2007PBWC NLH FinalTable - Bot mistake? or misfortune?

[RIIT]

This HH is from the 2007 PokerBot World Championship NLH final table.

The players:
$1 chair - AsbakAlpha - a Canadian bot (owner: professional programmer; age: 30+)
$10 chair - Guaran - a student (age: 19)
$100 chair - HTC - a veteran WSOP dealer (age: 40+)

At only 18 hands into the event, the bot made a very aggressive push with A9o. Guaran stayed in with KK and won (he was a 60% favorite).

Asbak, (the bot owner), said that his bot was primed for regular SNG play and so at 3 handed it believed it was ITM and is coded for increased aggression at that point.

The bot saw 40% chances preflop with A9o against all possible hands and risk of ruin was 46%.

Is this a case of bad luck or bad play?

Hand# 2740 2007-09-28 18:14:12 GMT PBWC-Valor
Table# 3 Table3 PlayMoney 1 2 2 NoLimit Holdem
Guaran Chair8 1051.50
HTC Chair2 990
AsbakAlpha Chair5 958.50 BUTTON
Guaran SBLIND 1
HTC BBLIND 2
DEALER DEAL
Guaran HAND Kc Ks
HTC HAND 8h Ts
AsbakAlpha HAND 9d As
AsbakAlpha RAISE 6
Guaran RAISE 10
HTC FOLD
AsbakAlpha RAISE 952.50 ALLIN
Guaran RAISE 1040.50 ALLIN
DEALER FLOP Qs 7s 2c
DEALER TURN 5d
DEALER RIVER 9s
Guaran POKER KKQ97o One Pair - Kings - Queen Nine Seven
AsbakAlpha POKER 99AQ7o One Pair - Nines - Ace Queen Seven
DEALER POT 2012
Guaran RFND 93
Guaran WIN 1919

RIIT

[Slim Pickens]

Quickly, what is this $1/$10/$100 "chair" business?

That push with A9o is pretty bad. You haven't mentioned any payouts, but even in the best case for explaining it (that would be 100% to the winner), shoving any hand for close to 10x it's S-C value is very bad unless your opponent is so awful that it could figure out how to fold AA there. The shove doesn't even make sense if you know he'll only call with exactly AA and fold everything else.

Even the call with KK there is pretty exploitable.

What sort of min-reraising range would you expect most people to give their bots? I would think (not that I know anything about making bots, just from a logical perspective) that unless there's a fair amount of random min-reraising in your algorithm (which is hard to make very profitable anyway), that min-reraise would rightly trigger a lot of ZOMG AA/KK set mining, especially with stacks that deep.

[RIIT]

[ QUOTE ]
Quickly, what is this $1/$10/$100 "chair" business?

That push with A9o is pretty bad. You haven't mentioned any payouts, but even in the best case for explaining it (that would be 100% to the winner), shoving any hand for close to 10x it's S-C value is very bad unless your opponent is so awful that it could figure out how to fold AA there. The shove doesn't even make sense if you know he'll only call with exactly AA and fold everything else.

Even the call with KK there is pretty exploitable.

What sort of min-reraising range would you expect most people to give their bots? I would think (not that I know anything about making bots, just from a logical perspective) that unless there's a fair amount of random min-reraising in your algorithm (which is hard to make very profitable anyway), that min-reraise would rightly trigger a lot of ZOMG AA/KK set mining, especially with stacks that deep.

[/ QUOTE ]

Slim,

Find the diagram of the final table structure here:
2007 PBWC TOURNAMENT STRUCTURE

They are playing for a trophy ring at this point. The cash had already been awarded in the qualifying events.

The PBWC awards final table chairs and stacks to winners of qualifying MTT events. There is no entry fee for the final table - you have to win a qualifying MTT.

The final table chairs have increasing qualifying values based on the entry fee of the MTT qualifier: $1, $10, $100, $1k, ... , etc.

There were no entries at the $1k level this first year otherwise there would have been a fourth player in the $1k chair (or a second stack for AsbakAlpha or Guaran or HTC had at least one of them been willing to throw down for the $1k chair/stack - HTC almost did).

If you look at the results for the other events you'll see that HTC had 2 stacks at the FLH final table and Tyche had 2 stacks at the PLH final table because they each won 2 of the qualifiers.

2007 PBWC RESULTS

-----
As to the wisdom of the bots decision, I think it's a case of bad luck for a bluffable hand. Bot's have to be willing to bluff at some point otherwise they become too readable. Guaran probably would have folded any other hand except AA or QQ.

RIIT

[Slim Pickens]

This is effectively a winner-take-all structure. The action looked like this, right?

Blinds 1/2
AsbakAlpha (t958.50)
Guaran (t1051.50)
HTC (t990)

Preflop: Dealt to AsbakAlpha 9[img]/images/graemlins/diamond.gif[/img] A[img]/images/graemlins/spade.gif[/img]

<font color="red">AsbakAlpha raises to t6</font>, <font color="red">Guaran raises to t16</font>, <font color="grey">HTC folds</font>, <font color="red">AsbakAlpha raises all-in to t958.50</font>, Guaran calls.

AsbakAlpha is shoving almost 950 into a pot of 22. What kind of hand range do you think Guaran's standard reraise from the SB implies? How much of that range is he folding to a push?

[HajiShirazu]

It was just a sick, sick cooler, only 500 times the big blind with A9? No way you can get away from that there.

[RIIT]

Slim,

Yes it was a winner take all situation.

(one very minor clarification: the spends were 1 2 6 10 and so there was 19 in the pot before the bot pushed)

I agree with you if this had been a regular cash game where pushing A9o preflop after a reraise from the bb seems way too risky - as I said already, the bots ROR was already at 46% against all hands.

I'm totally speculating as to what Guaran would have done but I think it's safe to say he'd have folded his entire range except for AA - QQ.

They started with 1k big blinds and it's only the 18th hand; I'd tend to not view any preflop raise as necessarily representing an ultra-premium hand (although in this case it was and G. played it well).

RIIT

[Slim Pickens]

[ QUOTE ]
I agree with you if this had been a regular cash game where pushing A9o preflop after a reraise from the bb seems way too risky - as I said already, the bots ROR was already at 46% against all hands.

[/ QUOTE ]

Effectively, it's a cash game. ROR doesn't sound very important unless you think this bot's overall skill edge is huge, which I doubt it would be against other decent bots. If I know it'll call with enough hands such that A9o is +cEV against that range, then shove away.

[ QUOTE ]
I'm totally speculating as to what Guaran would have done but I think it's safe to say he'd have folded his entire range except for AA - QQ.

[/ QUOTE ]

Yeah... that's the easy part. The hard part is his opening range, but there's no way to answer the question without it.

Also, it's so cute that no one even attempted to buy the 1k seat.

[DevinLake]

RIIT, you already have your mind made up that the A9o bot just got unlucky....so why post?

Even though ur conclusion is wrong and it's a bad play.

[RIIT]

[ QUOTE ]
RIIT, you already have your mind made up that the A9o bot just got unlucky....so why post?

Even though ur conclusion is wrong and it's a bad play.

[/ QUOTE ]

Devin,

No I really don't have my mind made up. If this was an isolated case of the bot looking to create table image early on then it seems like some bad luck (i.e. the bluff has a reasonable chance of seeing everybody fold). If in fact the bot is willing to always make this same play then that seems like a mistake. There's no way to know this without hearing from the bot owner himself.

For me personally, I definitely would not have pushed.

RIIT

[DevinLake]

Well, what makes this a bad push is that, IMO, it does not have a reasonable chance of getting a fold. Certainly not reasonable enough to overbet shove over a min raise.

Like slim pointed out, what range of hands did this program see fit to min 3-bet?