www.cleverpiggy.com

I've played like 2500 hands against it and am down a lot, mostly because of tilt.

if you're tilting against what i assume to be a play money bot, then your HU game is the least of your worries

too many HU questions here, it's a ******ed format which is rightly being restricted in most decent rooms

No is not! Its pure poker! Maybe u suck @ HU?

WAT

And that bot is pretty awful, i played like a 100 hands, it might be worse than Governor of Poker bots, the flash game. If you can't beat it, you prolly have to work on your HU

Get out.

If sixfour isn't trolling, than I assume he's referring to the bum hunting nature of HU and how fast the fish go broke in it.

Those are valid issues to have with it.

Obviously as far as a pure test and skill and ability HU is a beautiful game.

But as far as protecting the poker ecosystem goes, I can see where he might have some issues with HU.

???

Are you talking about the limit hold'em bot or the NLHE, anyway?

ding ding ding ding ding we have a winner, but additionally I'll never get why certain people feel the need to play formats with less and less players just because they're impatient for action

It's very +EV to start fresh tables. Starting tables will mean playing HUNL at a decent frequency i.e. when one person joins your table. Therefore it is well worth understanding HUNL (and shorthanded) strategy.

Even if you don't start tables, HUNL gives you a better understanding of how to defend blinds against a wide BU range, especially postflop.

How about some people just like playing HU?

It's the NLHE bot unfortunately which is the worser bot. It's just sickening to know that if I make a sizeable bet on the river that it already knows if it's gonna call or not based on a formula, system, and rules someone else created to decide if I'm bluffing or not and that it doesn't actually affect him whether he wins or loses. Playing against other players I always feel the tension a bet or something can create and getting snapcalled is just like god what the ****.

I liked playing double or nothings, doesn't mean any sensible site should offer them

I played one hand against it. I turned the nuts, overbet 3x pot and it folded ace high. Seems solid.

EDIT. Played a few more hands. Betting 9x pot on 9753 made it fold 82. It's such a nit.

Only reason for this sort of thing is collusion, though. If collusion could be stopped, many sites would offer them. Really leery about "games with high winrates shouldn't be spread." Granted, I think HU Cash should have some forced action stipulations (this is why husngs seem fine to me as well, because you can't just sit all day and drain money from fish) and why PS Zoom HU seems like a good idea (except for the rake). But these seem like minor things that could be solved as opposed to deep structural problems of HU.

Also, for me (and many players) it's not really about action, it's about the ranges being much more interesting.

Are you being sincere?

The bot was programmed by Alan Cunningham and is a decent opponent.

Your Bankroll starts with 100k, and you're playing with a 20k stack (200BB) that refreshes every hand.

Anyone who says it sucks, show me a screenshot of you at 300k, a feat I've been unable to accomplish.

Alan says the bot does not adjust to opponents.

I assume the bot uses Minimum Defense Frequency to decide whether to call.

Let's say it went check/check on the turn and flop. Then you bluff the river for pot. The bot may call a (sizable bet) with just a weak pair, because it's in the top 50% of hands it has in that spot.

If you bet half pot flop and turn, then potted the river. The bot should only call with the best 22% of it's range (it takes to the flop)

0.666 x 0.666 x 0.5 = 0.222

Example:

I opened 20 buttons in a row for 3x
CleverPiggy folded 6 times, called 10 times, and 3bet 4 times.

If I never 4bet, it's continuing to the flop 14/20 (70% of the time)
with an estimated range of:
22+, A2s+, A2o+, K2s+, K2o+, Q2s+, Q3o+, J2s+, J5o+, T2s+, T6o+, 93s+, 96o+, 84s+, 86o+, 74s+, 76o, 63s+, 53s+

Flop comes [952], no flush draw

I c-bet half pot, clever piggy continues with 66% of hands, including:
-every pocket pair
-every 9-x, 5-x, A2, K2
-every A-x, K-x,
-QJ

Turn comes a T, still no flush draw [953 T]

I 2-barrel half pot again:

bot continues with top 66% of remaining hands

22+
A9+, A5, A3
KT, K9, K5, K3
Q9, Q5, Q3
J9, J5, J3s
T9, T5s, T3s
every 9 (down to 93s, 96o)

totaling a massive 65 hands!

River pairs a 3. [953 T 3]
You bluff pot.

Clever piggy calls 99+, 55, 33
AT, A9, A3,
KT, K9, K3,
Q9, Q3, J9, J3s,
T9, T5s, T3s,
98, 95s, 93s,
63s, 53s

That's 37 hands!
How often did you have 37 different hands you'd call a pot sized bet with on the river??


Looking at a final board of [953 T 3]
The worst hand it might call with is 98, but it'll fold 97.

Had you bet pot on the turn instead of 1/2 pot, then potted river:
Bot is now folding J9 and sometimes Q9.

You may (without analyzing it) think the Bot is stupid for folding Q9 on a board of [953 T 3] one hand, and calling 98 on another.

How do you know it uses MDF?

If that’s how it decides how often to continue that’s not a great strategy.

It seems to only use 75% or 200% sizings post.

It's heads up where ranges are super-wide. On that runout, anything that was top pair on the flop will serve as a decent bluff-catcher.

I doubt the bot just uses MDF and some hand-strength rankings, but I haven't read all the literature.

Edit: FWIW, Cleverpiggy is apparently a fixed limit bot. The NLH version is called Hibiscus Biscuit.

Figuring out a bot's game is the same as figuring out a humans game without the variance. once a leak is found it becomes relatively profitable.

FWIW I thought the Hibiscus bot was tough. I played 200 hands with it and barely broke even. Most bots have some pretty glaring weaknesses, but this one seems pretty well balanced so far.