I recently purchased a HEM application called Leakbuster PLO. I wanted to create this thread to discuss the program in detail and try to take a critical approach towards some of its features and assumptions. In particular, I want a discussion on how well the program is suited for different stakes. Also, simply understanding the different stats is one goal.

Leakbuster is a program that analyses all the different aspects of your game in great detail to find out which ones are losing you money, and by how much. It gives you school grades from A-F and shows you how far off you are from optimal play and how much it affects your winrate. For example, my flop aggression at PLO200 6max is 41%, when the optimal range is 30-36%. This gives me a D+, affecting my winrate with a factor of 3.5. I also found out that I'm calling 3bets too light in position and not opening enough from the small blind.

All reviews of this program are welcome ITT, hopefully we get some fruitful discussion.

I tried the trial version and my first impression was that it's a program built on great generalisations and while it may help some people, it is not the most optimal way to improve your game.

I also think that if you play varying stack sizes (ie sometimes play normal tables, sometimes play deep tables, sometimes play shallow tables) and your hands are all blended together than the information it gives is useless.

I would love to be proved wrong though. Like I said, these are based on very first impressions.

The different stack-sizes are a good point. I only play regular/deep tables, but someone playing CAP/shallow should def try to filter. Interesting to see if the results vary.

are these grades based on the optimal stats of your own database or based on the stats the developers have gathered from multiple rooms?

Im asking that because I play on ongame which is 5 handed.. I think it would affect the stats significantly...

Good question. My guess is that the optimal stats suggested by the program are the result of a consensus by a bunch of pros. AFAIK, LearnedFromTV is one of the creators.

No idea how 5max is taken into account.

Definately want to hear more about this. Would love to see LFTV's input.

I bought this in preperation for this thread and have played around with it a bit. I ran it specifically on my PLO25 hands (about a 25k hand sample since Jan 1), which were all at regular 6M games, IIRC.

I graded out to an A- despite losing at a ~11bb/100 rate (I was a good number of BI under AIEV FWIW). And my game changed a good bit during the middle of it. I'd be curious as to what an adequate sample size is (for this sample, it said the accuracy of my results was 36%). I'm also curious as to the effects of any game changes within said sample.

That said, I think some of the advice it offered is beneficial for my game and I'm going to work on incorporating it.

I'll reserve my recommendation as to who it would best benefit until after this thread's gone a bit longer and I've had more of a chance to play around with it.

The way I understand it, they look at very large databases of hands (not your database) to determine the "optimal" ranges for various stats. I haven't used the PLO versions, but I do have the NLHE version. It can help when you are grossly out line in a specific area, but it doesn't give you specifics. For example, you might get an indication that you don't 3bet enough, but it won't help much finding additional spots where you should 3bet, or how to play the flop after you get called.

I thought one of the really interesting aspects was the videos/content tailored to a some of the leaks it identifies. Unfortunately, one of the things it said for me (river call efficiency) didn't have any videos that I saw. I may not be doing it right, though.

Still, knowing what I need to focus on in session reviews is handy and (lolsamplesize) I have brought my WTSD down a bit from ~28/29 to closer to ~25. <- Possibly a limit-dependent stat.

so it teaches you how to be average. woop de do

For me the program gave advice by listings the groups of hands that I spew with when calling 3bets. You can obv take the advice and optimal stats FWIW, but at least you'll find areas where you're losing money.

It teaches (or attempts to teach) you to be in the average of the top performers. I'll take that for a start.

I can try and help answer any questions you guys have. I'm one of the primary creators so I know the program well. Our intention with the program is to have a resource that can quickly let any level player know areas they need to address in their games, and provide solid advice on how to correct those areas. You can spend several hours yourself investigating these areas in your database. Even if you do spend say 30-40 hours, I don't think you're going to pinpoint all the areas LB will find in what will take it 10-15 mins of analysis (that's my honest opinion).

One thing about tackling PLO in this model is the variance is much wider than no-limit and even limit. So sample sizes are extremely important. In no-limit we recommend a minimum of 50k hands (which is still small, 100k or more is better). In PLO we're looking at more like 100k (200k or more is better).

As far as teaching you to be average that's obviously not the case at all. Leak Buster's goal is to make you aware of areas that are potential leaks and obvious leaks. I think it does a really good job of this overall.

A lot of attention is put on steps 2/3 (where the scoring and the grading are). These stat areas are general over views that if you're really out of alignment somewhere, it will let you know. But the grading area is one small part of what you should take away from LB as a PLO player. A lot of the value if you're a more advanced player is in the filtering, troubling hands, and the 3-bet pot grids.

Leak Buster isn't a perfect tool, and it's not a direct replacement for coaching, but I think if you do use it you'll find really good value in it. So if anyone has any specific questions, I'll be happy to answer. We will be adding even more training content (this is just initial release), and I think you guys will like some of the names we are close to bringing in.

Most importantly, I don't want anyone owning the software that doesn't believe after they've used it that it's helped their game, or was a very +EV decision. You can just contact me if this was ever the case and I'll insta-ship your money back. We only want satisfied customers and we do stand behind that as a company.

Awesome! I'm sure this wil help us all in learning to understand the intricacies of the program.

I think a lot of people would appreciate if you explained how stack-size is taken into account, for starters.

Also, how are you taking into account adjusting to different stakes?

teh avg of the top players will automatically change as they chagne who they compare you to.

a better qeustion to ask is if the parameters for judging top players changes based on stakes (if their sampled naems become more numerous, etc)

100k is a lot of hands and would take a long time for me (4-tabling and I play a relatively good number of hours for a non-pro). IMO, that sort of makes the line for beneficial at 20-30k hands/month with some lower utility for those that play less. I think it's just too likely that someone's game changes over a 6-10 month period that a high level of accuracy might be too difficult to achieve. Or as GGARJ says, people move up and the higher limit necessitates a change in our approach and the lower limit play ends up skewing the data.

Sure. So the only part of the program where stack size is an issue is in step 2/3. When we compile stats is based on table size using millions of hands in normal PLO games. So stack sizes over the course of 100k or more hands for a particular person is going to have an accumulation of many different stacks for a particular play (how often they 3-bet for example). If we take into account just specific stack sizes for every statistical category the variance based on the amount of PLO data out there is too great. So just like in no-limit though, this issue (being deeper or shallower) evens out assuming you're not JUST playing deep or shallow tables. Hope that makes sense, but in a nutshell for large samples it's really a non issue because it evens out.

As far as stakes, just like we do for limit and the other versions there are stats accumulated at micro and mid stakes. Right now there is one average range for statistics that is spread across these stakes, with the scoring being adjusted based on what winning players maintain at those respective stakes. In future versions will separate out ranges as well like we do for other versions of LB, but overall with the point of steps 2/3 it's not the most important point when you're looking at these stats. You really want to know, am I really out of alignment in particular areas, and if so, how should I look to address it (which is what LB will offer in terms of advice).

I hope that helps, but if you have other questions, just let me know.

Keep in mind this is recommended for accuracy in steps 2/3. Other steps you can do much smaller sample, even 5k. 50k you'll get good info back for steps 2/3, but the accuracy (and you'll see this in the accuracy meter) isn't going to be excellent, so it's something you need to keep in mind. If something is border line close on being a leak, don't start completely changing your game if you're looking at 40k hands. But use it as a guide to take a closer look at that area. If it's way out of alignment, and showing a score of a 1 or 2, then it will likely be a leak even in larger samples, so you'll want to start making adjustments.

Yes, they do change when moving up stakes, but surprisingly not by as much as you'd think. We've found this true in no-limit and limit as well. One example is in no-limit as you move up, the games become more aggressive, and the opposite is true in limit, the games actually become more passive. If you're a winning player though, most of the statistical areas don't vary that much whether you're playing micro or mid stakes. There are certain areas that do, but not by a lot. Others are almost exactly consistent no matter the stake.

We take this into account when creating the ranges, and that's why the algorithms used to compile the stats and scoring allows you to see how the winrate change, even if you go higher or lower from the Leak Buster recommended ranges. There are several statistical areas that will actually score higher than a LB recommended range, but it's not recommended by LB because it's much less likely that winrate can be maintained. Ultimately it's up to you to decide if you're comfortable with they style you're looking to play, but you'll get to see how good players are playing, and where the most profitable ranges do occur.

Again though, the stat areas in LB are great. I honestly think as you're moving up you'll get much more value from the other steps. Tom did a great job with the analysis and advice, and Tom is extremely thorough with his research.

FWIW, I think the leakbuster HUD is pretty damn sexy. Just from seeing my own stats on that, it seems I need to work on the non-blue ones (blue is TAG).

Edit: PFR, 3-bet (which I knew) and fold to 3-bet (which I'm going to look at more in-depth).

2nd Edit: And my c-bet flop is a touch high.

i havent played around with it, but do you give the error bars/confidence intervals of stats/projections? you say its less likely to to maintian that winrate in some cases, i think ppl could visualize this easily with some sort of way to appreciate the variance associated with a change to x stat

I have bought PLO Leakbuster about 1 month ago, just directly after, when it has been released.

I think the application is really well built, "LearnedfromTv" has designed and built it. There are some cool videos also at step7 (filters), and also about 3betting and more.

But I do not understand one part, and thats step 4 (pre flop). Somebody could explain that to me?ty

I believe, but don't quote me, that the development team had initially considered another number (when LB no-limit 3.0) was being designed, but when they went to the line bar and had the winrates in there they thought it was way too much info on one screen. I think it's a reasonable idea personally, I'm just not sure exactly how to present the data, but if you have an idea I'll pass it on to the development team.

Right now, LB is doing this, but you aren't seeing a statistical "confidence" meter for areas outside of the recommended ranges. It's a good idea.

Glad you're enjoying it.

Step 4 is basically a visual overview of losing hand groupings pre-flop. This way you can see if there's specific ones you're really struggling with. In other versions of LB there's advice provided in those sections, but it's not quite that clear cut in PLO so we didn't add anything there, but we're still thinking of re-visiting that area.

Not to derail the thread too much but these guys are serious about customer service and being helpful.

After purchasing LB, I had a problem reading the advice. Half of the text was being cut off in the letters, so I wrote tech support.

There was a quick response with thoughts on what it could be and how to correct it. A suggestion had things working beautifully. Apparently, it was a font sizing problem due to my windows desktop configuration.

The developer was made aware of the problem and had a fix for it which was included in a patch a couple of days later.

I'm very impressed that not only do they offer a free trial version but this guy has offered to insta-ship your money back for the full version if you don't feel it's helped you.

+EV