Why the solver widens for antes — and tightens the button for straddles

There was a story I remember Matt Berkey telling about Phil Ivey. Phil Ivey signs up to play a tournament without knowing what the game is, shows up at the table to play a structure he's never played before, and proceeds to open the exact optimal size the first hand he played.

While this may feel like magic, there is actually a predictable pattern as to how open sizes are supposed to shift based on the format of the game you're playing. And if you don't understand how preflop strategy is supposed to shift as things change, you can very easily make large blunders you won't be able to overcome.

This article is going to show you exactly how you should adjust your preflop strategy as the structure of your game changes — whether that be with straddles or antes — so that you can construct a strong preflop strategy in any format.

Straddle games are everywhere now. They are no longer the exception — they are the norm. In Western markets, WPT Gold and WPT Global both run classic UTG straddles. GGPoker and its sister room Natural8 have built-in straddle functionality on cash tables. CoinPoker, BetOnline, and TigerGaming all support straddle formats. In Asian and international club ecosystems, WePoker has quietly become one of the highest-volume online poker environments in the world — and it runs straddle-default. PPPoker, PokerBros, ClubGG, X-Poker, and QQPoker round out the club-app side. Live high-stakes games have defaulted to straddle-on for years. Operators are embracing the format because it aligns with how people want to play: more action, more flops, bigger pots. That's great for casuals, and it's profitable for pros who know how to navigate changing game dynamics. More dead money makes the game fun to play. This is where poker is going.

The pattern — how preflop strategy should shift as the format changes — isn't symmetric across antes and straddles. Antes widen all preflop ranges. Straddles do something much more interesting: they widen early and middle positions modestly, but tighten the button dramatically. Not because of dead money. Because the straddle silently moves "last preflop actor" from the button to the straddle seat itself. Every strategic adjustment flows from that.

Here's what the solver actually does — position by position — and why the same "dead money" can widen one seat while narrowing another.

The format we're talking about

We're covering the classic UTG straddle — the 3-blind format. SB posts 1bb, BB posts 2bb, and a dedicated straddle seat posts 2bb more. The straddle seat acts last preflop, after the button. This is the format most common on WPT Gold, WPT Global, GGPoker, and Asian club rooms like WePoker, and it's what our solver data is trained on.

On other straddle variants — the Mississippi straddle (usually posted from a chosen seat with action restarting from left of the straddler) and the button-straddle (where the button is also the straddler) aren't in our data. The core mechanism generalizes: wherever the straddle sits, it adds one more live hand behind everyone and typically plays the last-voluntary-actor role. But the specific magnitudes and the position-by-position adjustments in this article are calibrated to the classic UTG 3-blind variant. Mississippi and button straddles shift which seats get hit hardest; the direction is the same, the numbers aren't.

When we talk about an ante game, we mean an additional chip each player posts before the hand. Standard live ante sizes run 0.25bb to 1bb per player — that's the range clubWPT, WePoker, CoinPoker, and most online straddle tables run. Our solver data covers 0, 0.25, 0.5, 1.0, 2.0, and 2.5bb, so the tables in this article lean on the 0 to 1bb range where the data overlaps real-world structures. The 2.5bb value is included for mechanism continuity at the extreme end, not because it matches a common live game.

Antes widen everyone. Straddles don't.

Look at what happens to each position's opening range when you add a 2bb straddle to a no-ante 6-max Cash game.

Source: book-3 §1.5. 6-max, 100bb, 0-ante. VPIP = combo-weighted per-hand play rate across all 169 hands. See Methodology & caveats for full conditions and limitations.

UTG, MP, and CO each widen by 2.5 to 3.8 percentage points. This is the dead-money effect most players expect — chips in the middle before the hand starts means better pot odds for opening, so you open a bit more. Completely unsurprising.

The button is the story. It moves 14.4 points in the opposite direction — from 43.3% of hands down to 28.9%. That's roughly a third fewer opens. In a straddle game, the solver opens fewer hands from the button than it does from the cutoff. The button has been demoted.

The reason is cleaner than "the straddle stole the button" suggests. In 2-blind, when folded to the button, it faces two live hands behind: SB and BB. Both are forced-action defenders — they've already posted blinds and can only react to the button's open. In 3-blind, the button faces three: SB, BB, and the straddle. One additional voluntary-actor behind, and the additional actor has the best preflop position at the table.

The button's value in a 2-blind game comes from one specific strategic role: it's the last seat that gets to decide whether to open a hand from scratch. After the button, the only players left are SB and BB, defending a bet someone else made. The button holds the "last-to-decide-to-initiate" privilege with full information on everyone earlier in the hand.

In a 3-blind game, that privilege moves. The straddle posts 2bb mandatorily but acts last preflop — it sees the button's open, SB's decision, BB's decision, and then decides. Functionally, the straddle is now the last voluntary actor with full information on the table. That's what made the button valuable. The straddle has it now. The button doesn't.

This is the sense in which the straddle "steals the button": the role, not the seat name. The button still sits in the same chair. What changed is that there's now one more live hand behind it, and that extra hand plays the strategic role the button used to own. Our apples-to-apples comparison confirms this below: the button in 3-blind plays like CO in 2-blind — same live-hand count behind, same VPIP (~28%).

Here's the same finding at the hand level — the solver's full 169-hand BTN opening grid for both formats.

The high-card offsuit hands are the biggest casualties in the 3-blind grid. Hands like K7o, Q8o, J8o, T9o, the whole 98o / 87o / 76o offsuit connector column — all standard button opens at 2-blind — drop out entirely or become rare in 3-blind. Small suited aces thin out too. What survives: premium pairs, big suited broadways, and strong suited aces and connectors — hands that have clear postflop value even without positional information to lean on.

Is it really position, or is it the dead money?

Here's where the story sharpens. The 14.4pp BTN drop we've been pointing at is measuring two things at once: the straddle introduces both more dead money and one more live hand behind you (3 instead of 2). If you want to isolate the dead-money effect from the extra-live-hand effect, you need to compare two scenarios with the same number of live players behind.

The cleanest apples-to-apples available in the data: CO in a 2-blind game and BTN in a 3-blind game. Both have 3 live hands behind the opener. Blinds differ by 2bb (CO-2b = 1.5bb dead money, BTN-3b = 3.5bb dead money). Here's what happens:

Fresh solver queries, April 2026. Both scenarios have 3 players to act after the opener, so live-hands-behind is held constant; the +2bb dead-money difference is the only varying input. Isolation isn't perfect (CO vs BTN still differs mechanically). See Methodology & caveats.

When you hold live-hands-behind constant, the dead-money effect contributes only 0.8 percentage points to VPIP. The remaining ~13 points of the BTN 2-blind-to-3-blind drop come from the extra live hand behind — not from the extra chips in the pot. The straddle's "dead money" framing is almost a red herring. What's really changing is that the button now has three potential opponents behind it instead of two, and the extra one has strictly better position.

Sizing tells the other half of the story: even with identical VPIP, BTN-3b opens to 4.4× what CO-2b opens to. The raise size is the solver pricing out the last-acting opponent behind the button — the straddle — not buying more of the pot.

Which hand classes benefit (and which get cut)?

Aggregating the per-hand grid by standard coaching classes:

The premium end (top of the table) is unaffected. Every premium pair, big pair, big-ace suited and offsuit, and strong suited broadway already opens 100% in both formats — their EV goes UP in 3-blind because the pot is larger, but their play rate is already pinned at 100%.

The cuts happen in the middle and below: small pairs, suited connectors, suited 1-gappers, and weak offsuit aces lose 20–45 percentage points of open frequency. These are the speculative hands — they need postflop runway and positional edge to realize equity. Take away position, compress SPR, and they stop clearing the profitability bar.

This is the concrete answer to "what do I cut as BTN in a straddle game?" The speculative middle of your 2-blind range disappears. Premium hands stay. The playbook flips from "open everything playable" to "open only hands strong enough to stand on their own cards."

Sizing makes the same point

The button doesn't just tighten. It sizes up. Average button raise size jumps from 4.8bb in a 2-blind game to 14.5bb in a 3-blind game at zero ante, and climbs to 19.6bb at 2.5bb ante + straddle. Nearly one in five button opens at that extreme are all-in. The button is no longer raising to isolate and play a pot postflop — it's raising to price out the straddler or to limp-fold quietly instead.

⚠️ Review note for Brad and Thanh — remove before publish. The 4.8bb baseline button open size in 2-blind is worth a sanity check. In previous versions of the model, BTN opened smaller than 4.8bb predominantly; the current version sits at 4.8bb and the shift seems meaningful. Brad — could you look at the range view for BTN opens across a few situations and see if there's an explanation for why the baseline sizing moved? Not blocking the article's asymmetry story (the 4.8→14.5bb delta is the point, not the absolute baseline), but it's worth understanding before we publish.

Contrast the other positions: UTG, MP, and CO all open between 2.2× and 3× normally whether a straddle is in play or not. Their sizing barely moves. Only the button sees a regime change.

Why the asymmetry: three mechanisms

Three effects compound to tighten the button while leaving (or slightly loosening) every other position.

1. Three live hands behind, not two — and the extra one has the best position

In 2-blind, the button faces two live hands behind: SB and BB. Both are forced-action defenders. They've posted blinds and can only react to the button's open, not initiate. In 3-blind, the button faces three live hands behind: SB, BB, and the straddle.

That's one more potential 3-bettor — which alone matters — but the more important thing is which seat got added. The straddle acts last preflop. It gets to see the button's open, SB's decision, BB's decision, and then decides. Strictly better preflop information than the button has.

In 2-blind, the button held the "last seat with an open-or-fold decision" role, with full information on the CO / MP / UTG actions ahead of it. In 3-blind, that role shifts to the straddle. The button still acts before SB and BB, but now the straddle behind it has strictly better info than the button does. The button isn't choosing to open as the best-informed seat anymore. It has to price against the straddle's last-word option.

The button compensates by tightening (fewer marginal hands, since they face a strictly better-informed opponent behind) AND by sizing up (price the straddle out of calling with marginal hands). Both are visible in the data — the 14.4pp VPIP drop and the 3x raise-size jump.

(A small but important nuance: in 2-blind, BB technically has the last preflop action — but BB is reacting to a raise, not making an open-from-scratch decision. The button was the last seat with an "open or fold?" decision. In 3-blind, that decision moves to the straddle.)

2. SPR compression

A classic UTG straddle at 100bb effective reduces preflop stack-to-pot ratio from 66.66 (1bb SB + 1bb BB) to 28.57 (1bb + 1bb + 2bb straddle). That's less than half the postflop runway.

At lower SPRs, raw equity starts to matter more and positional edges start to matter less. An extreme version of this is obvious: if everyone has 2bb in the big blind facing a button shove, there are no postflop decisions to make — the button's positional advantage is zero, because the whole hand collapses to preflop equity. The same directional effect operates (in milder form) going from 100bb-in-2-blind to 100bb-in-3-blind.

The button's willingness to open marginal hands depends heavily on its ability to exploit position postflop. Compress the postflop runway, and marginal hands stop being profitable opens.

3. Open-size scaling

This one is more a consequence than a mechanism, but it feeds back. When the button does open in a 3-blind game, it has to open larger to price out the straddle (4.8bb becomes 14.5bb). That larger open discourages defense from the blinds — fewer callers, more 3-bets-or-folds. Which in turn means the postflop pots that do materialize are more polar and harder to navigate. Which means the button has to tighten further. The loop runs tighter → larger → tighter until the solver settles at 28.9% opens + 14.5bb sizing.

Note on "equity compression"

It's tempting to lean on "equity compression" language here — hands running closer in value at lower SPR. That's really a PLO framing and it doesn't cleanly transfer to NLHE cash games. The effect at play here is positional reshuffling plus SPR squeeze, not the equity compression from four-card combinatorics you hear about in PLO content.

Ante + straddle: the two effects oppose each other at the button

Here's one of the cleaner findings in our research: antes and straddles don't compound at the button. They subtract.

Source: book-3 §1.5. Ante sweep covers 0 to 2.5bb; live games typically run 0.25–1bb. See Methodology & caveats.

Ante pulls the button wider (there's more dead money in the pot to capture). Straddles pull the button tighter (squeeze threat from the 3 players still to act behind). The ante effect is stronger in magnitude, so the button still widens with ante + straddle — but always 19-25pp below the equivalent 2-blind game at every ante level. The gap never closes.

For UTG, MP, and CO, both levers widen the range. They compound cleanly. At 2.5bb ante plus straddle, CO opens about 67% and UTG opens above 44%. These positions get easier with each dial you turn up.

The button gets harder, and at the extremes, it gets weird.

The limp-jam regime at the extremes

At 2.5bb ante + 3-blind, 19.1% of button opens are all-in, and the average button raise reaches 19.6bb. The dead money is so large, and the squeeze threat so real, that mid-sized button opens just don't work — anything not close to all-in gives the straddle excellent odds to call and outplay postflop.

The grid shows the regime change hand-by-hand. Premium holdings — AA, KK, QQ, JJ, 99, AKs — dominantly limp in (call 2bb), trying to see a flop cheaply with a strong hand rather than bloat the pot preflop and face a straddle 3-bet. Off-suit broadway aces (AJo, ATo, KQo) and suited connectors (T9s, 98s) dominantly jam all-in, trying to realize their equity in one move or win the dead money outright. Weak hands fold. The middle of the chart — hands that in a 2-blind game would be comfortable 3x opens — simply has nowhere to go. There is no mid-sized raise that works.

So the solver collapses the button's strategy into three options: limp in with speculative hands that want to see a flop cheaply, jam with premiums that want to realize equity fold-or-get-it-in, or fold. The 3x-to-4x button open that works in every normal game simply disappears.

This is where Brad's Phil Ivey observation really lands. The "correct open size" in a big-action game isn't a linear adjustment. It's a regime change.

What to do at the table

Different seats, different prescriptions.

If you're UTG, MP, or CO in a straddle game

Widen modestly. The dead-money instinct is right for you — these are the seats that benefit from both levers. At 0bb ante in a 3-blind game, add 2 to 4 percentage points to your opening range per position. At compound ante + straddle, compound the effect: CO opens two-thirds of hands in a big game, which means virtually every suited hand, most connected offsuits, and a deep value range.

Don't overdo it. +4pp is real; +15pp is a leak. Stay grounded in what the solver actually widens to, not what "big action" feels like.

If you're on the button in a straddle game

Tighten sharply. This is the biggest instinct-trap in these formats. "I'm on the button, I can open anything" — that habit is a disaster in 3-blind games. At 0bb ante, cut your button range by roughly a third: play it more like CO of a normal game. At compound ante + straddle, expect your button decisions to collapse into limp / fold / jam. Stop raising to 3x — it doesn't work, because the straddle has pot odds to call with too much.

If you're the straddle seat

You bought the button. Use it. Your opening range when folded to is analogous to what the button opens in a 2-blind game — you're the last preflop actor, you have position info, you should be playing aggressively and pressuring everyone ahead of you. This is where most of the money gets made in straddle games by people who understand the format.

If you're in the big blind in a 3-blind game

Defend very wide — 75.5% at 0bb ante, climbing to 92.7% at 2.5bb ante plus straddle. The straddle's forced 2bb is pot equity you paid for (indirectly), and folding it away is leaving money behind. In big-action games the big blind is almost always defending.

Takeaway

Antes build bigger pots. Straddles build bigger pots but add an additional defender from the blinds. Ante widens everyone's ranges. The straddle widens UTG–CO, but the button is forced to play tighter. The straddle widens early seats and steals the button.

If you take one thing from this piece, let it be the button adjustment — it's the biggest single strategic shift going from 2-blind to 3-blind, and it's the thing most players get wrong. You're not on the button anymore. The straddle is.

Coming up next

Dead money reshapes the ranges you should be playing, but it doesn't tell you whether the game is beatable in the first place. In the next piece, we'll look at when rake and pool composition combine to make a game unbeatable — and the specific conditions under which even a perfectly-calibrated player can't beat the house and the field simultaneously.

Methodology and caveats

This section is deliberately verbose for the review pass so Brad and Thanh can verify how every number was produced and understand the limitations. Before publishing, most of this gets trimmed into a short footnote.

Data source

All numbers in this article come from the QuintAce trained-solver family, queried via strategy_grid API. Measurements as of April 2026, from the model captured in book-3 Cash Format Transitions v1.8.0 and supplemented with direct live queries on the same serving endpoint.

Base conditions unless otherwise noted:

• Game: 6-max NLHE cash
• Stack depth: 100bb effective, equal stacks
• Open size: 2.5bb (standard small-blind-denominator open)
• Rake: 3% capped at 3bb
• Blind structures: 2-blind (SB + BB) or 3-blind (SB + BB + 2bb UTG straddle). Straddle seat acts LAST preflop, after BTN.
• Ante: 0bb unless a specific ante sweep is named (0, 0.25, 0.5, 1.0, 2.0, 2.5bb)

What "VPIP" and "open %" mean

In this article, open VPIP is the fraction of dealt combos where the solver takes any non-fold action (limp, raise, or all-in) when folded to the specified position. It's measured by summing the combo-weighted per-hand open frequencies and dividing by the total combos in the range. "Open %" is the per-class or per-hand version of the same metric.

For hand-class aggregates, the calculation is a simple arithmetic mean across the hand names in the class (one vote per name). This is hand-name-weighted, not combo-weighted — e.g., in the "Offsuit trash" class of 44 names, each offsuit hand counts equally. For single-type classes (all-pair or all-suited or all-offsuit), this gives the same answer as combo-weighting; for classes mixing types, it introduces small distortion that we haven't corrected.

What "avg raise size" means

The mean raise amount across all non-fold actions the solver takes for that hand/position/scenario, weighted by how often the solver takes each action size. A solver that splits between raise-to-4bb 40% of the time and raise-to-8bb 60% of the time would show avg raise 6.4bb.

Per-hand range grids

The 13×13 grids render 169 NLHE starting-hand names (pairs on diagonal, suited above, offsuit below). Each cell's color represents the per-hand play frequency (100 − fold %). The 5 color bins (open / mostly open / mixed / rare / fold) are coarse by design — bin boundaries at 95%, 60%, 20%, 5% mean some edge cases can look different than the aggregate class average. The goal is pattern-at-a-glance, not precise frequencies; hover a cell for the exact number.

The apples-to-apples comparison

The comparison of CO (2-blind) vs BTN (3-blind) isolates the effect of added dead money (+2bb from straddle) while controlling for live-hands-behind (3 in both cases). Ideally we'd compare same-position-with-different-blind-structure (e.g., SB in 2-blind vs BB in 3-blind), but the strategy_grid client's Preflop.open doesn't cleanly set up a BB-opening-in-3-blind scenario because it doesn't reorder preflop action to put the straddle last. Writing a custom payload is on the follow-up list.

Known caveat: EV baseline in 3-blind

The solver returns a per-hand EV metric alongside the action frequencies. We originally intended to use this to show "which hand classes benefit most from the format change," but we discovered that folded hands show EV = −0.500 in 3-blind but ≈0 in 2-blind. The difference is consistent across all folded hands regardless of hand strength, suggesting a format-dependent baseline we don't fully understand (likely a rake contribution, an equity-realization correction, or a normalization constant). Until the gameplay-ai team confirms what the baseline represents, cross-format EV aggregation isn't defensible, so we removed EV columns from the class table and rely on play-rate deltas instead.

Hand class definitions

The 19 hand classes used in the class table are our own groupings (not the poker-knowledge KB's canonical T4 taxonomy). They're designed to align with standard coaching language:

• Premium pair = AA, KK, QQ
• JJ, TT = the first tier of pairs that don't always 4-bet
• 99, 88 = middle pairs with set-mining equity
• Small pair = 77, 66, 55, 44, 33, 22
• AK–AT suited = premium suited aces
• AK–AJ offsuit = premium offsuit aces
• A9s–A5s, A4s–A2s = mid and wheel suited aces
• AT, A9 offsuit, A8o and below = mid and weak offsuit aces
• KQ–JTs = suited broadway
• KQ–JTo = offsuit broadway
• Suited connector = hands like T9s, 98s, 87s, 76s, 65s, 54s
• Suited 1-gapper = T8s, 97s, 86s, ...
• Suited 2-gapper = T7s, 96s, ...
• Other suited = wide-gap suited hands
• Offsuit connector = T9o, 98o, 87o, 76o, ...
• Offsuit 1-gapper = T8o, 97o, ...
• Offsuit trash = everything else not ace- or broadway-involved

If a hand falls on a boundary (e.g., ATs is classified as "AK–AT suited" not as "suited ace mid"), the classifier uses the first matching rule top-down. Brad or Thanh could override any group definition before we publish.

Things we didn't measure or verify

• Only the current model version (April 2026 serving snapshot) was queried. Numbers would differ on older model checkpoints.
• We didn't cross-check these open-range frequencies against an independent solver (PioSolver, GTOWizard) at matching conditions. The order-of-magnitude numbers align with published solver output, but we haven't confirmed them cell-by-cell.
• The "aligned game popularity" framing in the hook (WPT Gold, WPT Global, WePoker, etc.) is editorial; we haven't back-checked each platform's exact straddle rules or volume numbers.
• "Classic UTG straddle" in our data is the "3-blind" format with the straddle seat acting last preflop. Real-world UTG straddles sometimes act just-after-BB (second-to-act). Our numbers apply to the acts-last variant only.

Numbers flagged for follow-up before publish

1. Baseline BTN 2-blind avg raise of 4.8bb — this is a recent shift from earlier model versions that Brad flagged during review. Needs sanity check before we lean too hard on the 4.8→14.5bb regime-change claim.
2. EV baseline in 3-blind (−0.5 for folded hands) — needs a definition from the gameplay-ai team.
3. Cash BB vs BTN 3-bet 18.1% — at the high end of published benchmarks (12–17% typical). Not in this article but affects Article 2 (the rake/fishiness piece) which also cites Cash baselines.

Research by QuintAce. Writing by Brad Wilson (Chasing Poker Greatness).