Solvers for 2-7, badugi, razz and stud.

From dedicated learners moving up through the mixed-game stakes, to established high-stakes and tournament regulars. The solver doesn't tire, doesn't tilt, and gives you the same answer every time.

Most players see improvement in the first few sessions — preflop ranges and the fold and value thresholds in common spots are quick to absorb. The players who get the most out of it put in hundreds of hours.

2-7 single draw, 2-7 triple draw, badugi, razz and stud hi. The three draw games are available heads-up and 6-max. Razz and stud hi are heads-up and 4-max, with simplified 6-max coming soon.

Every package is a blueprint: the whole game pre-calculated with Monte-Carlo CFR (MCCFR) over a bucketed abstraction. It is solved once, ahead of time — bucketing groups strategically-similar hands so the entire tree (every position, street and format) is covered end to end, rather than solving a handful of spots exactly and leaving the rest. Nothing solves while you use it.

A separate, non-bucketed resolver is the next part of the project — in development for higher precision in heads-up, two-players-left late-game spots, where an exact solution matters most. It complements the blueprints rather than replacing them.

Maybe — reach out and we can discuss it.

No. This is a draw and stud specialist. For hold'em and Omaha there are other tools built for them.

No. Customer identities are confidential, and it's not in my interest to hand information to your opponents. Anyone mentioned, publicly or privately, gave permission first.

Yes. There's heavy overlap between tournament and cash strategy, and the ranges transfer directly. Bring the structure you play and study the spots that actually come up.

For amateur players, a coach can be a good choice — reading solver output well takes some grounding in the game.

For semi-professional level and up, the few players ahead of you aren't going to coach you honestly without a steep hourly plus a share. The solver trains for unlimited hours, stays consistent, is stronger than any human, and won't teach you bad habits.

The algorithm behind every blueprint, MCCFR, is proven to converge: given enough iterations its average strategy provably approaches the game's equilibrium, and the remaining exploitability is something we can measure. Blueprints are trained until that gap is small.

Coverage isn't uniform, though — the solve spends its iterations where hands actually go, so common nodes are sharply resolved while rarely-visited nodes get fewer samples and stay coarser. That is exactly what the resolver is for: it re-solves a low-frequency node on its own, without bucketing, so it can approach very high accuracy.

Beyond the theory, the blueprints have been tested extensively against other established mix-game solvers: run the same spot in both and the action frequencies for both players match very closely — an independent check that the strategies hold up.

You can also check it yourself. Solve some simple spots and compare — start with lines you already trust, confirm the solution agrees, then let it surprise you on the ones you don't.