New preflop ranges: what changed in the RTA software this week, with the numbers

1. What a preflop range is in an RTA, and why it comes first

A real-time assistant answers the spot in front of you, but it can only answer it as well as it knows the ranges. When the app tells you to call a 3-bet with ace-jack suited, that answer rests on two things: the hands the 3-bettor holds, and the hands you would still have after calling. Both come from preflop. On the flop the solver's ranges are those preflop ranges filtered by what each player did; on the turn and river, the live solve runs on whatever reached it. A preflop chart that is a few percent off is not a few percent off at showdown. It is wrong in a way that compounds, because every street after it is solved against the wrong picture.

That is the case for getting preflop from a solver rather than from a coach's memory, and it is the case for publishing the numbers rather than hiding them behind a button. The free poker RTA charts page shows the ranges as tables; this post is the changelog behind them.

2. The solved six-max set: what is in it

The new set is a solved six-max game at 100 big blinds with a 2.5bb open, the format most cash players sit in. It covers four families of decision: the open from every seat, the defence facing an open (every seat against every earlier seat, fifteen spots), the opener facing a 3-bet (fourteen spots) and the 3-bettor facing a 4-bet (fifteen spots). Each spot carries the full mix for all 169 hands, so a hand that 3-bets 12 percent of the time and calls the rest is played at those frequencies by the app's roll, not rounded to one action.

A few of the opening ranges, to anchor it:

The defences are where a chart written by hand tends to be furthest from the solver. The big blind against a button open 3-bets 12 percent of hands and calls 45 percent, so it continues with well over half of all hands. Against a cut-off open it 3-bets 8.5 percent and calls 33.5. Against an under-the-gun open it still calls 26 percent, with a 5 percent 3-bet. In position the defence is tighter and more polar: the button facing a cut-off open 3-bets 9 percent and calls 12. Ace-jack offsuit in the big blind against the button is a 12 percent 3-bet and an 88 percent call; ten-nine suited is a pure 3-bet there; king-five offsuit is a fold.

3. A 3-bet is a multiple of the open, never a number

The old charts carried nominal sizes: 3-bet to 9bb, 4-bet to 22bb. That is right against a 2.5bb open and wrong against everything else, and real tables open to 3bb, 4bb, sometimes 5bb, with limpers and callers in the pot. The app now sizes the 3-bet as a multiple of the open actually made. The solved set's own multiples are 3.4 times the open in position (8.5bb over a 2.5bb open) and 4.6 to 5.2 times from the blinds (11.5 to 13bb), out of position needing the bigger size because the caller will realise less of their equity. Where the set does not chart the spot, the standard 3x in position and 4x out of position apply.

Two adjustments ride on top. The multiple eases as the open grows: a 3.4x 3-bet over a 5bb open would be 17bb, more than the spot needs, so the multiple comes down toward 2.2x in position and 2.8x out of position as the open gets bigger. And every caller behind the open adds one open to the 3-bet, which is the squeeze rule most solvers converge on: over a 3bb open and one caller, the in-position 3-bet is about 12bb rather than 9. The 4-bet follows the same logic as a multiple of the 3-bet that was made, 2.1 to 2.7 times it in the set (the button 4-bets the big blind's 13bb 3-bet to 27.5bb), and the 5-bet is a jam unless the stacks are deep enough for a size.

The frequencies facing a 3-bet are small and specific. The button, 3-bet by the big blind, 4-bets 3 percent of the hands it opened and calls 13.5 percent; the rest folds. Under the gun, 3-bet by the big blind, 4-bets under 1 percent and calls 6 percent. Facing a 4-bet, the big blind that 3-bet the button jams 2.7 percent of what it 3-bet and calls 4.3 percent. These are the numbers a player feels in their stomach as too tight, and they are the equilibrium.

4. Straddled pots are played in straddle units

Club games straddle, and a straddle changes the game more than a bigger blind does. The straddler acts last preflop, every other seat is effectively one position earlier, and the stacks are half as deep in the unit that matters. A flag this weekend from a Telegram club table showed what happens when an RTA ignores that: the 2bb straddle was read as an open raise, and the hero behind a limper with a pocket pair was told to fold to a "UTG1 open". Then the real raise arrived and the engine thought it was facing a 4-bet.

The app now recognises the straddle as a posted blind and plays the hand in straddle units. A 2bb straddle makes the straddler the big blind of a doubled game, the real blinds play as the small blind, each other seat moves one spot earlier, and the 100bb stack becomes a 50-straddle stack for the purpose of the charts. The advice comes back in the table's own big blinds, so "Raise to 8bb" means what the table shows. The result in that flagged hand: the same pocket pair, limped and then raised by the small blind with two callers, is a call at the price rather than a fold.

5. Limped, then raised: the chips already in set the price

That hand exposed a second rule. A player who limped and is raised is not a player facing an open; a blind's worth of their chips is already in and the pot usually holds several limps. The old logic answered the raise from the 3-bet chart's value half and otherwise folded, which is how a pocket pair with 7bb to call into a 43bb pot and 180bb behind got "Fold 100%". The engine now prices it: a pair calls when ten times the call or more is behind and the call is thirty percent of the pot or less, because a set is paid often enough at those odds; a suited connector or suited ace calls at twenty times the call and a fifth of the pot. Everything else still folds, and the limp-raise with the top of the range stays.

Antes work the same way, through price rather than through a separate chart. A table with a 1bb ante from every seat has 8bb of dead money before a card is dealt, and every open and every defence is priced against the pot it wins. The charts widen with the dead money, capped so that a chart never becomes a guess, and the ante is read off the table on the rooms that take one, including the Telegram clubs, where it was being logged as eight limps.

6. What the RTA software still plays from the older charts

Honesty about scope matters more in an RTA than in a study tool, because a player cannot check the source mid-hand. The solved set covers heads-up, three-handed and up to six-handed tables at 100 big blinds for the four families above. The squeeze and cold 4-bet families, the small blind's open with limps, seven- to nine-handed tables and 200bb play are still answered from the older charts, which are good but hand-built. Those families are being pasted in over the coming days, and full-ring and 200bb solutions follow once the flop library for six-max is seeded. Each release note on this blog says which spots moved.

Multiway pots are the one place no solver, ours included, has an equilibrium to offer: a three-way flop has no published solution, and the app says so in its note rather than inventing a frequency. The multiway post covers what it does instead.

7. How to read the preflop advice at the table

Preflop advice in GTO MAX AI shows the full mix, not a single verb: "Raise to 8.5bb 64% / Call 36%" with the roll's pick beside it, and a one-line note naming the spot it came from, such as "Facing BTN open to 3bb (40%): 3bet 11%, call 6%". The percentage in brackets is the opener's range as the set has it; the frequencies after it are your hand's. When a straddle is on, the note starts with "Straddle to 2bb: the hand is played in straddle units" so you know why the sizes look the way they do. When you limped and were raised, the note states the price and the stack behind that made it a call.

If a preflop line looks wrong to you, flag the hand from the app. Every one of the three rules in this post came from a player's flag within the last week, and the frame that showed the problem is now a test that stops it coming back.

More from the blog

• Bet sizing in solver trees: what 33, 75 and 125 percent mean
• GTO Wizard vs an RTA: the study tool and the live tool
• Playing GTO with real-time assistance: the edge and its limits
• Multiway pots: why no solver covers them and what to do instead

Longer reads: the RTA poker guide and all guides.