. The more frequent stealers, unsurprsingly, fold to 3 bets more often (generally 75% to 100%). These are, of course, extemely small samples for individual players, but the trend looks constant.
Interestingly, the player with the highest steal rate (82%) only folded to a 3bet half the time. Clearly he was getting used to being 3bet and had a second string to his bow. However, he still only went to showdown 20% of the time. So, playing back against this kind of guy probably needs some gutsy commitment if you want to stand up to him successfully.
The average win rate for players with a VPIP of less than 22% was 56 cents a 100, btw, mainly at $100 buy in with about a third of the hands being $50 buy in and about a 10th being $200 buy in. The average $ per hundred for players with a VPIP of less than 22% and a PFR of at least 9% was $2.41 per 100.
______________________
I played today for about 90 minutes with Auto HotKey rather than using the mouse, and, sad to say, the strain on the hand seemed about the same. However, I can swap from mouse to keypad and back to mouse, which seems to use slightly different muscle movements.
Ninety minutes in and Party Poker spotted that I was running a script. A popup appeared asking me to type within 90 seconds the five letters displayed. This failsafe anti-bot system is fairly pathetic, to be honest. I could quite easily have had a bot_script (Pokerpad merely exchanges mouse movements and mouse clicks for keystrokes) running and be doing other more interesting things like watching TV or reading a book. But I guess it filters out the "fire it up and go out to work" botters.
The trouble is, the guy who wrote PokerPad is no longer updating it, which is causing the donks on 2_+_2 no end of irritation. This seemed a clear challenge, so I downloaded WindowSpy and started looking through the PokerPad code. It seems fairly obvious where the numbers are that have caused IPoker to stop working (or, rather allegedly stop working; I haven't actually tried AHK on IPoker yet) and it seems to be to be a matter of finding the right screen co-ordinates for the various boxes.
This should also solve the PartyPoker "won't bet a proportion of the pot" problem. I just have to find the right co-ordinate on the screen that displays the size of the pot. I suspect at the moment that AHK is just looking in the wrong place, and therefore is trying to calculate 75% of _blank_.
++++++
I made some progress with Pokertracker 3 reports as well, and I managed to generate a list of players with a VPIP of less than 22% and an attempt to steal of more than 35% (min sample size, 200 hands, which might be a bit low). I leave it to the reader's imagination to what use I might put this.
My various att_to_steal stats are 29%, 27% (NoIQ and Betfred), 25% (Party) and 25% (PokerStars). The average of all qualifying players was 19.5%. I might put in another filter to have a minimum pre-flop raise percentage, to see how that changes things.
(Just threw in a minimum PFR of 9% as an additional filter. Average att_to_steal went up to 27.5%.)
Of the 20 big winners per hundred (once again, in very small samples of hands -- you really need about 3,000 hands as a minimum, I reckon) 12 raised less than the 19.51% and eight raised more than the average 19.5%. Which seems to indicate that it's not how often to attempt to steal (or, rather, how tight your range is), as how much you raise and how you play it when you get called and/or reraised.
Fold to 3 bets among the big winners ranged from 50% to 100%. The more frequent stealers, unsurprsingly, fold to 3 bets more often (generally 75% to 100%). These are, of course, extemely small samples for individual players, but the trend looks constant.
Interestingly, the player with the highest steal rate (82%) only folded to a 3bet half the time. Clearly he was getting used to being 3bet and had a second string to his bow. However, he still only went to showdown 20% of the time. So, playing back against this kind of guy probably needs some gutsy commitment if you want to stand up to him successfully.
The average win rate for players with a VPIP of less than 22% was 56 cents a 100, btw, mainly at $100 buy in with about a third of the hands being $50 buy in and about a 10th being $200 buy in. The average $ per hundred for players with a VPIP of less than 22% and a PFR of at least 9% was $2.41 per 100.
______________________
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2008-09-03 01:12 pm (UTC)"Yes, Leonardo, I understand the symbolism and the perspective and all, but why do those three pixels make it look like the silly bitch is scratching her nose?"
no subject
Date: 2008-09-03 01:17 pm (UTC)Help and suggestions would be appreciated, but not expected.
PJ
no subject
Date: 2008-09-04 08:36 pm (UTC)Still a damn silly way of targeting bits of an application, though. On the Web, you have the DOM (which basically lets you say "do this to that button there"). On a Windows program, you have something slightly similar, but needlessly complicated (which basically lets you say "find this window, assuming it has a name, which it almost certainly hasn't, but you can guess, and if you're lucky you can send a message to ask it questions in an arcane language, and then you can send it another message which ... bzzzt ... crackle ..." Perhaps a co-ordinate based system makes more sense). Anyway, it's doable, and probably amenable to explanation in a single, not very technical, email. Or indeed comment back at t' blog.