Monday 19:00 GMT Sweat Session
Thursday 17:00 GMT HH review
Saturday 17:00 GMT Topic of the week discussion
Sunday 19:00 Homework of the week
Yesterdays transcript:
<@Jon> Ok, I found one decent resource http://poker.cs.ualberta.ca/publications/ you should try to save those because these publications are pruned once a year or so. We can brainstorm some ideas from there.
<@Jon> Back to topic, we talked about an idea to purchase a large quantity of handhistories online and analyze the game based on those and hopefully create a winning strategy that would be the most optimal versus an average player.
* m4Ntiz (~mkeiriss@212.93.114.156) has joined #pokerstudygroup
<m4Ntiz> hi guys
<Ted_Forester> hello
<none888> привет
<m4Ntiz> privet none888
<@Jon> Sup Maris, I'm telling the group about our new project.
<m4Ntiz> aight cool
<m4Ntiz> https://www.boomplayer.com/8721955_DD718C33C9 hand i played for review later
<m4Ntiz> doing good at super tuesday
<m4Ntiz> would be nice to win me some 100k
<@Jon> Keep it up!
<@Jon> So, as long as there's one player at the table who's not hidden from the search, we are able to get that handhistory. My guess for market coverage is approx. 75% which is more than enough to get all the data versus an average player.
<@Jon> 500k hh per day, thats more than you can handle in the real time. The work has to be done off-tables, Holdem Manager or PokerTracker won't work with that many hand histories.
<RK> we could still have starting hand charts and best frequencies if we ran through 10 million hands or so?
<@Jon> Absolutely, that's the least we could do, in fact, we've already done some work on those.
<@Jon> But a home computer won't work to make these calculations. 10 mil took forever. Any of yall still going to univeristy? Got access to a supercomputer?
<RK> Lol uni, we're all degens here?
<m4Ntiz> I still do, but I'm planning to postpone the studies for a little after my internship
<m4Ntiz> will start an intership at a bank in 2 weeks or so
<m4Ntiz> unless I take down super tuesday lol
<RubItOnMaTitties> Me too, still at the uni
<m4Ntiz> http://www.lu.lv/eng/ not the western type of university but we have one supercomputer
<m4Ntiz> I dont know anything about it though
<@Jon> Great! I think Christian will know how to use it, if you can get access to it.
<@Jon> So, again, the goal is not to win against an individual player, but to create a strategy that crushes your average opponent.
<@Jon> Individual players are not that important at the lower stakes, because there are too many of new players, non-regular players and fish. We will start from the fundamentals and then go more into details.
<RK> could we use this data for learning more about how to perceive stats on hud after 10/50/200 hands or so?
* krempi_vladimir (~krempi@85.10.53.22) has joined #pokerstudygroup
<krempi_vladimir> Good day, sorry if I'm late.
<@Jon> Hi!
<@Jon> No problem, we will upload the transcript to the group blog, so that everyone is on the same page.
<@Jon> Everything is possible, you have all the market data, just need to know how to extract it: analyze top winners at every stake, reverse engineer their games, you can precisely see how poker evolves over time and you won’t waste your time playing millions of hands, reading all the poker books/articles and watching poker videos, preflop opening ranges, understanding how and when do other players playback at you.
* m4Ntiz (~mkeiriss@212.93.114.156) Quit (Ping timeout)
<RK> and if sites were to ban huds
* m4Ntiz (~mkeiriss@212.93.114.156) has joined #pokerstudygroup
<RK> that wouldn't hurt that much if you have all the data.
<m4Ntiz> motherfucking internet acting stupid when I'm playing 1k tourney...
<@Jon> Good point, RK.
<@Jon> Let's have a look at the files that I've sent you.
<@Jon> Remember only 10mil hands, everything is just an example.
<m4Ntiz> k
<RK> ok
<@Jon> "Action trees", you can simulate any preflop situation and you will know the exact percentages what actions your average opponent might do.
<@Jon> A and B columns always stay the same, they represent the sb/bb.
<@Jon> "Regulars vs fish" you can find what hours are best to play, when there is the most amount of fishes online and lowest amount of regulars.
<@Jon> "Stats distribution within intervals" - how many hands do you need on a certain stat to know that it is worth looking at. You have got to input the number of the stat in A1 field.
<RK> nice
<krempi_vladimir> (clap)
<@Jon> "Flops" is our system to analyze most the frequent flops. The first 5 lines show you how to read the file. Just ignore the first number under the var, look at the zeros and ones, the last number shows the percentage of the times. if you look at the line 43 010 00001 001 0010 01 12.54 – it is explained. This is the most frequent flop. So, instead of going through 100 different hands and thinking how you could have played, you could just perfect your play on most frequent situations. You learn the first 5 flops - you know the right play on almost half of the situations. Things like these can save you hundreds of hours.
<@Jon> "EV" - player pool average profit/loss in big blinds by holecards and position, you can understand how people perceive certain cards, for example people overplay A6 vs A7/A5, they give to much credit for 6.
<m4Ntiz> holy shit thats gold
<@Jon> You get the idea :)
<@Jon> Remmember people - the most efficient way is to analyze poker market as a whole, not individual players. And we're talking about data from the past here, you can build on it and make much better predictions than your average "Play LAG vs TAG" player advice.
<HoldemLady> Maris, can I still buy 25% of todays action?
<m4Ntiz> sure like i promised earlier
<HoldemLady> thanks
<m4Ntiz> k
<m4Ntiz> u can ship em for me
<m4Ntiz> youve got my paypal?
<HoldemLady> think so
<m4Ntiz> mariskeiriss@gmail.com
<HoldemLady> 262.50?
<m4Ntiz> y
<HoldemLady> give me a couple of minutes
<m4Ntiz> k
<RK> Jon, what about "Data Parsing" and "PT4 st"?
<@Jon> These are for Christian, ignore those for now.
<RK> https://forumserver.twoplustwo.com/164/cash-game-poker-coach-listings/pokermetrics-advanced-database-analysis-shnl-827925/ this looks like a simplified version of your vision Jon
<@Jon> 2 mil hand db.
<@Jon> That's nothing, too small samplesize.
<@Jon> You guys should still read everything on that thread, maybe we could implement something to our project.
<RK> ok
<krempi_vladimir> sure thing!
<@Jon> People still don’t think outside the box, and the vast majority never will, poker is such an easy way to print money if you look a little bit deeper!
<@Jon> Well, that's all that I wanted to say.
<@Jon> I'll upload this chat to our blog for those who couldn't come at this time and I will be waiting for some feedback, ideas when I'm back from Greece in one week!
