---

Poker Video: No Limit Hold'Em by terp (Micro/Small Stakes)

The audio quality makes me sad

07:20: "You don't want to be playing balanced... if you play balanced, you can't make money, you can't get your opponents to make mistakes when you are balanced. If your opponents are making mistakes on their own, you're not even profiting."

I don't think this can be true. If you opponents make mistakes then the money naturally flows to you. A lot of the profit in poker comes not from you actively exploiting your opponent but then just losing value or making poor calldowns. For example, if an opponent doesn't value bet enough on the river, you make money (by not getting value-towned correctly in a spot where you should) despite you don't nothing to exploit your opponent. Another is opponents playing too passively preflop (limping, coldcalling): we know this to be a generally losing strategy (side note: limping every hand preflop is a perfectly balanced strategy) regardless of how our opponents re-act.

How much we make from our opponents making mistakes over us actively exploiting them is a heavily debated. In the games i play (LHE), I believe a lot more money comes from our opponents making natural errors over us exploiting them. But regardless of the game, I'm sure that if a perfect GTO bot existed, it would crush both against strong and weak opponents.

19:20 - in the discussion in roshambo you kind of expand on this idea. Certainly in the case here if we don't exploit our opponent then we still have a breakeven expectation; but this is a simplistic grid and it's just not the case in poker.

A simplistic example: Let's say we bet on the river and we strive for balance so we have alpha combos of value bets and 1-alpha combos of bluffs: this makes him indifferent to calling with bluffcatchers (hands that lose to all valuebets but beat all bluffs). But that doesn't mean we don't make money on this bet. For example, if he calls us with a hand worse than some of our bluffs, we make money. Or if he doesn't value-raise correctly.

A more subtle example: Say the rolls are reversed. If our opponent bets in to us then we call with alpha bluff catchers and folder the bottom part of our range: this ensures he is indifferent to bluffing. But we still might make money if he doesn't value-bet enough. Or doesn't valuebet a sensible amount.

32:10: Re. the discussion of overbetting: "A common misconception... if you want to be balanced your overbets should be stronger... the funny thing is, small bets correspond to strong value ranges and large bets correspond to very bluffy ranges."


I had to stop and do some quick maths of this to prove what you said was correct! This is a really good point and probably worthy of some maths. If my quit calc is right, if you are half pot size betting on the river, you should have a 75% value range (25% bluffs). If you are 2x potting on the river, you should only have a 40% value range and 60% bluff range.

32:10: Re. the discussion of overbetting: "A common misconception... if you want to be balanced your overbets should be stronger... the funny thing is, small bets correspond to strong value ranges and large bets correspond to very bluffy ranges."


I had to stop and do some quick maths of this to prove what you said was correct! This is a really good point and probably worthy of some maths. If my quit calc is right, if you are half pot size betting on the river, you should have a 75% value range (25% bluffs). If you are 2x potting on the river, you should only have a 40% value range and 60% bluff range.

Ignore my maths above it's off and you cover it later in vid.

We can mix bet sizes remain balanced right?
For the # game, 55:00ish:
2PSB: 20, 30, 100
1/2PSB: 40, 96, 98
Check: 75, 80, 85
Is still a balanced strategy?

Time Link to 00:08:28

90 mph = 144.84096 kph

Been looking forward to more of your stuff terp.

Once I have watched it again or after another ep - I will try and give some presentation style criticisms that you asked for.

Thanks for your work.

Time Link to 00:56:42

What if on the 2PS bet we double pot 100% of our range. We are still balanced with 100, 98 and 96 as value hands and the other 6 as bluffs. Is this unbalanced because some of our better bluffs could be the same or better than some of his value hands?

Terp has rerecorded this episode and we are reproducing now, will post as soon as we can. Thank you for your patience.
-Rusty

In your Roshambo example, you assume Villain lost with Rock so why is he choosing scissors 2/3 time if he supposedly chooses the combo that beat him last time (must have been paper to beat Rock)2/3 of the time? Shouldn't he be choosing Paper 2/3 of the time and Scissors 1/3 of the time?

Secondly, Hero has higher EV by choosing the counter to whatever he chooses 2/3 of the time rather than randomnizing between his two choices since the other choice does not involve a Villain loss. This is because when Villain loses more often we can exploit him on the following contest assuming he does not adapt. Your EV calcs only include one round of betting instead of including the sequences that follow where Hero can exploit. One contest of betting is only sufficient for EV calcs when you can assume following contests are unpredictable/unknown.

Does that make sense? Or what am I missing?

(BTW I found audio fine if a bit eccentric).

Terp has rerecorded this episode and we are reproducing now, will post as soon as we can. Thank you for your patience.
-Rusty

sorry about that guys.

i also spotted the roshambo error rereading it and luckily fixed it quaddeuces is right on though.

sorry about that guys.

i also spotted the roshambo error rereading it and luckily fixed it quaddeuces is right on though.

At least you made me think. LOL.

07:20: "You don't want to be playing balanced... if you play balanced, you can't make money, you can't get your opponents to make mistakes when you are balanced. If your opponents are making mistakes on their own, you're not even profiting."

I don't think this can be true. If you opponents make mistakes then the money naturally flows to you. A lot of the profit in poker comes not from you actively exploiting your opponent but then just losing value or making poor calldowns. For example, if an opponent doesn't value bet enough on the river, you make money (by not getting value-towned correctly in a spot where you should) despite you don't nothing to exploit your opponent.

You kinda have a point, but it's more complicated than you think.

If we play fully balanced, we should call the river such a % of the time that villain will be indifferent between bluffing or giving up (same EV for him). Now, if villain doesn't valuebet us enough there, that means his bluffing ratio will be too high, and we're making a lot of incorrect folds. If we know he doesn't valuebet enough we adjust by bluffcatching more, but this creates an imbalance in our strategy as we're now calling too much, so villain can stop bluffing completely as they all become -EV now, and only valuebet us.

A GTO bot might make some money from fish, but it will be very close to 0 as the BB/100 you get from simply not bluffing the fish at all when you know he has top pair is WAY higher than betting a balanced range. And poker is also so complex that every strategy has counter, so the GTO bot could be beaten by a human opponent. You can program a GTO bot against a certain strategy, but the person employing that strategy could switch to something different at a moments notice, and there's no way for the GTO bot to know that he did it. Say we decide to only 3b with AA for a single hand, the GTO bot is likely to make mistakes against our flatting range and our 3b range alike. But there are way more complex possibilities than that.

Your claim would be true in a less complicated game, like 6bb deep HU, where there's a single optimal strategy and anything deviating from it will lose to the optimal strategy. There a fish would lose to the bot, and a human opponent could at best break even by employing exactly the same strategy. But that's definitely not the case for 100bb+ poker, and possibly not even for 20bb poker.