It's focusing on the biggest losing hands first. So you will have hands that are winners, but they typically won't be listed in that list (sometimes they will).

The other ranges are showing ranges that are possible to show profitability, but are difficult to sustain. The recommended ranges in step 2 focus on a slightly tight to slightly LAG style. You can play a more aggressive LAG style, and Leak Buster will show what ranges become too aggressive to have profitability. Those more aggressive ranges are very difficult to maintain long term, and this is why we show them, but don't initially recommend them. However, if it's something you want to explore, you can use Leak Buster as a guide to know where the edge of very aggressive games really start to fall apart. And if your stats are in those outer ranges, LB won't score you very poorly. It will just let you know that it's a difficult range to be in.

Glad you're enjoying.... GL on the virtual felt!

Agreed, and we had tried this, but the query time to produce this with the leaks got pretty long. It's perhaps something we can add as a setting option for those who want to see it. And it may be possible to have as default in the future as the DB becomes more refined.

Right above you someone had a similar question, and I answered it in my post just above this one.

Thanks for the reply FreakDaddy.

I'm still confused in step 4. Are you saying that the listed hands are ALWAYS the same because on average these are the most trouble/profitable hands? Because it's not in order of winrate nor how often I've had the hands.

It will always be the biggest losses shown. In the last update the developer put in a feature to switch between bb/100 or $, and in process apparently the sort order was messed up. It's already fixed, but won't be out in a release until Monday as far as I know. But in general, it's always just shown the biggest losses first (at the bottom of the thermometer).

Ah alright that makes sense now. Thanks!

You're welcome. I'll post in here when the update is out (likely Monday).

I don't use HM and I'm not going to. So, the price of Leak Buster for me is $80+$90=$170. Am I right?

Yes, unless you're playing micro stakes. Then it's 50+55.

New Features in 3.0.05d - Release date: 1/31/11

• Step 4 Data Sorting.
• Error on startup: Fix
• Date Filters - Apply to all (Added to all steps).
• Fix Step 9 - Opponents.

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

Cliffs notes: some suggestions for the developer. You can skip this if you're a customer.

Just purchased leak buster the other day and am evaluating it to see what I think. I thought I'd offer a couple of my initial impressions.

It strikes me how ambitious of an undertaking it really is to put something together like this, so first, hats off for finding a way to eat the whale bite by bite.

A couple things are just screwy about the recommendations and results. One thing in particular that sticks out is that the recommendations are almost always based off of whether you are outright winning or losing money. They ignore the impact of the blinds. So if folding in the blind costs me -100 BB/100 and leakbuster shows me losing -20BB/100 it'll tell me something like "you're losing money with this hand. Maybe you should play it a different way." Fact is I'm banking it with that hand and changing would be a very bad idea.

My 2nd biggest irk comes from the fact that I can't filter out hands I rarely play. There are lots of things I get dealt that I only play occasionally in Blind vs Blind, shorthanded situations, or vs opponents who have some specific hole in their game. This results in an incredibly small sample size for those hands and crazy BB/100 numbers. I would argue it's just not useful to look at my results for a hand unless I have a VPIP of say 60% (I just pulled that out of the air. don't focus on the # but the point behind it). I don't need to be warned that I am losing money with 34o when my VPIP with it is 10% or less. All it does is clutter up the displays, make it hard to tell from a glance how it is I ACTUALLY play (since nowhere in most steps is there any way to tell if you habitually play that hand).

The thermometer display would probably be WAY more useful with this VPIP filtering in place. It's really not significant that I played a single hand only twice and happened to lose both times. At a glance it appears to be a massive "leak" from the step's display. That's just not relevant or true.

If you ask yourself "what is the purpose of creating a graphical representation of the data in the first place?" you'd probably guess that it's to A) reveal something that would not have been obvious otherwise B) inform the user where they should direct their attention out of the nearly infinite candidates vying for review. At least according to these criteria, including hands that you barely play accomplishes neither of these goals (they're just not relevant). Worse yet, they clutter the display making it less useful and obscuring what value would be more apparent otherwise.

Finally, I'll say that I really liked Step 2 and it's representations of what optimal ranges were for various stats. Of course, some of them are a bit funky: how much control does one really have over flop cbet % success? Yes, I know that the width of your range pre and the sizing of your bets and the frequency with which you barrel all come into this, but especially over smaller samples it might as well just say "you run bad/you play loose opp's/you habitually ran into hands that could peel this sample" If I'm criticizing it then, would I change it? Probably not. Maybe I would include some of the above caveats clearly in the explanation of the stat. A confidence interval for how quickly this stat converges might be useful where it can be calculated. Still, explaining the concept to any newbies is probably more important than printing an exact CI itself.

Also, if you sell a version of LB for small stakes and higher stakes, then why is all of the analysis done off of 50 NL? The games change a lot when your opponents start reading hands better and certain holes you could drive a truck through become leaks if you continue to try to exploit them vs better opponents. There are some recommendations I'm outright ignoring from Leakbuster currently because I'm not convinced the data set is relevant to the games I'm playing in. It's a shame, because it's very possible I'm wrong, but I've got no real way or knowing. It'd even be nice to see the trend in how these ranges shift from low to mid stakes or over the years as games have gotten tougher.

What would truly be awesome (but lots of work), would be to do something similar to step 2 for the filters. Yes I can run 52 filters on my game and that's great (and actually where I think mot of the value of the product comes from), but that is an ENORMOUS amount of data to sift through entirely manually. If you could compare your results in those filters to that of other solid winning players it would make it obvious, which of those filters you should turn your attention to first.

This is my take after some initial use. Again, very impressed with the scope of the project and what you guys have done. Hope my suggestions are useful.

Thanks for taking the time to provide feedback. I'll try and address some of the things you said below.

All of the blind play analysis is done when you're voluntarily put money into the pot. So it's looking for hands you were involved in actively, thus discarding hands you folded without voluntarily putting money in. Also, the analysis for blind play takes into account (and actually analyzes) what the average losses are from the blinds. It doesn't expect you to be winning, and doesn't say this anywhere I'm aware of.

Again, if you are actively playing a hand, Leak Buster is going to show you your biggest losses. If you're folding say, 73o every time, it's not going to list this hand unless you actively play it. And if you are, you should be made aware that it's an issue imho. I understand what you're saying here, but I only think this would be an issue if you were looking at very small sample sizes. Perhaps you played a hand badly one time in 10k hands that you normally wouldn't play (like 34o) like you said. Then it would likely show up, because the sample of hands it has to compare against is small. But if you have 100k hands or more, then this shouldn't be an issue.

Trying to think of a way of what you're saying that would be useful as an option setting, but I'm not sure yet. Maybe you can clarify some more.

Understood. We try and illustrate this with the accuracy meter, and explain that 50k+ sample sizes are needed to get reasonable results. That's just kind of how it is in analyzing raw stats from a game w/ a decent amount of variance.

The analysis is done off of primarily 400NL-50NL. I'm not sure where you got the 50NL. Some of the advices and scoring change (not the stat ranges) for different stakes. We extensively analyzed stats that were winning ranges for say 200NL players, and saw how those related to winning players at say 25NL. The ranges were very similar except for a couple of specific areas. And this is why there's a range in there, and a scoring algorithm that adjusts based on what data you're analyzing. For 400NL, a much more aggressive 3-bet calling strategy is needed then say at 25NL. It takes this into account, but LB also shows you the possible range of profitable ranges, and how much more players win when inside of those ranges (some of the ranges outside of LB win even more). But we list the most profitable ranges that account for the least variance.

Also, since the initial release, a complete new analysis of the current games was done when version 3.0 was launched. So all of the stat ranges and data were updated to keep up with the current games. Additionally, the people involved in generating the advice and analysis are all very good long term winning mid and high stakes poker players. If you had a question about a specific leak, we always address it and are happy to help.

I'm one of the primary initial designers so I know the software very well, and I use it myself all the time. I follow a lot of my own advice that sometimes needs to be pointed out and set in my face.

It does this already though. There are 2 ways of looking at the filters. After a complete analysis you can look at step 1. It will sort the filters you're having the biggest issues with. And then within step 7 it will compare this info to other winning players. There may be value in adding the comparison +/- inside of step 1 as well, and it's something I can pass on to the development team.

Thanks, and keep in mind that we have more features planned. I know once LB Omaha is released, there's already plans on the tables for some really exciting data analysis.

Would you guys mind voting for my Leak Buster Video, in the Leak Buster Contest?

http://leakbustervideocontest.strutta.com/entries#96636

No problem. Thank you for a great reply!

Ok. Let me take a stab at this. Let's say someone has opened and action is to me in the BB, and I am considering two options: fold or CC. If I fold, then I will show -100BB/100. If I CC then we play the hand and see how we do. When I looked at some hands I cold call with in the blinds I show a 52bb/100 loss, which actually means I am making back 48bb's I would have otherwise forfeited by just folding outright.

I double checked this behavior by running a filter in normal HEM to tell me what my results were with a typical hand I cold call from the blind. The number was -52bb/100, which matched what showed in leakfinder. But the recommendation in leakfinder said "You're losing money cold calling, perhaps consider something else."

What you say makes abundant sense, but what am I missing here? Perhaps 3betting is still better than cold calling, but without a doubt in the situation above cold calling is far better than folding.


I'm not sure where we're missing each other on this one. I see a lot of hands in the thermometer that I rarely play. I wondered if by default you were showing hands in your recommended preflop ranges (I'm a bit tighter than you recommend). That didn't hold true. As an example, I have a very few examples where I OR T7s OTB. That's not part of your recommended chart, and I have a VPIP of about 15% in this spot. I probably don't need to see it and would rather focus on situations I'm routinely getting in. As it is I stare at the thermometer and have no quick visual way to distinguish small sample size and rarely played from what I commonly play and am leaking money on. The utility of presenting the data that way goes down because there's too much noise to signal in the graphic. The recent sample I'm using is 150K+ hands so, it's not THAT small.

It would just be nice to say "only display a hand in the thermometer or in this step if I VPIP with it 50% or more." My tired eyes would not have to hunt through to see what outrageous bb/100 #'s were volatility due to a 5 hands sample and what was a recurring leak that needs attention.

Yup, I noticed the accuracy meter. Great addition. This was a minor point of mine, and I'm guessing if we went back and forth we'd probably just be agreeing more.


Why isn't this more prominent in your marketing material!? It all sounds great and makes me feel a lot better. I have no idea where I heard the 50 NL, but for some reason I believed it. At the least I still didn't encounter any of the above, which certainly would have been a strong selling point.

Cool! Looks like I haven't discovered all the features yet. Just looked over Step 7 and not sure how I missed the +/- vs Avg Results column. Very happy you brought it to my attention. I also now see the overview of "most important filters for you" on Step 1. What is the sort order there? I'm hoping it's by how much you're trailing the avg winning player's bb/100?

Do you want to tease any of the additional features?

Oh..one additional thing...I can't resize the LB window! That drives me crazy. I have a huge monitor and yet I am doing all this analysis in what appears to be 600x1000 of pixel space. That's my last comment. I promise I'm done now!

It does account for this, but I'm unsure what step your referring to specifically. I talked with the developer about this and it's doing this in step 6, but not in step 4 for example. In step 7 it's using the comparison part so you'll know better if it's a leak or not. Same thing applies with folding to a 3-bet. It's not specifically accounted for simply because the analysis would take too long. We had talked about putting what the average opening size is in for a person, and then just multiply that by total hands played, just as a simple way of knowing profitability.

But thanks, it's something we're aware of, and when the HM database speed allows for it, we're going to be making that much more clear. I think for now, we offer guidelines of ~ where your losses should or shouldn't be.

Even if a hand like your T7s example if showing up if you're playing it, but advice is going to be presented unless there's some repetitive pattern to your play. If you just have one or two hands like that which are significant losses, LB isn't going to point this out. If you read the sidebar advice, it's pretty intuitive. That step isn't meant to over analyze every hand you're losing with. But it does provide a visual over view in case there's something that stands out for you. If not though, you can allow LB to let you know what you should really focus on, or not focus on. Hope that makes more sense.

What are you referring to specifically. We're trying to convey to people that LB is a little more than just some stats and grades, and that it's built by winning players for other online players. It actually has a ton of value to any online player. I mean, for mid stakes players, the ranges and things aren't going to be as interesting, but there's a lot more that you can quickly learn about your game if you do a complete analysis using Leak Buster. But if you have any ideas, please let me know.

It sorts by your most significant losses in that step 1. But we are going to add a feature that you can choose a sort order based on the winning player analysis for that area.

We are confined to that space for now until there's a HM update. We want to maximize the space, and it's something we're hoping to do soon.

As far as new features, I can't say too much, but there will be a new step 8, and an in-depth analysis function that will be very sweet.

Again, appreciate the feedback and comments. We listen to them all. If you have any other suggestions please let me know.

New Features in 3.0.05f - Release date: 2/04/11

• LB will auto adjust the length of data to your screen
• Columns can be re-sized in step 2
• Leak History sorting fixed

NOTE: Very soon we'll have a beta update program we will post in the forums here. If you want to keep up with the latest beta, you can come to the HM forum and download the latest update. When beta is completed, a new version will be uploaded in the automatic update process in LB, so you won't be prompted as often to update LB.

I've been asked to make a public service announcement.

I understand there are several people excited about the video contest, but please check the forums rules and regulations that you're posting on if you're going to post asking people to vote for you, etc... This goes on any forum, and some forums are a lot more lenient than others. 2+2 is a bit more in the non-lenient category compared to other poker forums. If you're unsure, please ask a mod before posting. It's fine to post in this thread, but if you're starting other threads, just please ask first.

Thank you, and good luck to everyone!

This has been a public service announcement. Beeeeepppp

Is there any option to add supporting of PT database in the future or this issue is closed?

Sorry, missed this.... no, we will not be adding PT support.

I've heard of a "get it free" option for Leak Buster. If this exists, would you mind pointing me towards it?

Right on their website. http://www.holdemmanager.com/store/poker-sites.php

Thanks. Didn't realize it was for all HEM apps as well.

Just bought LeakBuster today - it has already pointed out a problem I didn't know I had. One question and one suggestion:

question first: when looking at the filter for Top Pair (Step 7), it shows my bb/100 at 243.33, but then says my +/- vs. Avg Results is -340.09.(!) Is this saying that the "average" player is winning 583 BB/100 hands with top pair? That seems like a really really high number to me.

now the suggestion: when selecting the time range for an analysis, please add the "last x hands" to the available selectors, similar to the way HEM does this now, so that I can always run my last 30000 hands or so.

thanks again.

Correct, those are the actual average winning results with each filter. Mine is somewhere over 700bbs personally (can't remember off hand). Your sample size will effect this, but this is one of the filters that should have a good amount of hands.

I'd suggest going into step 1, scrolling down to the filters area and see what it says as some of the areas you should be working on first.

Yeah there's just about every filter except X hands. I know it was something that was talked about being added before, and I'll pass it along. Thanks.

Just a random suggestion but could you maybe add a filter or two in Step 7 dealing with 3bet pots postflop (cbetting/facing a cbet/etc.)?

I'd also love if Step 6 wouldn't just be a 3bet screen, but could include VPIP. For those who are used to HEM, it makes WAY more sense to me to have the interface in this step be modeled after the "Preflop Cards" screen in the regular part of HEM. I understand that steps 3 & 4 accomplished something similar to Preflop Cards, but I'd love it to be a little more in sync w/ the rest of the program.

(FYI, I always found step 4 a the least useful, even baffling at first, imo.)