Thursday, February 26, 2009

playing wet flops as pfr oop

here's a link to a discussion about a pretty common spot on the deucescracked forum: http://www.deucescracked.com/forums/17-Intermediate-MS-No-Limit-Hold-em/topics/16184-JJ-overpair-vs-flopraise?page=1&per_page=30#posts-119386

here's a link to a spreadsheet where i tried to generalize from that: http://spreadsheets.google.com/pub?key=pvjFKb7H91RqCUjhFzychIw

cliff notes: 
  • folding overpairs on wet boards is bad. 
  • regardless of your hand, dont give up on these flops! usually cbet or c/r w/ your entire range, generally cbetting v. passive (few bluffs in range) and c/ring v. aggressive (bluffy) opponents.

bb defense/donk

i haven't written up any analysis, but here's a link to a spreadsheet addressing donk betting in various board textures:

http://spreadsheets.google.com/pub?key=pvjFKb7H91RoEfcoL3uqVOg

BB Facing Steal

bb facing steal situation: button opens to 3bb, bb calls, bb checks flop, button bets 5bb, bb raises to 16bb laying ~2.5:1. button calls w/ one of three ranges. if button calls bb is assumed to check/fold the turn.   

question: for different button ranges, which flops are best for check-raising? 

cliff notes: surprisingly enough, some flops hit more
starting hands than others. if you are playing against players who
don't fully appreciate this shocking discovery, the following analysis
should prove useful. if not, it may still be worthwhile to skim the 'results' and then move to the 'more discussion' section, as i thought this analysis (and the accompanying spreadsheet) were somewhat helpful for framing questions about good/bad double barrel spots. 

method: simulations were run using stoxev, where button was given a specific opening range (top x%) and flop texture was defined. the button's c/r calling range was then set and fold % was determined. 

results (opening %, fold % range 1; fold % range 2; fold % range 3) (disclaimer: i'm not so great w/ details sometimes, so take this fwiw and please let me know if something looks wrong):         

1. flop: Ah9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (41.1); [56.7]; {70.9}
  • 30, (38.0); [48.9]; {71.9}
  • 40, (47.9); [59.4]; {79.9}
  • 50, (47.1); [58.2]; {74.6}
2. flop: Kh9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (25.4); [61.3]; {72.6}
  • 30, (31.5); [56.7]; {77.4}
  • 40, (45.9); [64.7]; {83.6}
  • 50, (51.9); [69.5]; {84.5}
3. flop: Qh9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (25.6); [54.4]; {63.1}
  • 30, (32.4); [53.6]; {73.2}
  • 40, (42.4); [62.4]; {79.8}
  • 50, (49.9); [69.5]; {83.5}
4. flop: Jh9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (25.3); [54.2]; {61.6}
  • 30, (31.2); [52.7]; {69.9}
  • 40, (39.2); [58.9]; {75.3}
  • 50, (46.6); [65.0]; {78.3}
5. flop: 8c9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (43.9); [67.9]; {69.1}
  • 30, (34.6); [53.5]; {59.9}
  • 40, (32.1); [52.1]; {63.8}
  • 50, (37.7); [56.1]; {69.2}
6. flop: 5d5c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (65.6); [67.7]; {70.2}
  • 30, (69.8); [73.9]; {76.0}
  • 40, (66.4); [71.8]; {74.1}
  • 50, (63.7); [71.6]; {75.7}
7. flop: AhQc3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (3.70); [50.0]; {70.1}
  • 30, (26.0); [56.3]; {72.1}
  • 40, (42.7); [68.2]; {80.1}
  • 50, (42.2); [64.7]; {75.1}
8. flop: 8h8c7d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (62.7); [80.8]; {81.8}
  • 30, (57.6); [74.8]; {79.1}
  • 40, (46.3); [66.2]; {75.4}
  • 50, (50.9); [67.3]; {77.8}
9. flop: ThTc3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (53.6); [56.3]; {72.2}
  • 30, (67.1); [68.9]; {79.9}
  • 40, (74.2); [75.3]; {84.4}
  • 50, (76.8); [77.8]; {87.0}
10. flop: JhTh9d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (1.35); [27.5]; {38.8}
  • 30, (5.50); [31.1]; {38.0}
  • 40, (13.9); [38.9]; {44.4}
  • 50, (22.4); [45.4]; {50.1}
some observations:
  1. the effectiveness of check-raise bluffs on some flop textures is WEAKLY dependent on button OPENING range, see flops 1, 5, 6, 8 and 9.
  2. the effectiveness of check-raise bluffs on some flop textures is WEAKLY dependent on button CALLING range, see flop 6, and to a lesser extent 8 and 9 (note that these flops are all paired and all included in observation 1).
  3. the effectiveness of check-raise bluffs on some flop textures is STRONGLY dependent on button OPENING range, see flops 2, 3, 4, 7 and 10.
  4. the effectiveness of check-raise bluffs on some flop textures is STRONGLY dependent on button CALLING range, see flops 1-5, 7 and 10.
some discussion: the obvious general cause of observations 1 & 2 is that these flops are hard to hit (few draws and 2pr+ hands) and that 'any pair' ~ >=mp ~ >=tp. that is, the ranking of starting hand strength is relatively unchanged after the flop. for example, on flop 1 Ah9c3d it should be obvious that once the button opening range includes all Ax and pp(TT+), there is no way to further increase the number of top or 1p>mp hands. similarly, observations 3 & 4 identify flops that interact strongly with the top x% of starting hands. i suspect that the ability to quickly and accurately adjust to these types of flops is a major source of reciprocal gold

more discussion: consider flop 2,          

Kh9c3d; (btn calls any pair, fd, gutshot+); [btn calls >=mp, fd, oesd]; {btn calls >=tp, fd, oesd}
  • 20, (25.4); [61.3]; {72.6}
  • 30, (31.5); [56.7]; {77.4}
  • 40, (45.9); [64.7]; {83.6}
  • 50, (51.9); [69.5]; {84.5}
two obvious points stand out: first, the button is very susceptible to being c/r bluffed if he is not calling w/ mid pairs. second, c/r bluffing against a btn capable of calling light or floating greatly reduces the profitability of an isolated flop c/r. considering turn play, if we take the open-raising range of 40% and use the second c/r calling range of [>=mp, fd], we find that btn is left holding <=mp on the turn 51% of the time. thus, if btn continues past the turn only w/ a range of {>=tp, fd, 1p+>=gs}, they will be folding 43.8% of the time, and provide the bb an opportunity to fire a profitable 2nd barrel. considering a complete decision tree, where bb fires all turns, btn calls w/ {>=tp, fd, 1p+>=gs}...bb bets river with >=tpmk (checks otherwise), btn calls w/ >=tp, or bets when checked to w/ >=2p (bb always folds river after checking). we find that the flop c/r ev = 4.8bb, and the turn 2nd barrel ev = 9.4bb. interestingly, there is not a -ev decision for either bb or btn in this particular tree if the bb calls preflop w/ atc. if we apply this identical analysis to other flops, we find:     

     1. Ah9c3d
  • btn cbet ev = 1.6bb
  • bb c/r ev = 3.9
  • bb 2nd barrel ev = 9.6bb
    3. Qh9c3d
  • btn cbet ev = 1.0bb
  • bb c/r ev =  4.5bb
  • bb 2nd barrel ev = 8.3
    4. Jh9c3d
  • btn cbet ev =  1.4
  • bb c/r ev =  4.2
  • bb 2nd barrel ev = 7.3
    5. 8c9c3d
  • btn cbet ev =  2.7
  • bb c/r ev =  2.7
  • bb 2nd barrel ev = 7.3
    6. 5d5c3d
  • btn cbet ev =  1.4
  • bb c/r ev =  4.3
  • bb 2nd barrel ev = 0.5
    7. AhQc3d
  • btn cbet ev =  2.4
  • bb c/r ev =  3.2
  • bb 2nd barrel ev = 2.7
    8. 8h8c7d
  • btn cbet ev =  2.4
  • bb c/r ev =  3.2
  • bb 2nd barrel ev = (1.9)
    9. ThTc3d
  • btn cbet ev =  1.5
  • bb c/r ev =  4.2
  • bb 2nd barrel ev = (0.7)
    10. JhTh9d
  • btn cbet ev =  6.9
  • bb c/r ev =  (1.6)
  • bb 2nd barrel ev =  (1.7)
unless i'm missing something, the thing that stands out here is that c/r bluffing and 2nd barreling are rarely significant ev mistakes v. most btn open ranges. that's not to say that these plays are anywhere near optimal. [note: btn cbet ev is biased by the bb pf call + flop c/r range of ALL HANDS.] i've expanded this ev analysis to include the ten flops above across the different opening ranges in this spreadsheet: http://spreadsheets.google.com/pub?key=pvjFKb7H91Rrpd-aisSImAQ




Comments on Analysis of RR Ranges...

the following discussion refers to a spreadsheet found here: http://spreadsheets.google.com/pub?key=pvjFKb7H91RqBvslkH-zcJA

bb cbetting frequency makes all the difference
: the fact that all players do not c-bet all flops is mentioned in the original article, but the effect of this may be somewhat understated. consider a flop of Js92s facing the 4.7% rr range and button calling range of 7.9% {88+, ATs+, AQ+, JTs+}. if the bb c-bets all flops then the button min-r returns ev(100%) = 15.4bb (assuming that bb shoves, and button calls, with a range of [>=tp, fd, oesd, >=1p+gs]). in contrast to when the bb check/folds overcards and check/calls mid pairs (assumed to check down after flop), the bb will check ~57% of their hands and the button min-r returns ev(43%) = (18.7bb).

applying these same assumptions about checking, calling and shoving ranges to additional flops we get the following ev(cbet%): 
  • 874t: ev(100%) = 10.2bb; ev(47%) = (14.3bb)

  • Q94t: ev(100%) = 9.6bb; ev(50%) = (15.4bb)

  • K73t: ev(100%) = 10.9bb; ev(69%) = (1.6bb)

  • A86t: ev(100%) = 10.0bb; ev(54%) = (11.9bb)

  • KT5r: ev(100%) = 6.9bb; ev(77%) = (1.6bb)

  • T63m: ev(100%) = 3.0bb; ev(49%) = (16.6bb)

  • QJ2t: ev(100%) = 4.6bb; ev(72%) = (6.6bb)

it is possible that these negative expectations for raising over a constricted c-betting range are largely due to the conditions that btn 1) always bets when checked to and 2) raise/calls when facing a bet w/ [>=tp, fd, oesd, >=1p+gs]. obviously, other lines (e.g., checking behind, calling or raise/folding) could be better in many of these spots. even so, the (unsurprising) point stands that many factors other than board texture have large impacts on the ev of any particular play. in this scenario, the c-betting tendencies of the bb have a huge impact on the profitability of raising.
other comments on the analysis of rr range article:
  1. the big message for my nit-brain here
    is that most 3-betting ranges are not all that susceptible to flop
    bluff raises. so, i can basically c-bet a ton of missed flops and quit
    worrying about being exploited when i have to fold to a raise. unless,
    of course, villain has a peculiar 3-bet calling range...
  2. bluff-raising K-hi flops IP is basically printing v. all but the tightest 3-bettors.
  3. forget about bluff-raising Q-hi, 2-broadway and monotone boards (for the most part).
  4. i feel that floating is (or could be) a big part of these situations
    with some players/dynamics, albeit a much more complicated part. it may
    be tough to address this neatly, as one would have to think about
    pfrr's flop and turn c-bet frequencies, potential for caller to
    improve, etc. i suppose this would get messy and may be difficult to
    use anyway, but might still be worth trying...
  5. an obvious point that was mentioned in the text is that c-betting
    tendencies vary widely from player to player. it might be possible to
    do some more analysis and try to isolate this variable?
  6. one other problem is that from the bluffer's perspective, there is no
    account for 3-bet re-bluffs. maybe this doesn't come up too often, but if we use this article as a guide, an observant opponent could certainly exploit us.
  7. there is also no account for reads, recent history, metagame, etc. i guess i likely overrate that shit anyways...
  8. the assumption that players are folding 'weak' hands may be increasingly questionable in today's games. for example, i've had small raises called/re-raised by JJ on Kxx 2-tone boards 2x recently by 'good regs'.