Poker Coverage: Poker Tournaments Casino News Sports Betting Poker Strategy

Bracelet Winner Justin Saliba On Deviating From GTO Tournament Strategy

by Craig Tapscott |  Published: Jul 27, 2022

Print-icon
 

Justin Saliba Justin Saliba learned the game of poker around the family kitchen table at age 10. His older brother would have friends over for home games and to watch poker on TV, and the younger Justin would worm his way into the action.

His play moved online, although Saliba’s focus turned to chess and soccer, where he was a standout goalie. He ended up playing soccer at the University of Dayton, where he won a league championship and also graduated with a degree in chemical engineering.

In 2019, he decided to get serious about poker, and moved to Las Vegas to play cash games. When COVID shut down live poker, however, Saliba began to concentrate on tournaments online. He found quick success, especially at the World Series of Poker Online series in July 2021, where he took down the $5,000 buy-in freezeout to earn his first career bracelet and the $253,800 prize.

The 27-year-old has since branched out to the live tournament circuit, and so far in 2022 has 21 cashes and eight final tables. When he’s not playing, he’s teaching the game as a six-max cash game and GTO tournament specialist for PokerCoaching.com/CardPlayer. You can find him on Twitter @Justin_Saliba.

Card Player caught up with Saliba during the WSOP to get some tips for implementing game theory optimal strategies in tournaments.

Craig Tapscott: How do you adjust from non-ICM adjusted GTO strategies to situations where the Independent Chip Model applies? Can you give some examples of how you should adjust when there are payout implications in a tournament?

Justin Saliba: Adjusting for ICM is all about understanding the balance between expected value and risk. When you’re just playing a strategy purely based on maximizing your expected value, you should take every single profitable spot, even if the return that you’re expecting is 1/100 of a big blind in value.

As ICM becomes more important though, that 1/100 of a big blind in value isn’t worth the risk of taking on variance and possibly busting out of the tournament. An example would be if you were second in chips with six people left at a standard ICM situation. You have 50 big blinds, the chip leader has 55 big blinds, and there are four stacks under 10 big blinds. Action folds to the small blind chip leader who goes all-in for your 50 big blinds and turns his ASpade Suit QSpade Suit face up.

You peel the 10Club Suit 10Diamond Suit and know that you’re a favorite to win about 54% of the time. For chips, this is great, and would be a snap-call. But, going broke here is a nightmare given the payouts, so the risk outweighs the expected value in this spot, and the correct play would be to fold. 

The balance between expected value and risk goes much further than just simple preflop spots though. It affects your opening ranges, continuation betting frequencies, river bluffing frequencies, bet sizes, and much more. Typically, when you are the big stack, you are the one who can put other people’s stacks at risk, so you can open wider, continuation bet more, bluff at higher frequencies, and show more aggression across the board.

When you are covered though, the strategy flips. You have to play a much tighter, aggressive strategy. You don’t want to take passive actions preflop like cold calling or limping because this takes on more risk on future streets. Break-even calls, such as big blind defends all, aren’t worth it to play due to the risk that you’ll have to take on future streets and you’ll have to over-fold flops, turns, and rivers playing out of position as the covered player. 

A spot that comes to mind that shows this concept comes from a recent online high-stakes tournament that I played. We were five-handed, and the stacks were HJ: 53 bb, CO: 21 bb, B: 25 bb, SB: 59 bb, BB (me): 34 bb. The hijack opened to two big blinds and I defended the QClub Suit JHeart Suit in the big blind. The board came ASpade Suit JSpade Suit 3Diamond Suit and the HJ bet 5.1 bb into a 5.1 bb pot.

I simply folded and moved on to the next hand. On the surface, it seems somewhat trivial, but if I only cared about winning chips in this spot, I would have an easy call on the flop. If we look at this using ICM tools such as MonkerSolver, we’re able to learn that my BB-defend range is significantly tighter preflop than it would be if we were just playing for chips. Also, my flop defense frequencies are significantly lower because it is simply not worth the value in chips to take on the amount of risk that’s possible by playing a large pot out of position against a covering stack with a bluff catcher that rarely improves to a nutted hand. 

If you can constantly try to balance the value of gaining chips with the risk of ruin or changing your future profitability in decisions, you’ll be able to make much better decisions at final tables than someone not keeping it in mind. 

Craig Tapscott: How do you decide whether to try and implement GTO strategies or if you should use more of an exploitative strategy? Can you give some examples of adjusting from what you thought was correct in theory to what you did in game? 

Justin Saliba: In general, I want to do my best to execute theoretically sound strategies, while always paying close attention and constantly looking for reasons to deviate and play an exploitative strategy.

If I had a rubric of how my opponents played, I’d never play GTO. If they folded too much, I’d bluff more. If they were calling stations, I’d go for thin value more and bluff less. But poker isn’t that easy.

I constantly want to be putting my opponents in difficult spots, and that happens when playing tough GTO strategies with a proper understanding of the ranges at play. As I collect and observe more data points and information, I’ll look to use that to my advantage, but I will always stay paranoid about the possibility of counter-exploitation. Every deviation that you make opens the door for them to counter you and really punish you.

Imagine you thought someone folded too much, but that somehow, they knew that you thought that. You decide to c-bet more often, especially with bluffs. They counter you by increasing their check-raise frequency with bluffs, and all of a sudden two months go by before you realize that your strategy is being ravaged by their counter-strategy. Overall, I want to be extremely aware at all times and look for spots to make small deviations, and the weaker the opponent, the happier I am to make larger, very exploitative deviations, but I’ll always try to keep my default strategy very theoretically sound and difficult to play against, no matter the opponent. 

A clear deviation that I made recently was from a recent $25,000 buy-in against a world-class player. I think it’s a fun hand to show that even when playing against the best, sometimes, it’s unlikely that they’re going to play like a computer. 

I raised two big blinds with KSpade Suit 9Diamond Suit from the CO off of 45 bb and the player called in the big blind. The flop came JHeart Suit 6Spade Suit 5Diamond Suit. This is a board where you have three real options for your overall strategy. You can bet large, pushing your advantage with very strong hands like good J-X and overpairs. You can bet small pushing your overall equity advantage, getting value with weak jacks, middling pocket pairs, and even good ace-high hands, and you can check. All three strategies should be used on this flop with different parts of your range.

For both bet sizes, you need to have bluffs, and my hand fits nicely as a bluff, so I chose to bet 5.5 big blinds (100% pot), and the big blind called. The turn brought the JDiamond Suit, pairing the top card and bringing a back-door flush draw. 

On this turn, we really just want to use one bet size (a medium bet size of 60-70% pot). We don’t really have many thin value hands that want to bet small, and the board can’t change all that much to incentivize a larger size. When thinking about our hand, the K-9 does well to block the opponent’s K-J and J-9 combinations, so mostly bluffing with this hand on the turn seems great. I bet 10 big blinds into the 16.5bb pot and the opponent called again. 

The river was the ADiamond Suit and my opponent checked. I had 26 bbs back and the pot was 39 bbs. My 9Diamond Suit is now a great card to hold, preventing my opponent from having the 9Diamond Suit 6Diamond Suit flush that could be possible. I knew my combo needed to bluff at least sometimes, but I decided to pure check back and give up on the bluff. My opponent rolled over J-8 suited and I lost the pot. 

So… why did I deviate from what I thought was the best GTO play? Well… it starts on the flop. Even at the high-stakes, people don’t use large sizes in a ton of spots. I think that when I bet 100% on the flop, this takes most people by surprise, and they end up over-folding. The second reason is that the JDiamond Suit turn isn’t a very high EV turn for me here. My opponent has more J-X combos and my overpair advantage loses value because now I lose to trips.

I think in general, even at the high stakes, most people under-bluff low EV turns like this one. In my opinion, this leads to a +EV bluff on the flop and the turn… but by the river, I think my opponent’s range is going to be much tighter than a computer would play here causing them to have very few folds on this runout.
 
If we further examine what hands become indifferent from the out-of-position player in a GTO solver, the largest portion of the range are offsuit 6x (K-6 offsuit thru 7-6 offsuit all have to mix some amount of calls, but are heavy folds to the all-in bet on the river). This means that my opponent needs to look at Q-6 offsuit, call a 100% pot bet on the J-6-5 rainbow flop, and then continue to a 60% pot bet on the JDiamond Suit turn as we’re nearing the money in a $25k buy-in. I just don’t think that happens very often in game, so I chose to purely give up on my bluff due to the fact that I didn’t think my opponent would have a high enough fold frequency on the river. 

This can lead to me being counter-exploited, and if my opponent were to have shown down Q-6 offsuit, I would have to re-evaluate my deviation. But for now, I think it’s a very sharp strategy to implement and a nice spot to deviate from what a computer would do here. ♠