In-Position Raising In 3-Bet Pots as the Preflop Caller (Part II)

Part II – Practical Applications

As I mentioned in the first part of this series, raising in position in 3-bet pots as the preflop caller is the most underplayed and misplayed high-frequency line in the entire game tree.

In Part I, I focused on the underlying mechanics of the spot. In this part, I’ll shift toward how you can capitalize on your opponents’ mistakes and start making a lot more money when you find yourself in this situation.

Let’s dive in!

The Methodology

To analyze this situation in a scientifically valid way, we need to rely on hard, cold data. Don’t worry—I won’t go into deep detail here, just surface-level information.

The correct approach to analyzing patterns of deviation in pro-level human play is to first establish a baseline of what is optimal, then perform a comparative analysis against that baseline.

To achieve this, I compiled data from a representative sample of 184 flops using GTO solutions calculated with a solver (both preflop and postflop). I then imported this data into my custom-built Hand2Note pop-up, which I will use as the benchmark for what is “correct.”

For real-life strategies, the dataset I’ll use in this article comes from my own database of hands played at 500nl and higher on mainstream sites—very high levels of expertise.

Note: This comparison is effective because, on the aggregate, 500nl+ regulars demonstrate the same caliber of expertise preflop. They match the preflop frequencies and range compositions of the GTO strategy. Additionally, the postflop bet sizes they use—both for c-betting and raising—are the same as those in the GTO strategy, which confirms that these players have very high expertise in this domain.

GTO Poker vs Human Poker – Results

So, let’s first take a look at how the solver approaches continuation betting and how it plays against raises.

I’m just going to focus on what’s relevant for this article. I want you to pay attention to two things:

1. The top part labeled “Next actions,” where it says “FCR 28/46/25.” This shows the average fold/call/raise strategy against a raise (across all flop textures).
2. The column on the right-hand side of each flop composition, which also shows “FCR.” This represents the average strategy across the textures within that definition (see explanation and example below).

The numbers after “FCR” reflect the solver’s reaction when it gets raised after c-betting.

For example, on Ace-high rainbow (A-high RB) flops, when facing a raise, the solver folds 30% of the time, calls 50% of the time, and 3-bets 20% of the time.

Now, let’s shift our focus to how the expert human population plays this same spot.

Let’s first compare the averages:

• GTO – 28% Fold, 46% Call, 25% 3-bet

• Humans – 43% Fold, 42% Call, 16% 3-bet

What do we notice? Humans are playing much more risk-averse compared to the solver. They fold 15% more often, call 4% less often, and 3-bet 9% less often.

Now, you might say: “Well, what if they’re incentivized to play this way by the player pool?”

To which I would reply: “Very astute observation!”

To verify this, we can compare the solver’s reaction to c-bets with the expert humans’ reaction to c-bets.

In my GTO data, the solver raises 8% of the time against a c-bet, while in my reg population data, that number is 7%.

Is this a significant enough deviation to warrant such an overfold and under-3-bet?

I would argue that it is not.

That’s because other relevant metrics—such as won at showdown—are almost the same (within the 95% confidence interval). However, the fold vs 3-bet rate is much higher for the reg population. In the GTO strategy, the solver folds to a 3-bet 27% of the time, while humans fold 39%.

This final fact could be explained by four possible reasons (in no particular order):

1. The population is raising with too many hands that can’t call raises – highly unlikely, given the won at showdown stat and the known bias toward risk aversion.
2. The population is adjusting to the tighter 3-betting range of the Small Blind by over-folding – possible.
3. The population is 3-betting with a larger size than the solver – highly unlikely, when comparing the FCR frequencies of GTO vs. reg profiles.
4. The population is over-folding with hands that should mix between calling and folding at equilibrium – highly likely, due to the well-documented human bias toward risk aversion.

The importance of these factors could be determined with greater certainty through more in-depth research, but that is beyond the scope of this article.

What we can say, however, given the known human bias toward risk aversion, is that the most likely explanation involves either point number 2 and/or point number 4.

While this is insufficient evidence to completely rule out the deviation as an adjustment, I would argue that it is more likely a leak caused by a structural human bias toward risk aversion.

Moreover, overfolding appears as a consistent strategic bias across all flop textures—we can also describe it as a texture-invariant pattern.

GTO	Reg population

How to Exploit This Pattern

Now we’re getting to the actual EV-maximizing strategic upgrades.

To make the most efficient use of your time, I’m going to show you a simplified version of the process for obtaining these upgrades, so you can scale beyond the example I’ll provide.

Let’s dive in!

Ace-High Rainbow

I chose this texture for a few reasons:

• It’s one of the most common textures.
• It’s one of the most infamous textures.
• And, most importantly, because the real-life optimal strategy when facing a c-bet on this board is going to shock you.

Let’s start by reviewing the data on this texture. First, the GTO strategy:

We see that the solver c-bets at a 94% frequency across Ace-high rainbow textures (these do not include double broadway or wheel flops). It also folds versus a small raise (30% pot, roughly 2.5x) 30% of the time, while 3-betting 20% of the time.

Now, let’s compare that to how the reg population plays:

We see that high-level regs under-c-bet by 12%, overfold by 8%, and under-3-bet by 12% when faced with a raise.

So, let’s model these deviations into the solver on a fairly neutral As 8d 2h rainbow flop.

But first, here’s the GTO flop c-betting strategy:

And the reaction against the 2.5x raise:

We can see that the action frequencies for this flop texture are extremely close to those of the average Ace-high rainbow flops.

The next step is to model the deviations into the sim using the tool’s simple node-lock function—all while maintaining balanced range compositions, c-betting at a lower frequency, and 3-betting at a lower frequency.

So now, get ready for the grand reveal. This is the local maximum EV* strategy you can implement against this type of deviation:

Isn’t that something to behold? Not only that, but the expected value gained increased from 98.3 chips to 109.8 chips. With a big blind worth 10 chips, that means the net EV gain is 115 bb/100 hands compared to baseline.

There is a distinction to be made between local and global maximum EV strategies. The former (local) only considers deviations that occur in the sub-tree node, with the rest of the game tree re-calibrated toward a new Nash equilibrium strategy. The latter (global) accounts for the entirety of deviations across the game tree. With the tools currently available, only local maximum EV strategies can be calculated.

Now, let’s take a look at the EV comparison between different actions. Here’s the breakdown between raising and calling:

For reference, EV is expressed in bb/100 hands. So, if you look at 54s where it says “R 191 +114,” it means that raising to 191 chips generates 114 bb/100 more than calling.

This comparison helps in two ways. The first is that it establishes which hands perform relatively better by taking the aggressive line compared to the calling line.

The second way it helps is…

Resiliency

I’m going to define resiliency as the extent to which an exploitative strategy is likely to provoke a structural adjustment in the opponent’s play that reduces its own expected value.

Without going too in-depth, in this specific case, raising with 100% of your range may or may not create resiliency issues, depending on other factors.

If you’re concerned about playing such a strategy, one solid approach is to raise only with GTO-approved hands.

This approach greatly increases the resiliency of your strategy, since even when you reach showdown, the hands you reveal will both make sense and align with the GTO response for your situation.

So, let’s do that now. Here’s the GTO strategy against a c-bet:

We can see that the solver will raise with AJ+ for value, and it will also raise hands with backdoor flush draws such as QJs, QTs, JTs, 76s, and 65s. In addition, it raises hands without backdoor flush draws such as 98s, 87s, 86s, 54s, and 22–55.

If you pushed it to the max with only those hands, you’d be raising with roughly 42% of your range while still maintaining very high relative resiliency.

Wrapping Up

You’re now equipped not only with the knowledge of the exploitative opportunity and the process required to unlock tactical upgrades, but also with a strategy for flying under the radar longer—without giving up much of the potential EV gain. From here, you can scale the process to include tactical upgrades on every other board imaginable.

That’s all for this article! I hope you enjoyed reading it and learning from it. I had a blast writing it. As always, if you have any questions or feedback, please let me know in the comment section down below.

Till next time, have fun crafting your strategies!

To learn more about how to use uncommon aggression in your games, read: Beyond the Flop Overbet: Elite-Level Turn Play Explained.