3k Post: How To Learn Good Them SNGs (tl;dr)
11-05-2009
, 08:16 PM
I'm reformatting this post from my blog. It may be easier to read over there than here (see link in profile ldo)
As I've said many times before, getting good at SNGs is _hard_. Not only do we have to play competent FR or 6-max poker, but we also need to learn many additional sub-games: mid-level, bubble, and heads up. The games are tough now, and you need to be knowledgeable at every stage of the SNG to squeeze out your profit.
When I played SNGs, I went through many iterations of game review using HEM and SNG Wizard to try to hone by general game and my ICM-based decisions. Here I'm going to outline my final method I used, which allowed me to review a very wide range of hands I played in a methodolical, systematic way. I was playing 50-80 SNGs a day, so it was impossible to review all hands. But at the same time, it is important - as you add on more tables and play more hours a day - to be confident you are not leaving unseen holes in your game.
The main tools I used were HEM and SNG Wizard. I'm sure PT3 will do most - if not all - of what I lay out below (except where my own custom tools come in to play).
Along with these 2 main programs, I developed various AHK scripts. Any mention of a script can be found in the "tools section":http://www.sickread.com/blog/tools/ of this blog. I'm not here to pimp them out - I wrote them specifically for my own use as the needs arose, and have since made them available publicly. All the scripts are free (with the exception of a 'pro' version of WizHUD, but that's an optional extra).
I am splitting this in to two distinct sections:
* The daily review
* Systematically learning ICM from real hands
This might seem like a lot of work. But in practice it's much simpler than it sounds. Normally I would study an hour or two before playing each morning, then play for 5-6 hours. At the weekend, I would do the weekly 'systematic' review for a few hours. I think at least 20% studying time is the least you should be doing if you want to continue moving forward. Put in place your own strict study session and, using some of the techniques I outline below, you should be getting through a review session much quicker, and making much better use of your time.
I don't play SNGs much any more. And I was not that hot at them. I think by the end I was playing the mid-stakes 6-max games on stars and beating them for some, but I was far from the best at the tables. A lot of my problems came from concentration issues playing multiple tables. I was never great at that. But what I was damn good at was my ICM. Gimme a situation and the stack sizes and reasonable calling ranges I could nail the correct range of hands to be shoving. I could give a pretty close approximation for nash for shover, SB and BB in most standard 2-4 situations, and this information was internalised through many hours of study. I'm sure many regulars on the forum can do the same, but for the new player, the task seems daunting. I've read many posts from players saying, "How exactly am i meant to remember this?". Hopefully this post will help with that.
The Daily Review.
Marking hands and tournaments.
To be prepared for daily review, I would make extensive use of the 'mark hand' facility in hold'em manager whilst playing. I don't just mark hands I feel I might have played incorrectly, but also hypothetical situations such as "well here this is an easy shove with A4o, but what if I make my hand J4o?". I have a notepad by the laptop to make these additional notes. Using a hotkey (which is one of many I have in my "HEM Shortcuts" AHK script), I can mark the currently playing hand in HEM.
I would also irregularly 'mark' entire tournaments. Again I had a small script to do this (which isn't publicly available at the moment - but you can just jot down the last 3 #s of the tournament ID in your notepad for now). I would do this for when there was a string of interesting hands, say a long fought-out bubble with interesting stack sizes, or a long set of BvB situations vs. a reg.
The Review Session
I would review in the morning, reviewing yesterday's hands. I preferred this so that it would get my brain working before I started playing my first set of SNGs. You may work better reviewing at the end of the day, whilst the hands are still fresh in your mind. But it's important to get in to your regime and stick with it.
I will mention something now that will come up a lot in this: I never use the SNG Wiz 'game view' (the list of hands in one tournament with tick marks) at any point of the review. I think many players do their reviews by just firing up some HHs in wiz, quickly scanning for any red X marks, checking a few hands, and calling it a day. I found this a very ineffective method, and potentially destructive way to learn. I'll go in to this a bit later in part 2.
Reviewing marked hands.
So in HEM, set the date filter to 'yesterday', go to the tourney 'hands' tab, clear any filters, and check the 'only show marked hands'. Click twice on the 'date' column to sort in chronological order (oldest first). Then right click and 'Review all hands in replayer'.
There's no science to the next part - step through the hands, have your notebook ready, and check out those problem spots. This post isn't about actually 'how' to improve, rather the methods I used to do a review, so I won't go in to what exactly you should do at this stage - but a few pointers: if you want to get outside help on a hand, do it now when the hand is up in the replayer and you have stats. Post it on 2p2 (don't use the '2p2' format that HEM provides, it's barely readable. Copy the 'txt' format, and paste that in to the 2p2 converter). Share it with a friend (I paste that same text in to weaktight.com). Paste it in an IM to a friend. Make use of HEM's mini-hand history, that gives you your equity vs known hole cards on each street.
If you have an ICM decision, send this hand to SNG Wizard. Locate the hand back in the HEM hand list. If you have WizHUD, you can middle click on a hand and it will be send to wiz for you. Otherwise, right click, copy hand to clipboard, then over on SNG Wizard, use the 'paste' button on the toolbar.
Once I've been through marked hands, I do the same process but for marked tournaments. If I have no marked tournaments, I may just pick a couple at random, or maybe the last ones I played. Find the hands in Tourney > Results > Data View, find the tourney based on the # you wrote down, then sort the hands in chronological order, right click and replay all hands.
Using all my shortcuts in HEM Shortcuts for the replayer, I can review a tourney _very_ fast. I could go through one and review/post important hands, check ICM spots in wiz, in 5 minutes.
Daily hand filters.
The second part of my daily ritual was running specific filters across the day's hands. This would be a 4 or 5 filters saved in HEM to check common trouble spots. Often these would be checks to ensure I'm not making any misclicks or board misreads due to multitabling. For example, one time during a random tournament review, I missed an easy cbet on the flop. This was probably due to my focus on another table, and missing the action. So I set up a "missed flop cbet" filter, and ran it daily. Of course, some times I wanted to check back a flop, so I would have some 'false positives' in the filter, but having these filters, and knowing I would check them daily, gave me confidence to ignore these issues ingame because I knew they would surface in the review later.
Similarly, at various times I made some changes to my default preflop decisions (for example, to open fold small PPs UTG in 6max). Because a lot of my preflop play is done automatically at a sub-conscious level, I was concerned I might not apply the changes. So I created a filter "vpipped 22-55 UTG/HJ with >20BB eff stack 4 players+". The HEM filter system is pretty powerful once you get the hang of it, so many scenarios can be modelled and saved as filters to be run daily.
Systematically learning ICM from real hands
The problem with using SNG Wizard for review.
As I touched on earlier, I never use the wiz 'game view' when reviewing games or when trying to learn ICM. I think there are two inherent problems with the 'scan the game view and look at the checkmarks, or step through the hands' process.
One is that simply the ticks and crosses are not accurate enough to use as a guide for whether a hand should be reviewed or not. There are too many X's that are obviously fine and, much worse, many ticks for shoves that were clearly wrong. There are some specific situations (such as when there is a shorty all in) where ranges are way off that the ticks are meaningless. This is no criticism of SNG Wizard, it's just the nature of the game.
But more importantly, this method is very ineffective at internalising the information in order to learn ICM and make the right play at the table. Clicking through hand after hand, looking at constantly changing dynamics and stack sizes and # of players, does no good to try and learn one certain spot. First, you see a hand where you shoved too loose in the cut-off. It should have been only top 40%, wiz says. Then you see a hand where you folded when you should have called out of the big blind. Any ace, broadway or pair, wiz says. How are you supposed to remember this for next time?
It's not really relevant whether you played that hand correctly or not. Patting ourselves on the back for getting a sea of green ticks, or chastising ourselves for getting lots of red X's does little to further our actual understanding of the game. What's important is that we can learn something from that hand to further our internal knowledge of the game and make a better, more informed decision next time it occurs.
A Systematic Approach with HEM filtering
Using the power of HEM filtering, and a little extra I built in to wizhud, we can turn our review process from looking at hands in a one-dimensional, linear fashion (stepping through hands in a tournament), to a dynamic and methodical process, using our real hands.
Each week, I would pick one ICM situation and create a specific filter through HEM. For example, I may feel I've been making some mistakes on the button with 8-12 big blinds when 4 and 5 handed.
Fairly complex filters can be created. You can look at 3bet situations with deeper stacks, setups with limpers, etc. Using the effective stack, you can also in some situations control the opponents stack size within certain bounds(this works only in blind vs blind situations). I then set the date filter (all hands that week or, if it's a very specific filter, I may look at all hands that month or even wider) and we get our list of similar hands.
Using the ctrl-middle click feature in wizhud, i can then send all these hand histories in one batch over to SNG Wizard. The result is that we now have a pretend tournament in the 'game view' of sng wizard, just showing all hands in a specific context.
The benefits of this should hopefully be obvious. We can now easily see how well we are playing in this situation, whether there are common errors (maybe over-valuing A-x type hands, generally shoving too wide, over-estimating our fold equity, etc). By seeing one specific situation, we are cutting down on the number of variables (our position, our stack size, blind level, etc) and limiting it to a few uncontrolled variables (out hand, opponents stack sizes, etc).
The result is really like a SNG Wiz quiz, except you have more control over the hands, you have real situations vs. real opponents, and you can review your play at the same time as internalizing the scenario.
The missing context.
One problem with this method is that it lacks the context of the situation. When going through hands linearly, you see how the SNG progresses, build up reads on the players, etc.
This was one of the reasons I created wizhud - it allows you to get some sort of read when reviewing the hands with various stats. In the pro version, you can also add blind steal stats and HEM note text and icon.
But when doing this contextual approach, the main goal isn't to see how well you've played vs. that specific opponent at that time; we use this approach to have a body of example hands in similar contexts to see how changing variables effect correct push/fold/call ranges. One such variable is the read on the opponent. It's not that useful to conclude, "Well this was a bad shove because this guy calls super loose.". Much more useful is to note how exactly our profitable shove range changes based on that player type, so next time we can make the right decision for all player types.
Cliff Notes At The Bottom
* Have a regular daily process for reviewing your hands. Mark hands and tournaments as you play, and review these first, using the HEM replayer to step through the hands, only sending a hand off to SNGW when needed. Next, apply specific daily filters to review certain trouble areas in your game.
* When trying to learn ICM, use filters in HEM to create specific scenarios, then send this batch over to SNGW. With hands in groups, it's easier to try to internalise specific ICM situations. You can do this exercise as part of a regular review, applying a set of 2-4 filters a week to review and learn, or irregularly when you have the time to study.
What's coming up.
There are two main areas that I haven't touched on yet, that I will in future blog posts. One is how to use the various features of SNG Wizard to interrogate the scenarios and aid in learning correct push-fold strategy, including using the graphs, playing with ranges, and the the nash equililbirum.
Secondly, I want to show a method I devised for constructing real ranges based on emperical data (your own hands) in HEM.
As I've said many times before, getting good at SNGs is _hard_. Not only do we have to play competent FR or 6-max poker, but we also need to learn many additional sub-games: mid-level, bubble, and heads up. The games are tough now, and you need to be knowledgeable at every stage of the SNG to squeeze out your profit.
When I played SNGs, I went through many iterations of game review using HEM and SNG Wizard to try to hone by general game and my ICM-based decisions. Here I'm going to outline my final method I used, which allowed me to review a very wide range of hands I played in a methodolical, systematic way. I was playing 50-80 SNGs a day, so it was impossible to review all hands. But at the same time, it is important - as you add on more tables and play more hours a day - to be confident you are not leaving unseen holes in your game.
The main tools I used were HEM and SNG Wizard. I'm sure PT3 will do most - if not all - of what I lay out below (except where my own custom tools come in to play).
Along with these 2 main programs, I developed various AHK scripts. Any mention of a script can be found in the "tools section":http://www.sickread.com/blog/tools/ of this blog. I'm not here to pimp them out - I wrote them specifically for my own use as the needs arose, and have since made them available publicly. All the scripts are free (with the exception of a 'pro' version of WizHUD, but that's an optional extra).
I am splitting this in to two distinct sections:
* The daily review
* Systematically learning ICM from real hands
This might seem like a lot of work. But in practice it's much simpler than it sounds. Normally I would study an hour or two before playing each morning, then play for 5-6 hours. At the weekend, I would do the weekly 'systematic' review for a few hours. I think at least 20% studying time is the least you should be doing if you want to continue moving forward. Put in place your own strict study session and, using some of the techniques I outline below, you should be getting through a review session much quicker, and making much better use of your time.
I don't play SNGs much any more. And I was not that hot at them. I think by the end I was playing the mid-stakes 6-max games on stars and beating them for some, but I was far from the best at the tables. A lot of my problems came from concentration issues playing multiple tables. I was never great at that. But what I was damn good at was my ICM. Gimme a situation and the stack sizes and reasonable calling ranges I could nail the correct range of hands to be shoving. I could give a pretty close approximation for nash for shover, SB and BB in most standard 2-4 situations, and this information was internalised through many hours of study. I'm sure many regulars on the forum can do the same, but for the new player, the task seems daunting. I've read many posts from players saying, "How exactly am i meant to remember this?". Hopefully this post will help with that.
The Daily Review.
Marking hands and tournaments.
To be prepared for daily review, I would make extensive use of the 'mark hand' facility in hold'em manager whilst playing. I don't just mark hands I feel I might have played incorrectly, but also hypothetical situations such as "well here this is an easy shove with A4o, but what if I make my hand J4o?". I have a notepad by the laptop to make these additional notes. Using a hotkey (which is one of many I have in my "HEM Shortcuts" AHK script), I can mark the currently playing hand in HEM.
I would also irregularly 'mark' entire tournaments. Again I had a small script to do this (which isn't publicly available at the moment - but you can just jot down the last 3 #s of the tournament ID in your notepad for now). I would do this for when there was a string of interesting hands, say a long fought-out bubble with interesting stack sizes, or a long set of BvB situations vs. a reg.
The Review Session
I would review in the morning, reviewing yesterday's hands. I preferred this so that it would get my brain working before I started playing my first set of SNGs. You may work better reviewing at the end of the day, whilst the hands are still fresh in your mind. But it's important to get in to your regime and stick with it.
I will mention something now that will come up a lot in this: I never use the SNG Wiz 'game view' (the list of hands in one tournament with tick marks) at any point of the review. I think many players do their reviews by just firing up some HHs in wiz, quickly scanning for any red X marks, checking a few hands, and calling it a day. I found this a very ineffective method, and potentially destructive way to learn. I'll go in to this a bit later in part 2.
Reviewing marked hands.
So in HEM, set the date filter to 'yesterday', go to the tourney 'hands' tab, clear any filters, and check the 'only show marked hands'. Click twice on the 'date' column to sort in chronological order (oldest first). Then right click and 'Review all hands in replayer'.
There's no science to the next part - step through the hands, have your notebook ready, and check out those problem spots. This post isn't about actually 'how' to improve, rather the methods I used to do a review, so I won't go in to what exactly you should do at this stage - but a few pointers: if you want to get outside help on a hand, do it now when the hand is up in the replayer and you have stats. Post it on 2p2 (don't use the '2p2' format that HEM provides, it's barely readable. Copy the 'txt' format, and paste that in to the 2p2 converter). Share it with a friend (I paste that same text in to weaktight.com). Paste it in an IM to a friend. Make use of HEM's mini-hand history, that gives you your equity vs known hole cards on each street.
If you have an ICM decision, send this hand to SNG Wizard. Locate the hand back in the HEM hand list. If you have WizHUD, you can middle click on a hand and it will be send to wiz for you. Otherwise, right click, copy hand to clipboard, then over on SNG Wizard, use the 'paste' button on the toolbar.
Once I've been through marked hands, I do the same process but for marked tournaments. If I have no marked tournaments, I may just pick a couple at random, or maybe the last ones I played. Find the hands in Tourney > Results > Data View, find the tourney based on the # you wrote down, then sort the hands in chronological order, right click and replay all hands.
Using all my shortcuts in HEM Shortcuts for the replayer, I can review a tourney _very_ fast. I could go through one and review/post important hands, check ICM spots in wiz, in 5 minutes.
Daily hand filters.
The second part of my daily ritual was running specific filters across the day's hands. This would be a 4 or 5 filters saved in HEM to check common trouble spots. Often these would be checks to ensure I'm not making any misclicks or board misreads due to multitabling. For example, one time during a random tournament review, I missed an easy cbet on the flop. This was probably due to my focus on another table, and missing the action. So I set up a "missed flop cbet" filter, and ran it daily. Of course, some times I wanted to check back a flop, so I would have some 'false positives' in the filter, but having these filters, and knowing I would check them daily, gave me confidence to ignore these issues ingame because I knew they would surface in the review later.
Similarly, at various times I made some changes to my default preflop decisions (for example, to open fold small PPs UTG in 6max). Because a lot of my preflop play is done automatically at a sub-conscious level, I was concerned I might not apply the changes. So I created a filter "vpipped 22-55 UTG/HJ with >20BB eff stack 4 players+". The HEM filter system is pretty powerful once you get the hang of it, so many scenarios can be modelled and saved as filters to be run daily.
Systematically learning ICM from real hands
The problem with using SNG Wizard for review.
As I touched on earlier, I never use the wiz 'game view' when reviewing games or when trying to learn ICM. I think there are two inherent problems with the 'scan the game view and look at the checkmarks, or step through the hands' process.
One is that simply the ticks and crosses are not accurate enough to use as a guide for whether a hand should be reviewed or not. There are too many X's that are obviously fine and, much worse, many ticks for shoves that were clearly wrong. There are some specific situations (such as when there is a shorty all in) where ranges are way off that the ticks are meaningless. This is no criticism of SNG Wizard, it's just the nature of the game.
But more importantly, this method is very ineffective at internalising the information in order to learn ICM and make the right play at the table. Clicking through hand after hand, looking at constantly changing dynamics and stack sizes and # of players, does no good to try and learn one certain spot. First, you see a hand where you shoved too loose in the cut-off. It should have been only top 40%, wiz says. Then you see a hand where you folded when you should have called out of the big blind. Any ace, broadway or pair, wiz says. How are you supposed to remember this for next time?
It's not really relevant whether you played that hand correctly or not. Patting ourselves on the back for getting a sea of green ticks, or chastising ourselves for getting lots of red X's does little to further our actual understanding of the game. What's important is that we can learn something from that hand to further our internal knowledge of the game and make a better, more informed decision next time it occurs.
A Systematic Approach with HEM filtering
Using the power of HEM filtering, and a little extra I built in to wizhud, we can turn our review process from looking at hands in a one-dimensional, linear fashion (stepping through hands in a tournament), to a dynamic and methodical process, using our real hands.
Each week, I would pick one ICM situation and create a specific filter through HEM. For example, I may feel I've been making some mistakes on the button with 8-12 big blinds when 4 and 5 handed.
Fairly complex filters can be created. You can look at 3bet situations with deeper stacks, setups with limpers, etc. Using the effective stack, you can also in some situations control the opponents stack size within certain bounds(this works only in blind vs blind situations). I then set the date filter (all hands that week or, if it's a very specific filter, I may look at all hands that month or even wider) and we get our list of similar hands.
Using the ctrl-middle click feature in wizhud, i can then send all these hand histories in one batch over to SNG Wizard. The result is that we now have a pretend tournament in the 'game view' of sng wizard, just showing all hands in a specific context.
The benefits of this should hopefully be obvious. We can now easily see how well we are playing in this situation, whether there are common errors (maybe over-valuing A-x type hands, generally shoving too wide, over-estimating our fold equity, etc). By seeing one specific situation, we are cutting down on the number of variables (our position, our stack size, blind level, etc) and limiting it to a few uncontrolled variables (out hand, opponents stack sizes, etc).
The result is really like a SNG Wiz quiz, except you have more control over the hands, you have real situations vs. real opponents, and you can review your play at the same time as internalizing the scenario.
The missing context.
One problem with this method is that it lacks the context of the situation. When going through hands linearly, you see how the SNG progresses, build up reads on the players, etc.
This was one of the reasons I created wizhud - it allows you to get some sort of read when reviewing the hands with various stats. In the pro version, you can also add blind steal stats and HEM note text and icon.
But when doing this contextual approach, the main goal isn't to see how well you've played vs. that specific opponent at that time; we use this approach to have a body of example hands in similar contexts to see how changing variables effect correct push/fold/call ranges. One such variable is the read on the opponent. It's not that useful to conclude, "Well this was a bad shove because this guy calls super loose.". Much more useful is to note how exactly our profitable shove range changes based on that player type, so next time we can make the right decision for all player types.
Cliff Notes At The Bottom
* Have a regular daily process for reviewing your hands. Mark hands and tournaments as you play, and review these first, using the HEM replayer to step through the hands, only sending a hand off to SNGW when needed. Next, apply specific daily filters to review certain trouble areas in your game.
* When trying to learn ICM, use filters in HEM to create specific scenarios, then send this batch over to SNGW. With hands in groups, it's easier to try to internalise specific ICM situations. You can do this exercise as part of a regular review, applying a set of 2-4 filters a week to review and learn, or irregularly when you have the time to study.
What's coming up.
There are two main areas that I haven't touched on yet, that I will in future blog posts. One is how to use the various features of SNG Wizard to interrogate the scenarios and aid in learning correct push-fold strategy, including using the graphs, playing with ranges, and the the nash equililbirum.
Secondly, I want to show a method I devised for constructing real ranges based on emperical data (your own hands) in HEM.
11-05-2009
, 08:17 PM
11-05-2009
, 08:18 PM
11-05-2009
, 08:37 PM
11-05-2009
, 08:46 PM
11-05-2009
, 08:52 PM
11-05-2009
, 09:05 PM
11-05-2009
, 10:34 PM
11-05-2009
, 11:07 PM
11-05-2009
, 11:53 PM
11-06-2009
, 02:10 AM
