GTO vs. Exploitative Poker: Whose side are you on?
In this wonderful Tweet from the unique character that is Scotty Nguyen, he makes what is actually a very important point - poker is not all about solvers. And yet of course every high stakes player worth his salt is using one for study - so they must do some good, right?
In this write-up we aim to settle the issue once and for all - do we have to take sides in this, or is a third answer possible - a hybrid approach? To answer this at all we must first understand a little about what these two types of poker are all about.
What is GTO Poker?
GTO stands for Game Theory Optimal, and it is essentially a way of thinking about modelling decisions made in a poker game. GTO is a model of a poker hand which assumes that every player’s range is transparent to every other player, meaning they all know exactly what each other’s strategy is, with their whole possible range of hands. The simulation then runs the hand, and allows the players to then adjust their strategies perfectly according to what they know of each others’ ranges. It then runs the hand again, and so on, in many, many iterations.
Such solvers can be run for thousands of iterations, but in many scenarios (preflop with many players, postflop typically with just two players modelled in a “heads-up” scenario) they will reach an equilibrium long before this point. After reaching this equilibrium, or balance point, no player will have any incentive left to deviate from the range of hands they are using, and their strategies will then cease to change.
In truth multiple such equilibria exist, since you can always alter bet sizes in NL. So there are always some outstanding assumptions which frame the solver setup.
Even artificial intelligence is only as good as the algorithm, which was written by someone. This applies equally well to poker solvers - they are only as good as their users, and designers.
The Flip Side - Exploiting in the Moment
Now, another problem with these static equilibria is that, as Scotty points out, they are, in a sense, dead. They have nothing of the cut and thrust of in-game decision-making.
It might help to know where the equilibrium lies for certain sizings mid-hand, as long as it doesn’t take us away from the immediacy of the game we are playing, and prevent us from using our instincts, from picking up reads, from playing on the right level for our opponent’s actual thinking.
The truth as always lies somewhere in-between, but Scotty has a really important point to make here, which he does of course in his unique and cool style. But the core point is this, that the solvers, as he says, “you know … put one hand out there, they have a million ways to calculate it!”
It is certainly true that if you get lost in the endless intricacies of solver space, it can have a downside of pulling you away from an awareness of how the game is played in real-time, and of understanding your opponents. The truth is that even at the highest stakes heads-up, there’s a ton of exploitative play going on.
Exploitative Play Defined
Simply put, exploiting your opponent means to play in a manner which is not balanced against them. This means that the imbalance opens us up, if noticed, to being re-exploited and therefore losing equity. For example, if we think the blinds are big old nits and we raise 100% from the button when we have the opportunity, that’s easily re-exploited if they notice and adjust. If they don’t adjust, it’s in-game optimal to keep doing it, but it is of course far more aggressive than GTO would go in that spot.
So it is clear that in many, many spots playing GTO poker which assumes our opponents are balanced will gain us far less than the maximum available, especially if the games are soft. So it must be looked at only as a point of reference (or multiple points of reference), against which we can understand our exploitative deviations all the better.
They Can’t Read, Baby
If you take one thing away from this article, and this clip, let it be this: Solvers can’t read the table, baby.
Even in online play there is just so much that can be gleaned from the game simply by tuning in, by showing up and being present, by observing. Timing tells, bet sizings, the flow of action in-game, these can all contribute so much to our instincts, if we’ve been honing them. Being willing to adjust, to play outside of GTO, is always important, but never more so than at the lower and mid stakes, where there are just so many exploitable mistakes being made.
Thanks Scotty for a wonderfully simple message that was so on point. We just had to go and write up the long version, but it was beautiful, baby.
By Lucky Luke
Image courtesy of WPT.com
Variance can be cruel, or uplifting, negative or positive. One thing's for sure - it's here to stay. Even the best players in the world are subject to it. As Richard Swift put it in the awesome song of the same name, "But I wish sometimes that Lady Luck, she would find some time to spend with me ”. We can relate, Rich! What is Variance? Variance is the term given to the uncertain short-term outcomes for which poker is well known, but it can of course apply to the swings in fortune you will see in any game where luck plays a role. The good thing about luck, of course, is that it is borrowed, not owned. In the long run everyone will roll an equal number of sixes, if we sit around rolling dice. The long run can be very long indeed (arguably millions of hands depending on the game format in question), but in the longest of long runs, luck will be distributed entirely evenly across all players, leaving only skill to determine who is ahead. This is the case with any game where the skill element is significant enough to play a major role, which there is no doubt that it is in poker, well established as a +EV skill game (positive equity value), meaning that it is beatable by dint of skill if you are sufficiently better than your opponents (and if you are not playing < 10bbs deep and/or with insanely high rake). Does Variance Matter? If you have infinite (or just very large) bankroll and the patience of a Buddha, variance doesn't really matter, as you can simply endure (or even enjoy!) the potentially seismic swings and treat it as the thrills of the rollercoaster. For all of us mere mortals, variance tends to matter, or at least feel that it matters, a good bit. It can be pretty rough to play through a big downswing, and it can also certainly cause you to second-guess your play, perhaps in spots where there is no need to do so. It's certainly true that most experienced downswings tend to be a combination of variance and human errors, which will of course compound any impact which variance has on your play. Telling one from the other can be pretty tricky. On the flip side, positive variance is also a thing, though fewer people tend to realise they are experiencing it. If you find yourself on the positive side of variance, especially near the start of your poker journey, you may simply feel that you are God's gift to the game, long before your poker skills are really honed. This may give you a feeling of complacency which can limit your ability to grow as a player, or cause quite a shock when you hit your first downswing. Measuring Variance The tools are now available to measure variance really quite accurately, at least in terms of understanding how much of it exists in different types of poker games. This very fine tool from PokerDope is the best one online for doing exactly that, and you can program the settings of this calculator to measure how likely you are to have certain outcomes at a certain skill level in a certain MTT or SNG. There's also a similar calculator for cash. To save you some trouble, we can provide some good rules of thumb for you right here in this article. If you're playing cash poker and you are confident that you have a skill edge on the field, you likely need at least 30-40 buy-ins to avoid much risk of ruin (going busto) especially if you are multi-tabling or playing zoom format. For a SNG grind that goes up to at least 75-100 buy-ins , which is also a good minimum for micro stakes MTTs . If you move up to the midstakes or higher for MTTs , you really should be sitting down to play with at least a few hundred buy-ins available. For all formats this should be measured against your average buy-in. Managing Variance If you choose to control the amount of variance which you encounter, instead of simply "embracing the ride” as remarked above, we'd be in agreement that this is generally wise, and a good way of having a lower stress existence and enjoying the game to the max. Should you go this route, there are a number of ways in which you can control the amount of variance in your poker grind right from the off, and a lot of it comes down to game selection. As noted above, choosing cash as your main grind is one way to go, as you certainly need the fewest buy-ins in your bankroll to weather the variance. This is simply due to the fact that chips are chips in cash poker, and you can sit and stand up as you please. MTT players on the other hand might run great at the start of a tourney and terrible at the end, or vice versa, and this is one reason there is a lot of variance in the MTT grind. If you do play cash, you can further reduce the impact of variance by buying in with 100bbs rather than 200bbs , or by sitting out when you reach 150 or 200 big blinds , and sitting a new table with 100 , especially if you have more skill edge with 100 big blind stacks. Bear in mind this will also cap your potential value gained in each pot. Another way to crush variance is simply to table select well and to sit spots where you have a large skill edge, as the impact of variance will be less the more profitable you are in a game. Long-handed tables also have less variance (and sometimes less value) than short-handed. If we're talking SNGs or MTTs , the smaller the field size, the less impact variance will tend to have. It is really not possible to measure your skill level according to your financial results in MTTs for example until you have a game sample of many thousands of games played, simply due to the size of the potential swings. So the gold standard for a low variance MTT grind is soft field, small field size, good structure. If you want to reduce it still further, knockouts and progressives reduce variance since you can win money before the bubble, though again your potential ROI (return on investment) will also be more capped in these games. Making Peace with Variance However much you opt to control the impact of variance on your play, it is important to realise that it's a baked-in part of the game of poker, and that this is a good thing. Frankly, the fun players wouldn't keep coming back to the game if they had no chance to win in the short-term. This is especially true of the MTT world, where beginners (with a bankroll!) can rub shoulders with the best in the world, something which you can say for very few sports or competitive endeavours on this Earth. Variance keeps the poker world turning, and without it we would literally just have a bunch of battle-hardened regulars out-grimming one another at the felt all day long and playing for razor thin edges. Every game would be the Hot $215 on Pokerstars , essentially. The good news is that it is possible for you to make peace with variance and not to sweat its impact on your game, past taking sensible steps to limit this as noted above. Beyond that it is important to find equanimity, and to accept that we are going to frequently face outcomes we don't want in a poker game, there is literally no way to avoid that. Don't go folding those Jacks just because you might have tough spots with them, it's all part of the ride. Some simple steps to achieving a zero-tilt mentality with regard to variance are to remind yourself, if you're an MTT player, that you will frequently finish 11th , or 9th , or 24th when you make that deep run. If you're a cash player, you'll have downswing days, don't sweat it and don't chase those losses. Suffering over variance has compounded many a poker player's losses as they go on tilt after a casual player hits a 2-outer against them in a 400bb pot and they get steamed and start to punt, throwing bad money after good (to reverse the old saying). If you'd like to read more PokerDeals guidance on mindset, check out our PokerDeals Mindset 101 post here . Breathe, remember it's better than working down the mines, and remember how lucky we are able to sit and play this beautiful game. In the long run you'll deal out exactly as many bad beats as you'll receive. You can't avoid that, but you can certainly avoid stressing out over it. Enjoy the game. If it's no longer fun, it's time to hit the beach, the mountain trails, or just sit down with a friend or lover and get some real life EV time. by Lucky Luke
Playing good postflop poker takes a good deal of work and study. In this series we're going to open it up, and explore a lot of key areas to help you on your way in developing this skill set. Of course many factors influence your ranges and your optimal lines postflop, including the game format. In a game such as MTTs , for example, you will often be shorter-stacked, and also with relatively large pots which include antes . There will also be ICM effects which factor into how tight or loose you should play, and these effects may be harder to measure postflop but they still have an influence. Being shorter will mean more one or two street postflop lines, where the opponents have simply got all before the river. In cash games, on the other hand, stacks are usually at least 50-100bbs effective, and there is usually no ante . Pots are smaller, ranges are somewhat tighter, but there's far more complexity to postflop play due to the deep stacks. There are more complex lines possible with deeper stacks, and hence arriving at the optimal line vs. tough opponents is harder. Couple this with the lower variance in cash grind and you have a tougher format, stake-for-stake, and one with a lot more to think about postflop. Still, the basic principles for postflop play apply equally to both of these formats, and we'll explore these in today's piece, an introductory voyage into the world of postflop poker. For the sake of consistency, we'll consider cash game spots in this article. Single Raised Pots These are often referred to by their acronym as SRPs , and refer to any pot which has been raised preflop once only, so pots in which no 3bet / 4bet etc. has occurred. We'll look today mainly at heads up pots to the flop, where there is an open raiser and a flat caller. SRPs Raiser has Position In these pots the raiser is in position (IP) and the flat caller is out of position (OOP) . The most common example for this scenario is an opening position between UTG and BTN , and the BB in defence. Let's look at a couple of spots where BB is defending against an IP open. Using some incredible software called GTOTrainer we can quickly calculate what the game theory optimal solutions will be for different postflop situations. Bear in mind that in a real poker game our opponents will very often be playing nowhere near optimally, in which case we would opt to adapt these ranges. Nonetheless using such software will give us a very good idea of what kinds of hands to include in our betting range in a given spot. Let's look at a situation where the Cutoff (CO) has opened the pot in a $1 / $2 cash game to $5 , the Button (BTN) and Small Blind (SB) have folded, and the Big Blind (BB) has called. Let's pull up an average enough flop that doesn't drastically favour either the BB or the CO , such as J ♦ 6 ♦ 2♠ . Game Flow Game Flow is an important concept in poker, and it refers to the fact that once a player has taken an aggressive line and his opponent has only called, he has what is called initiative in the hand. He is the driving force behind the progress of the hand at this point, and it is usual for the caller (who is in this scenario OOP and acts first) to check to the aggressor. This is called acting in flow, and for the flat caller to bet into the last aggressor instead is known as lead betting or leading out . This is often not recommendable, but it's something of a sign of a beginner if you see a player leading in a lot of spots, as they may simply not understand the significance of game flow. That said, there are occasions where it may make sense to lead bet for strategic reasons, and GTOTrainer itself may suggest some lead bets in certain spots. These can still be hard to manage as strategies since they are unfamiliar to most players, but that may still make them interesting lines as your opponent may also be unfamiliar and uncomfortable facing these lines. BB in defence vs CO open As we can see however, in this case BB does not want to lead, at least not more than 1% of the time, even in the GTOTrainer simulation. We can surely simplify this 99% checking strategy into a 100% checking strategy without losing too much equity! Continuation Betting The preflop aggressor will usually have the option to continue betting in position after facing a largely procedural check from the OOP player. Let's look at GTOTrainer's recommended play on the flop for the CO here. CO on the flop facing a check from BB As you can see from the above image, GTOTrainer wants to take a mixed approach with most hands on the flop, meaning that we are mixing bets and checks with almost every hand combination. In reality one can take a more relaxed approach to this at most stakes and just mix up your gameplay without worrying about the exact frequencies. What's most important is to note here is that overall CO is continuation betting (c-betting) about 58% of the time. One common mistake made by a lot of regulars is to c-bet with too high a frequency without good reason, building a big pot without a coherent plan. There are some boards where you can c-bet a very high percentage. Hands most favoured for a check here include mid-strength suited and offsuit aces except for those featuring the A ♦ which always bet. Some weaker top pair, underpairs to the J as well as some 6x tends to check here quite a lot also. GTOTrainer also recommends a mix of sizings here, but is mostly using 1/3rd pot as a sizing, which seems reasonable on a board with one paint card and some lowball cards. The more connected the board, generally you'll want to use a larger sizing more often, although your betting frequency may go down. Responding to a Continuation Bet Let's take a look at GTOTrainer's recommendations for the BB facing a small 1/3rd pot cbet from the CO . BB's strategy against a small 1/3rd pot cbet from CO As we can see above, BB is folding around 1/3rd of the time here. One mistake often made by students of the game is to overfold the BB . This is especially problematic if our opponent is cbetting even more than the 58% GTOTrainer recommends here, since we'll be folding against an even weaker range. According to the above we are supposed to call here with hands as weak as 33 and even some Ax and Kx hands with backdoors to hit flushes or straights on the runout. Another great takeaway from this data is the best check-raises BB has against CO's cbetting range. GTOTrainer prefers to check-raise a lot with top pair, as well as top two, and its sets, naturally. However it also recommends raising a high frequency with hands such as 65s and 64s, 54s, 53s and 43s . These hands block important two card or set hands, as well as having combinations of backdoor or gutshot straight draws. This makes them quite appealing check/raise bluffs since they can continue on quite a few turn cards as a multi-street semi-bluff or hit a big made hand on turn or river. How would we respond if we knew that CO was cbetting 100% ? In this case he would have a far weaker range on the flop, and would be forced to surrender many more hands if we raised, so we'd be incentivized to raise much more often, probably making any mixed raising hands into pure 100% raises, and adding in more marginal hands as well to this raising range. We would also likely flat call a larger part of our range, and fold less overall. If we didn't make these adjustments, CO would get away with making more money with his 100% cbetting range. In reality this is a flop where some less experienced players will cbet more than this GTO 58% , but it's unlikely many players will cbet 100% here these days, knowing that BB can play back with at least some strong hands and big draws. I'd expect to see lower than 70% cbet from most players in this spot across most stakes. Evaluating what the player pool tendencies are in a specific type of spot is also part of the skill set required for optimal decision-making. It is worth mentioning here the difference between GTO and optimal play. GTO , game theory optimal, is how we would play if all participants were perfectly aware of one another's ranges and strategic adjustments, and is an equilibrium which would be reached with this mutual knowledge and within a defined set of parameters, such as using a finite number of specific sizings. Change these parameters and you'd reach a different set of solutions, a different equilibrium. Optimal simply refers to what will be best in a specific circumstance against a specific player. So there are many simple examples of when we would want to play in a way which is optimal against a specific villain but nowhere near GTO . For example if we know our opponent never folds we can play very thin and very large bets for value, but rarely are we ever going to bluff them. This unbalanced strategy will be golden against a calling station, but suicide against a good observant regular. What about Turn and River? Naturally as we proceed down the streets, our decision tree is going to get progressively more complex. We will often face scenarios where multiple lines can be considered optimal, or where a mix of lines can be considered best, depending on what we think our opponents are thinking. Should we barrel turn after cbetting, or would a check/raise be better on a specific turn card? How often will villain call a turn bet on this card, but then fold to a blank river card (a card that changes very little with regard to the players' ranges) if we fire off a final bluff in a triple-barrel bluff line? These are some of the toughest questions in postflop poker, since the tree of decisions becomes ever more complex as we proceed across multiple decision points for each player. Rest assured, PokerDeals has you covered for this too, as we'll explore a wide range of specific spots in the postflop poker landscape in order to help you to map it all out. For more awesome hand discussion and theory, as well as all the latest in poker, head over to the PokerDeals Discord today! by Lucky Luke Image courtesy of PokerGO.com
If you're a tournament player, whether MTT or SNG, you just cannot do without studying ICM. Since it reflects the way ranges are impacted by payout pressure, we simply must make these adjustments to thrive in the MTT endgame or risk bleeding chips in all the wrong spots. ICM is a highly technical, and important part of tournament poker, particular endgame, as we'll explore below - thankfully there are some very simple ways to think about and study it which will help to make it far more accessible. We'll unpack these in the sections below. Just What is ICM? ICM stands for independent chip modelling, which sounds daunting but is really just a model for what your chips are worth at any one moment in a tournament. Unlike cash games, your chips don't have a fixed financial value in a tournament. That value is always shifting depending on the total number of chips in play and how they are distributed between the players, as well as the payout structure of the event. In a nutshell, you cannot simply sit out and collect your winnings as you can in a cash game. Stacks have leverage, the potential to translate into a cash, into a range of possible cashes in the payout structure. ICM is an attempt at a mathematical model which tells you what your stack is worth, and your opponent's stack, in a given spot, and how we should adjust our ranges based on that. In cash games we talk about chip equity value (cEV) which is the straightforward equation of $1 worth of cash game chips to $1 USD. In tournament poker we talk about $EV, which is an estimate of the dollar value of our chips in a tournament, in relation to all the other stacks and our position relative to the payouts at any one moment in the event. Think about the fact that survival has a premium in a tournament, especially so as you approach the bubble or a big payjump on a final table. Sometimes all you have to do is fold another orbit and you hit a payjump. This translates into the notion that holding on to the chips you already have (especially to your last BBs!) has more value than gaining the same amount of chips in the next hand. This is because your last chips represent your very existence in the tournament, once you lose those you are busto. Side Note: Stack Utility There's also a little extra feature which factors in here known as stack utility but this is not included in an ICM model. Stack utility is the notion that whilst going from, say, 10bbs to 20bbs in stack makes a big difference to what you can do in-game (you can 3bet non-all in, you can min open more often and fold to a raise, etc.) whereas going from 45bbs to 55bbs in stack makes almost no difference. This is another kind of undulating value to chips which is not even covered by ICM calculations, and is a little harder to pin down, but worth thinking about in certain close decision spots nonetheless. Does ICM apply only near or after the bubble? ICM effects can be much more pronounced as you approach the bubble or a big payjump, but that's simply because the premium on survival is going up the closer you get to that cash. Doing nothing can make you money in this spot, so there's a larger incentive not to lose your very last chips. However, ICM will still be around and having some subtle effect at other stages in the tournament. Even in the early stages it might make certain spots which were 2% profitable in a cash game breakeven in a tournament, for example. Bear in mind there are also effects in tournaments which make us want to play wider, such as the presence of an ante enlarging the pot. Basic Principles of ICM Whilst the actual detailed impact of an ICM effect will depend on exact distribution of chips and payout structure, and can be hard to accurately intuit in-game, the general principles of the impact of the effect on different stacks is quite easy to outline, and will give you a good idea to get started with. The main impact of ICM pressure is that there is an incentive not to bust, so especially when you are the effective stack in an all-in situation (the shorter stack of those in the hand, and therefore at risk of busting) one significant effect is that you must call off against all-ins with a tighter range than you would do normally. This would also be somewhat true for a stack which stood to lose almost all its chips in an all-in spot but have a few bbs remaining. If you are one of the larger stacks, the impact of ICM on your play when battling small and middle stacks will be lesser, since you aren't risking your entire stack when you get all-in against these players. The ICM impact will be at its greatest for you when you are contesting a big pot against another big stack. If you are one of the smallest stacks, the impact of ICM on your play will also be a little more limited than for middle stacks. This is due to the fact that your stack utility is very limited, and that the blinds will wear down your stack to nothing if you don't take edges available. ICM is especially minimal for the shortest stack in an event, especially if there are no other fairly short stacks around, since there's really no purpose in waiting around for a payjump (unless the larger stacks are playing very fast and loose, in which case someone might bust out sooner than you). The ICM pressure tends to be greatest of all for the middle stacks, who are under pressure from everyone for a significant portion of their stack, and tend to hit payjumps by outlasting the short stacks. This means they are more restricted in what they can do. If Everyone Has to Fold More to Shoves, I Can Shove Wider Right? Not necessarily. Because there is ICM pressure on the open shover as well, although less than on the caller of the shove (assuming they have equal stacks). This is because the caller will be all-in and going to showdown every time they call, but the shover will not be all-in as often, as they will get some folds. So they will bust out less often overall, relative to their range strength. Nonetheless ICM pressure still applies, so there are two forces acting on the open shover. One factor is the tighter calling ranges behind, allowing the open shover to shove wider. The other is ICM pressure, causing him to tighten up again. In some scenarios these forces actually cancel out fairly closely, meaning the shover will want to be slightly tighter than in cEV spots, but not much. In other spots they can still go super wide, it will depend so much on the stacks, numbers of players to shove through, and where you are in the payjumps. What Exploitative Plays Can I Make in ICM Spots? The most important thing to note is that if your opponents are calling wider than they should in ICM pressure spots, this will really kill your profit with your more marginal shoving hands and make them big losers. So if you have reason to believe that your opponent is calling off even a bit too wide, you'll have to tighten up your shoving range considerably. Let's look at a quick example of this in action. Here is a 6-handed spot from a final table, where the Button has 10bbs, SB has 10bbs and BB has 8bbs (the other stacks range from 20-40bbs). Note that the Button can shove almost 38% of hands including those as weak as J9o. Just for comparison, let's look at these ranges in a cEV scenario (no ICM at all and disregarding any rake): Note that, as discussed above, although the blinds can call off significantly wider here, Button still shoves very slightly tighter. Now, let's return to the first set of ranges in ICM on the FT, but adjust the calling ranges so that they are a bit wider. We'll give BB any pair, any ace and a couple of extra suited combos, and SB any suited ace and an extra offsuit Ax combo, just altering ranges by a few percent. As you can see, this has had a very strong effect on Button's shoving range, reducing it from around 38% to around 29%, as it must be reduced by around a third. The worst offsuit shove has gone from J9o to QJo. So if in doubt about how well your opponents understand ICM adjustments, you must tighten up! Other exploits include sizing big instead of shoving versus players who are exceptionally tight, as this will achieve almost the same benefit as a shove without risking as many chips. What's the Best Way to Improve at ICM Spots? Aside from grasping the fundamentals, the best way to study ICM is simply to grind out an absolute ton of spots in a preflop calculator such as Holdem Resources Calculator, used for the screenshots above. It's not very intuitive how stack distributions will change ranges in these spots, so a lot of brute force solving is required to hammer the ranges into your memory. There's no real replacement for this, so get solving today! We hope you've enjoyed this article from PokerDeals, please hop in the PokerDeals Discord and let us know! by Lucky Luke Image courtesy of PokerGO.com
Pokerstars Deposit Bonus
- OFFER -
Partypoker First Deposit Bonus
- OFFER -
Red Star Deposit Bonus!
- OFFER -
Subscribe to Pokerdeals.comPoker Deals
Join the PokerDeals community and take advantage of exclusive content and giveaways!