This thread is for me as much as it is for any of you. Blind stealing is a part of my game that I have been working on and I need to work on. To boost my confidence in raising pf with rags in order to steal blinds, I have done some work from a mathematical perspective. I hope some of you get some value from it.

PROFILES:

I) We'll call this guy Tightish (T). The range of hands he will typically call or raise us with is such:
Pocket pairs 77+; AJ+, KQ; A8s+, KJs, QJs, JTs
(you may disagree with this profiled range of hands, but the following is the important part)
This range of unique hands totals 136. There are 1326 total unique starting hands.
T will call/raise with 10% of possible starting hands.

II) We'll call this guy Shortish (S). The range of hands he will typically call or raise us with is such:
22+, A2+, KT+, QJs, QTs, JTs
I call him Shortish, because this guy might be somewhat short stacked. A situation in which he needs to make a stand soon, but isn't forced to right now (any pair, any ace mentality at this point). This guy could also just be a player looser than T, but tighter than the next profile.
Anyway, this range of unique hands totals 330.
S will call/raise with 25% of possible starting hands.

III) We'll call this guy Loose-ish (L). He's your ever so common donkey that plays way too many hands. The range of hands he will typically call or raise us with is such:
22+, A2+, any broadway, K8s+, Q9s, J9s, T9s, 98s, 87s, 76s
This range of unique hands totals 398.
L will call/raise with 30% of possible starting hands.

-------------------------------------------------------------------------

When faced with a standard raise...

T calls or raises 10% and folds 90%
S calls or raises 25% and folds 75%
L calls or raises 30% and folds 70%

Raising, first in, from the Button:

Based on the above %s, when is this likely to work? Well there are 2 left to act, who are they?

6 very general situations: TT (two tightish players), SS, LL, TS, TL, SL

Here is the % chance of a successful steal (i.e. they both fold) vs. each combo:
TT: 81%
TS: 67%
TL: 60%
SS: 56%
SL: 52%
LL: 49%
--------------------------------------------------------------------------

Raising, first in, from the cutoff:

TTT: 73%
TTS: 61%
TTL: 57%
TSS: 50%
TLL: 44%
SSS: 42%
SSL: 39%
LLS: 37%
LLL: 34%
-------------------------------------------------------------------------

This is very basic analysis. We are ignoring the possibilty that they recognize a blatant steal. We are strictly going by predetermined ranges of hands that will actually defend. Instinct, timing and table image play a crucial role in blind stealing. I'm just setting a mathematical foundation for it.

Basically, as you can see, a simple steal from the Button has a high likelihood of success. It's even just under 50% against 2 VPIP 30% players. Steals from the CO can be reliable against the right players.

Also, one variable I didn't mention - your cards. That's because they don't matter here!

A closer look at a common STT situation:

Blinds 100/200 Pot = 300

If you raise 3BB to 600 you:
a) win 300
b) lose 600
(for simplicity I'm ignoring further action)

How often do you need to be successful to be EV+ in the long run?

Breakeven success rate = 67%

From above, we know this works against:

TT 81%
TS 67%
TTT 73%
-------------------------------------------------------------------------

Add in the antes...

6 players left; Blinds 100/200; Antes = 25; Pot = 450

Standard raise to 600
a) win 450
b) lose 600

Breakeven success rate = 57%

From Button:
TT 81%
TS 67%
TL 60%
SS 56%
From CO:
TTT 73%
TTS: 61%
TTL: 57%
Other (can you say bubble time... table's tight):
TTTT: 66% (UTG+1)
TTTTT: 59% (UTG)
------------------------------------------------------------------------

Generally, you can figure out your Breakeven success rate (B) like this:

x = pot
y = raise amount
B = y/(x+y)

MTT 8 at table; 200/400 blinds; 50 ante; pot = 1000
raise to 1000

B = 1000/(1000 + 1000) = 50%

From Button:
TT 81%
TS 67%
TL 60%
SS 56%
SL 52%
From CO:
TTT 73%
TTS: 61%
TTL: 57%
TSS: 50%
Other (can you say bubble time... table's tight):
TTTT: 66%
TTTTT: 59%
TTTS: 55%
TTTTTT: 53%
TTTL: 51%

I'm in a hurry this morning so I'll post more on this later, but just wanted to say this is an excellent analysis.

I like to raise 2.5 bb for the simple fact that it almost always has the same effect as a 3bb raise (in fact against certain weak opponents it seems scarier) and reduces the breakeven point to 60% (no antes). In fact, I no longer raise to any other amount that 2.5 bb after the beginning of a tournament or SNG (unless that won't pick up pots, then I'll do the min necessary, but if 2.5 bb won't pick up a pot then you're likely at a loose table and should shift gears).

And one thing to take into account is calling equity (your equity when you do get a call). If you'd like you can run this through pokerstove (take a random hand vs. the calling range and assume 1 caller the % of the time only one has a hand they can call with, 2 callers the rest of the time).

NOTE: I said I don't raise to any amount above 2.5 bb in the previous post . . . what I meant was outside the blinds I don't raise to any other amount. In SB vs. BB I'll usually raise to 5 x bb.

I think YP's analysis demonstrates why it's most important to observe the three players to your left (and one to the right) . . . in live tourney play I try and keep track of how many pots the 2nd on my left and 3rd on my left play (ignoring the details of the player on my immediate left, who will be the SB on my button and button on my CO and whom I can usually play by feel) . . . . in online play pokertracker keeps track of everybody for me.

In online play, the easiest thing to do is just multiply the two opponents 1-VPIP and then also do it for the third to get the stealing success rate you need . . . this will often tell you if your raise with junk can be successful, or if you're going to need something of a hand to come at them.

For instance:
Player to immediate left VPIP: 25%
Player 2 to left VPIP: 15%
Player 3 to left VPIP: 40%

So on my button, my stealing success rate is:

(1-.25) * (1-.15) = 64%

In the cutoff, my stealing success rate is:

.64 * (1-.4) = 38%

So a raise from the button to 2.5 bb will succeed (no antes), while a raise from the CO with the same junk is ill advised, and thus we should tighten up our stealing requirements in the CO to hands like suited connectors and hands that are much more likely to have equity on the flop like Ax, Kx, JT/QT/QJ, Suited connectors, and 1-gap suited connectors (maybe add T6s/T7s/T8s) and any pair.

So this is the practical application of YP's posts above . . .

Originally Posted by Jason75

I think YP's analysis demonstrates why it's most important to observe the three players to your left (and one to the right)

In online play, the easiest thing to do is just multiply the two opponents 1-VPIP and then also do it for the third to get the stealing success rate you need . . . this will often tell you if your raise with junk can be successful, or if you're going to need something of a hand to come at them.

So a raise from the button to 2.5 bb will succeed (no antes), while a raise from the CO with the same junk is ill advised, and thus we should tighten up our stealing requirements in the CO to hands like suited connectors and hands that are much more likely to have equity on the flop like Ax, Kx, JT/QT/QJ, Suited connectors, and 1-gap suited connectors (maybe add T6s/T7s/T8s) and any pair.

So this is the practical application of YP's posts above . . .

Well said. Let's chart some things out:

Raise Amount - Breakeven success rate (no antes)
min (2bb) - 57%
2.5bb - 62%
3bb - 67%
3.5bb - 70%
4bb - 72%
5bb - 77%

Remember: B = y/(x+y)
x = 1.5bb with no antes

And here is the B with 1 limper (2.5bb in pot) - (and B when stealing from the SB with 1 limper)
min (2bb) - 44% - 38%
2.5bb - 50% - 44%
3bb - 55% - 50%
3.5bb - 58% - 55%
4bb - 61% - 58%
5bb - 67% - 61%

Here's an interesting philosophical and/or math question . . .

When do the blinds become stealable in a tournament?

I used to think that it was at the 5% mark, that is, that when the blinds were 5% of the average stack size it becomes profitable to steal them.

I'm starting to think that mathematically it's profitable to steal the pots at any level . . . .

The key for me is that I must understand the PF play of the 3 players to my left. This takes at least 20 hands online with pokertracker (IMO), longer in live play (because I lack pokertracker).

But really, I only have a gut instinct to tell me when to go after them . . . .

This is a really great thread, thanks.

My favourite steal? UTG with a minimum raise. Now only will this work a lot more than you can imagine but it's also cheap when it fails. You'd rather have a tight opponent in the BB when your try this but even if the BB happens to call because he's getting the right price (as he should be doing with most hands,) then you'll have position on him/her anyways.

Another important player profile (IMO) would be the very tight player, that'll reraise with less than 5% of hands (prob only QQ+, AK. Calling is unlikely here, since we can crush our opponent and he knows it)

This gets really important if we are the big stack, now steals from mid position may be profitable, if there's no maniac or short stack in the way.

Another interesting big stack problem is 3-handed. Should you always stick in a raise from the button here? Even if the others know exactly what you're doing, it's not easy for them to defend, the shortest stack will probably wait for at least Acex.