Hello

I think I might have a leak in my game but not sure maybe you guys can help me.

One thing that is good to know is I play mostly shallow stack and cap games this might have some effect on the stats.

filter 2/4-10/20 98% shallow stack and 2/4-10/20 cap

flop Cbet% 72% flop cbet success 43% flop agg% 42.2
turn cbets % 39.8 turn cbet success 43.9 turn agg% 19.9
river cbet % 75.2 river cbet success 51.1% river agg % 13.2


please ask for any more stats if necessary.

Thanks

nah, I'm saying the starting point is -300bb/100 so it might be as bad as you think.

I don't know of good stats for calling 3 bets, it probably varies quite a bit depending on how much ppl 4 bet vs flat their big winning hands like AA-KK. mine is -208bb/100 without including AA/KK/AK for last year and 64bb/100 including.


that's at mostly opening for 2.5bb

:/

Ahhhhh gotcha!! Now this makes sense. I was thinking the starting point would be 0ptbb, which makes my winrate absurdly bad (although last 100k hands it still is pretty bad).

Thanks for the help. Anyone else feel free to chime in, i wouldn't mind hearing some stnadard winrates for this

A "normal" win rate for calling 3 bets in position doesn't really exist; it depends a lot on the range you call with.

A normal range of win rates for the button when flatting 3 bets would run from -.8ptbb/hand for somebody who folds to about 50% of 3 bets to +2ptbb/hand for somebody who folds about 75% of the time he is 3 bet OTB.

As has already been pointed out, your baseline is -1.5ptbb/hand, so -.8ptbb is a good loss mitigation strategy. The guy who is winning at 2ptbb/hand when he folds 75% of the time is realizing 50% more 1.5ptbb losses than the guy with a -.8ptbb win rate who folds 50%; so you can't say one strategy is better than the other. The idea is simply to minimize the losses you take when facing a 3 bet by whatever YOUR optimal mix of calls, 4 bets and folds is.

IMO this stat is not that useful to compare between regs because there are people who rarely flat their premiums (so their 3bet flat win rate will be higher). Also a function of how wide you defend vs a 3bet, as someone mentioned. I think looking at "faced preflop 3bet is better" if you want to compare numbers.

So then does anyone have a baseline for what a decent number is for faced preflop 3bet? obviously anything better than -150BB/100 is better than folding but I'm not sure what I'm aiming for..I'm around -90BB/100 seems bad? maybe not? no clue

Actually I wrote my post backwards, but you all know what I meant. Higher was supposed to be lower

I'm around there too. not sure it's a leak since my win rate when I 3net is much higher.

The computer I have HEM on crashed the other day, so I had to run this filter in my laptop running PT3.

One difference between HEM and PT3 is that PT3 counts you as facing a 3 bet regardless of whether you have volunteered to put money in the pot yet, while I am pretty sure that HEM only counts you as facing a 3 bet when you have already put money in the pot.

So what I did was filter in PT3 for VPIP and Faced 3 Bet. Pretty sure this is the equivalent of Faced Preflop 3 Bet = true in HEM.

Here are my stats over the last 160,000 hands of NL $200 (mainly Rush):

Overall: 2368 hands -0.56 PTBB/Hand

There are rounding errors in some of these stats:

Button: 441 hands, +1.23PTBB/hand Fold: 56%, Call 33%, 4 Bet12%
Cutoff: 506 hands, -0.66 PTBB/hand Fold: 57% Call: 26%, 4 Bet: 17%

These stats are a little out of line for the fact that I call so often. There could be some run good going on here, but I also consider this to be a strong point in my game, and I have always had a better than average performance in this spot. I would expect most players who only fold 55% of the time to average being about break even in this spot, or to have a slightly negative win rate.

Note also: I steal from the CO and Btn with a 2.5bb steal, so my baseline here is 1.25ptbb/hand. From all other positions my standard raise is 3bb/1.5ptbb/hand.

Position off the button:

2: 317 hands -1.75ptbb/hand Fold: 50% Call 33% 4 bet 17%
3: 257 hands, -3.03ptbb/hand Fold: 51% Call: 33% 4 bet 15%
4: 248 hands, -0.57ptbb/hand Fold: 50% Call: 33% 4Bet 16%
5: 196 hands, +0.49ptbb/hand Fold: 47% Call 36% 4 Bet 17%
6. 190 hands, -0.37ptbb/hand Fold: 46% Call 37% 4 Bet: 17%
BB: 73 hands, -0.73ptbb/hand Fold: 15% Call 41: 4Bet: 44%
SB: 140 hands, -0.22ptbb/hand Fold 39% Call 24% 4 Bet 37%

As you can see, I have a massive leak in middle position; I definitely call too much. It's a constant struggle for me to fold anything reasonable when I will be in position relative to the 3 bettor.

One thing to keep in mind is that I usually play about a 13/10 style, so the fact that I am folding only 50% of the time is also a function of the fact that I have a pretty strong range to start with.

On the button, my WR when I call a 3 bet with a hand other than AA or KK is 4.10ptbb/100. In the small blind, my WR when I call with a hand other than AA or KK is .48ptbb/100.

In all other positions, I am losing money compared to folding when I call a 3 bet with a hand other than AA or KK.

What this means is that most of my loss mitigation actually occurs when I 4 bet bluff or 4 bet on the light side. My WR 4 betting hands other than AA or KK is -0.4ptbb/hand. My showdown line is hugely negative, so I am losing a lot of money with the bottom of my all-in range. But my redline is hugely positive.

Overall, I would say my performance in this spot is probably average. I'm definitely above average* in late position, but also getting my ass kicked in middle position. I've seen plenty of people who have positive win rates in early position, so I'm a bit under the average there as well.

These numbers should give you a reasonable basis for comparison. You'll have to adjust in your head for any difference between my preflop style and yours. Also, make sure that you break it down and look at your performance when you call a 3 bet, and when you 4 Bet. It's also worth filtering your 4 bets by starting hands so that you can see how you are doing when you 4 bet bluff and when you 4 bet for light value.

You should probably also analyze which hands in your calling range are making you money and which ones are losing you more money than if you had folded; but look at them as groups, because the sample size for any given hand will be too small to rely on.

*Average for solid winning regulars.

oh. whoops... I didn't realize you guys were talking ptbb. So mine is like -45ptbb, even tho I fold 64% to a 3bet. It's interesting because it feels like I give up on a lot of 3bet pots.

Mpethy,

What's a good winrate for UTG in six max?

11 or 12 bb/100, imo.

Mpethy or anyone with good stats from BB

im running at -38 bb/100 over a sample of several hundred thousand hands of 6m 200 nl rush, and relatively speaking this seems like my worst stat by position.

Have tried some standard filters eg [did cold call = true excluding TT=, AQ+] and [did 3b = true excluding TT=, AQ+] and the results look not too bad at all. So im wondering if there are some revealing other tests i can try from the BB. For instance:

For good stats from the BB, how much do you lose from the BB:
- versus single raise from UTG
- versus single raise from UTG+ 1
- versus single raise from CO
- versus single raise from D
- versus single raise from SB

Any other filters you could recommend that might be revealing for BB analysis?

I don't track stats for the filters you list. the reason for not doing so is that even in a huge overall sample, you'll be dealing with sample sizes that are so small that they'll easily be subject to influence by variance.

When analyzing a player's blind play, there are really only two filters that are necessary to expose most players' major leaks:

1. Cold Calling.

In HEM, I go to the Holecards report. On the main filters page I deselect all non-blind positions, and under more filters I apply "did cold call = true."

With this basic filter on, I look for trends in the player's winning and losing hands. Common leaks:

a. Sub-optimal performance with broadway hands.

In full ring, it is very common for players to have a losing win rate with broadways, or to be losing with some and winning with others, dragging down the win rate with these hands below what is optimal.

In six-max, it is less common to see an actual losing win rate with broadways, but it is very often the case that the win rate is below what it needs to be.

If I see a problem broadway hand or two, I will go to the holecards HE tab and select AK down through JTo and check the overall win rate. If there are problems here, you have no choice but to do hand history reviews among this group of hands to try to figure out what problems you have playing these hands.

b. Ragged Suited Aces:

Do the same analysis; this is a particularly important group to check in six max.

c. Suited connectors:

Ditto

d. Calling with Small Pocket Pairs (22-66)

What you have to do here is go to the holecards HE tab and select pocket pairs. Then, on the holecards report, you look at two things:

i. Win Rate Analysis.

Look for a pattern. Some players are losing with 22, 33 and 44, but win with 5s and 6s. More commonly, players' win rates within the group are random, determined simply by which hands in the group happened to have been paid off or set over setted most often.

If your win rates with these hands are all similar, or appear random, just analyze the hands as a group.

What is your win rate with the group? Any win rate better than -75bb/100 hands is preferable to folding, and the higher the better, ldo. But you should actually be in profit with these hands, and if you have a big sample with them and aren't winning at a decent rate, you are probably leaking (variance is always possible).

Assuming you have a leak, you should try to narrow down the precise situations that cause you trouble:

ii. Frequency Analysis

Compare your calling frequency with small pockets to your calling frequency with middle pockets. You should see a SUBSTANTIAL drop off in the frequency with which you call with small pockets. If you are losing and you don't see a drop off in calling frequency, that's your leak right there, most likely.

Try to determine which situations are causing you the most trouble. Go back to the main filters page and select position of first raiser is "early" and go back to the more filters tab and add the filter "players seeing flop is exactly___" and make it "2."

Check your results. If there is a problem, look at individual hand histories to make sure that it made sense for you to be calling that person to set mine OOP in a HU pot.

Repeat for MP, CO and Btn.

e. ZOMG, Pot Odds!!!

This is a VERY common leak. Something like this happens: You're in the BB with T6s, EP minraises, two people call, and when it gets to you, you are getting something like 7.5:1 on the BB you have to put in to see the flop. So you call, because, pot odds, ldo, maybe I'll flop trips.

Theoretically, the call is maybe justified. But some players (myself included) don't necessarily have a big enough edge in their game to play marginal hands OOP in a multi-way hand for a profit, and this winds up being a player specific leak.

If you have the tendency to call in spots like this, the clue you will see in your DB is a lot of random hands show up with low frequencies (just a few calls each) but amounting to a significant percentage of your calls.

The existence of a lot of these one-off calls in and of itself is not a leak. Some players have the edge necessary to make these calls profitably. If you see a lot of one offs with trash hands, you have no choice but to write down all of the starting hands you called with that fit the profile, go back to the holecards HE tab, and input all of them (because of the leak discussed below, you can't just run a filter for all hands other than your standard calling range).

So filter for all these hands, and check your win rate. If it is better than -.75bb/hand, your loose, speculative pot odds calls are justified, and keep on keepin' on. If your win rate is worse than that, one of two explanations is likely true:

i. Your sample size is too small to rely on.

An indication that this is the case will be if your winrate appears kind of absurdly high or absurdly low. This is a marginal spot from which we would hope to extract a small profit, or to suffer a small loss. If you're crushing or getting crushed, it is likely the result of run good/run bad. The closer your win rate is to break even, the more confidence you can have that it is accurate.

ii. Your skill edge is not sufficient to justify these calls.

If you have a decent sample of these calls and your win rate is, say, -1.3bb/hand, it is probably close to accurate, and you should probably take these loose calls out of your game until you have significantly improved your post flop skills.

f. ZOMG! a Fish!

This is another very common leak. basically, what happens here is that you overestimate your skill edge against a fish and assume that you can play trash profitably against them.

I usually just stumble across this leak in doing win rate analysis and hand history reviews.

If I were going to look for it specifically, I would filter for the group of hands that lies just outside your standard flat calling range, and look at the individual hand histories to see if fish are over represented in the sample of hands as the preflop raiser.

If they are, and you are losing with these hands, this is very likely what is going on--you just have a tendency to dip down too deep in calling against them. Tighten up a bit.

2. Three Betting Light.

The single most common 3 betting leak people have at 6 max is 3 betting the wrong hands against the wrong people. This gets them into all sorts of nasty, -EV spots post flop that exacerbates the leak.

Here is how you go about determining whether you have a 3 betting light leak:

a. Do you have a 3 betting leak (from the blinds, for this discussion)?

Go to the position report. then go to filters and add "3 bet preflop = true" Then go to the holecards HE tab, and select all cards, then deselect AA, KK, QQ and AK (both).

Look at your blind win rates. The following should all be true in a big sample:

i. You should have more raises from the BB than the SB.

ii. Your SB WR should be higher than your BB win rate.

iii. Your win rates should average about 1.2bb/hand.

Higher is better, lol, and the further you are below 1.2bb/hand, the bigger your potential leak (obviously variance is always a complete to partial explanation for any win rate in a small sample).

If you have a leak, here is how to figure out whether it is a preflop leak or a post flop leak. As I said earlier, the most common leak is 3 betting the wrong hands against the wrong people. So here is how you look for that one:

b. 3 Betting the Wrong Hand Against the Wrong People:

Basically what we are doing here is looking for your understanding of fold equity.

i. Filter for your pure bluff 3 betting hands.

Go to the holecards report and note all of the trash starting hands that you 3 bet in the sample. You know, the J5o, the 720, the 53s type hands that you 3 bet.

ii. Check your win rate.

If your win rate with these hands is below the benchmark 1.2bb/hand, you may have a leak.

ii. Do a Hand History review of these hands Analyzing Villain stats.

This isn't nearly as tedious as it sounds; it is actually kind of fun. In the hand display window below your stats, click on to highlight the first hand, right click on it, and click "replay all hands"

In the replayer, make sure villain stats are displayed from your HUD and then look at the fold to 3 bet and fold to flop c-bet stats on all of the villains (add them to your HUD if you don't have them on there already--if you don't have them on there already, their absence is probably your leak!)

What you are looking for here is confirmation that your plan makes sense. Every hand you see where you 3 bet trash against a villain who didn't have a high fold to 3 bet or a high fold to flop c-bet is almost certainly a bad decision. You should see a a parade of players whose stats are high for one or both of these -- either he folds preflop a lot, or he folds postflop a lot, or you made a mistake by bluffing in a spot where you didn't have sufficient fold equity.

iv. Filter for your light value 3 bets.

Depending on your habits, this range is probably stuff like AQ, AJ, KQ, TT, 99 and 88, maybe a few more hands in that range.

iv. Check your win rate.

If it is a problem, you may have a leak.

v. Cue them up in the replayer.

Check the stats of the villains you played the hands against. Are they people who will call the 3 bet with a range that is primarily comprised of hands you're ahead of? if so, good, but if you see a lot of questionable 3 bets here, this is likely the cause.

An interesting and useful exercise (and this one IS tedious, sorry) is to do this:

1. Pick a villain OTB on a steal who you 3 bet for light value.
2. Note his steal % (let's say it is 40%)
3. Note his fold to resteal % (let's say it is 70%).
4. Note that this tells us he continues with 12% of starting hands (30% of 40%).

We can assume that he will 4 bet his premiums most of the time for most people, so we eliminate those from his range. The remainder of those 12% are his calling range.

5. Run it through Poker Stove against each of the hands in your light value 3 betting range.

If none of the foregoing analysis has disclosed a leak (extraordinarily unlikely) then your primary leak in 3 bet pots is almost certainly in your post flop play.

c. Finding Post flop Leaks in 3 bet Pots

The easy place to start here is to go to your position page. If they are not currently displayed, go to your stat selection box and add flop c-bet %, turn c-bet %, flop c-bet success % and turn c-bet success %.

I have to generalize here, because people play differently. But basically the idea here is to check the profitability of your c-bets. If your average c-bet is 60% of the pot, you need a success rate of 40% to break even on the c-bet. So what you will do here is to figure out a guesstimate of what your average c-bet is, and figure out what success rate you need for it to break even. Then you look at your actual success rates to see whether they are at or above what you need them to be.

Assuming your success rates are below where you need them, then this is the explanation for why your win rate is not where you want it. But it is not the leak in your game.

To find the leak, there is no choice but to do hand history reviews, and look for patterns.

This is already tl;dr, so i won't go into that. But there is no short cut to improving your game; you just have to do the work and study the hands where you made mistakes. Maybe you are c-betting too much at people with low fold to c-bet percentages. Maybe you are value towning yourself when you catch a second best hand. It could be almost anything. If you are brutally honest with yourself, you'll probably already have an idea of what, if any, post flop leaks you have in 3 bet pots.

There is a TON more that I could write; this is maybe 1/3 of the analysis I routinely do for people who are leaking from the blinds.

Mpethy

Thank you for taking the time to write this up in detail, very helpful.

I don't follow this - my wr doing this is 0.6bb/hand. But if our win rate is as high as 1.2bb/ hand when we are 3 betting light (non premiums), isn't it a clear sign that we are not 3 betting enough trash?

Surely we should keep adding hands to our light 3b range until they are losing less than 1bb/ hand from the BB and 0.5bb/ hand from the SB? That is, assuming that 3b> calling, although that is v likely to be true with junk like T2s.

I don't think there is a correct answer to your question.

Good luck finding the sweet spot that minimizes blind losses. The only way to find it would be experimentally. While you were experimenting, you'd likely become exploitably unbalanced.

The 1.2bb/hand win rate reflects a compromise. It has you 3 betting a wide enough range of bluffs and semi-bluffs that you are marginally profitable. It should then be possible with experience to either a. increase your profitability with your bluffs and semi-bluffs or b. widen your range to bring your wr back down to around 1.2bb/hand.

But fundamentally, you are correct--you should be 3 betting the widest range that you can 3 bet for something around break even--for aggro 3 bettors, my empirical data suggests that the average winning reg strikes a balance at around 1.2bb/hand.

To look at it differently, it is irrelevant to our results whether we mitigate blind losses by either:

a. maximizing the number of hands we can play for less of a loss tan we would take folding; or

b. winning the same amount of money as that would save us by having a profitable range of bluffs and semi-bluffs.

Ideally, we would figure out which strategy gives us the greatest theoretical rebate on our blinds, and then pursue that strategy. Personally, I have no idea how one would even go about doing that, so I just get people to try to get the widest possible range they can play for a small profit.

Just to follow up on the precise question you asked:

Figuring this out would require that you have two reliably huge samples of light 3 bets.

You'd be looking for which range made you the most money overall.

Suppose in range 1 you have 100 hands and make 120 dollars.

If you add 30 hands and your win rate goes down to .60, then you have lost $30.

____________________

Also, the main reason why I wrote that section the way I did was because most people are 3 betting at a frequency that allows them to pull a decent profit from their bluffs and their semi-bluffs.

So I can say with a high level of confidence that if you are 3 betting around 5-8% of your blinds, and your win rate is lower than 1.2bb, then you are not mitigating you losses as much as other regs who are also in the 5-8 range, and this would suggest that you are leaking, and you have to play the hands more optimally before you can think about adding additional hands to your range.

Mpethy,

If I'm reading that right(and my calculations are correct), then an easy way to test whether adding more hands to your range beyond the 5%-8% range would be: (assuming fold/3bet preflop)

SB+BB = -1.5bb/hand ;starting blind loss rate
5%-8% range = 1.2bb/hand ;win-rate per 3 bet

3betlight%(winrate) = y - (3betlight%-1)*-1.5bb

5%(3bet per hundred) at 1.2bb/hand = +0.06bb/100 - (95% * -1.5)
8%(3bet per hundred) at 1.2bb/hand = +0.096bb/100 - (92% *-1.5)

so a solid reg light 3 betting strategy would yield +0.06bb/100 to +0.096bb/100 to win-rate per hundred.

To make it more accurate you could take out cold called preflop in % so it would sum to 100%.

Anything higher than that would mean you are achieving better than 1.2bb/100 on light 3 betting.

Does that look right?

3betlight% would = to the percentage with filter set to exclude AK/QQ-AA.

It looks right as a math problem, but, of course, you can't develop an optimal 3 betting strategy just by assuming a light 3 bet has X value, and then solving for N hands necessary to achieve Y reduction in your blind losses.

I know you didn't really ask this, but what TagWAG wrote really got under my skin and I have been thinking about it off and on all night.

Fundamentally I think he and I were talking past each other--he was talking about optimal strategy when he replied to my post by saying, "I don't get this," and in making the post I was focused on the process of identifying existing imperfections in an existing strategy (and I just assumed that as a solid 6 max player he was in the 5-8% range).

Theory and practice have to come together here to determine your individual optimal 3 betting strategy, because:

1. The enemy has a vote.

Suppose you lack the post flop skills necessary to play the 5-8% range for a profit of 1.2bb/hand, and you are only winning at .6bb/hand. So you think to yourself, "to achieve the blind loss mitigation effects that most 5-8% 3 bettors are achieving, I have to 3 bet 10-16%. You can't just go ahead and do it, because your opponents will adjust to a 3 betting strategy that extreme.

2. If you're doing it reasonably well already, you are picking the best spots.

Just adding 3 bets will require you to select more marginal spots which will have a lower +EV, potentially even a -EV. So you have two potential problems here: a diminishing returns issue related to the theoretical value of each situation and a skill issue that you might introduce a leak into your game if you add marginal situations.

Theoretically, there are two ways to make X more dollars: play the hands you are playing more optimally, and add more hands.

In my opinion, you have to optimize before you add. It's the same reason why you shouldn't play LAg until you have optimized TAg--you put yourself in spots where your EV is theoretically marginally profitable, but your skill renders the result as marginally -EV (this is why most small stakes LAgs have win rates essentially equivalent to small stakes TAgs).

When I see a 5-8% 3 bettor, and he isn't at 1.2bb/hand, what I normally see is that he isn't picking the best spots, and that is the explanation for why his win rate is lower than 1.2bb. His winrate isn't lower because he is playing additional marginal spots; by reference to the field of solid winning players, someone winning .6bb/hand simply isn't winning as much as the average solid winning regular with comparable 3 betting stats because he isn't optimizing his profits from those 5-8% of hands.

The thing that I said that got this whole discussion started was that the main leak people have in 3 betting light is that they 3 bet the wrong hands against the wrong people. They 3 bet air against people with low fold to 3 bet stats and they have a range weighted toward value against people who usually fold.

This is a very basic mistake*. No good will realistically come from compounding the problem by adding hands without first mastering the fundamentals. And most of us (I 100% include myself in this group) need to work a lot more on the basics of analyzing the situation than we need to worry about optimizing our frequency.

___________________________

*It's not really a basic mistake; it is a basic mistake as defined by reference to mistakes among solid winning regulars in the cut throat, intensely competitive world of small stakes internet poker. Even knowing to talk about it is a fairly advanced level of poker knowledge.

Mpethy

Over a decent sample (several hundred k hands of 6m $200 nl rush) it seems I am 3b (exc QQ+/AK) at 11% from the BB and making 0.6bb/hand when 3betting.

Tbh, I think anyone 3b non premiums and making as much as 1.2 bb/ hand is blatantly missing more marginal opportunities to 3b more from the blinds.

Taking a closer a closer look at my db I've come to the somewhat surprising conclusion that I should be 3betting a wider range of speculative hands from BB v EP and MP opens...... Also that calling with KQ is strangely unprofitable for me in all positions.

Also (non related) all over my db (nearly 1m hands) it seems KJ makes more/ loses less in similar situations than KQ.

Thanks Mpethy, v helpful.

Mpethy,

I'm leaking really badly from the blinds I think(six max).

I've tried changing my game. Gotten more aggro from the blinds. And wondered if you could look and see any obvious leaks/things. Or even if someone could post what good stats look like? I do occasionally spew large with bluffs, but any comprehensive stuff would be awesome.



Uploaded with ImageShack.us

@tagWag:

If you're at 11% from the blinds, then you probably should have a lower win rate than the one I quoted, which is standard for people in the 5-8% range, and which i have observed from dozens of small stakes six max regs.

The thing that interests me is who is making the most money. Could you post or PM me, whichever you prefer, a screenshot of your unfiltered blind win rates and a screenshot of your total winnings with your light 3 bets from the blinds? I'd appreciate it.

@Ace:

I'll take a look at you stats later; I have a session starting in a few minutes.

mpethy

pm sent with screenshots.

i'm quite unhappy with my results from the BB and i imagine you will think i am just playing too loose. There again, these days in 6max games so many players are stealing with 50% and more of their range that folding too much from the big blind is also going to be a mistake since folding costs us 100bb/100 without a fight...

i also suspect that both 3 betting or calling from the SB gets more respect than a 3b or call from the BB. Relatively speaking i think im doing better out the SB than the BB.

thanks, i'd be interested in your thoughts.