Balatro discard calculator
Compare possible replacement draws against playing now in a specified small-deck state.
Which cards should I discard?
The useful comparison is between playing now and the distribution after a discard. Specify the actual remaining draw pile: a smaller sample pile is a different state, not a faster full-deck estimate. Every repeated card in the pile is counted separately.
Four-card discard with five possible replacements
Discarding 2D, 4C, 9S and 9H retains AH KH QH JH. Four cards are drawn uniformly from a five-card pile. Four of the five equally likely replacements include TH, permitting a royal flush worth 1,208 chips. Without TH, the best available immediate score does not clear 600. The selected action value is 0.8 under this model; it is one of the optimal actions, not necessarily unique.
Exact inputs
{
"contract_version": "balatro-web-v1",
"state": {
"id": "worked-example",
"hand": [
"AH",
"KH",
"QH",
"JH",
"9H",
"9S",
"4C",
"2D"
],
"deck": [
"TH",
"9D",
"AS",
"KC",
"2S"
],
"hands_left": 1,
"discards_left": 1,
"chips_needed": 600,
"jokers": [],
"levels": {}
},
"explain_actions": [
{
"kind": "discard",
"cards": [
"2D",
"4C",
"9S",
"9H"
]
}
]
}Computed result
Optimal model value 0.8. 2 actions share that value; this page does not claim a unique recommendation.
discard 2D 4C 9S 9H
Action value 0.8. Weighted branch sum 0.8; probability mass 1; residual 0; tolerance 1e-10.
| Draw or terminal outcome | P(draw) | Conditional value | Weighted contribution |
|---|---|---|---|
| 9D TH KC AS | 0.2 | 1 | 0.2 |
| 2S TH KC AS | 0.2 | 1 | 0.2 |
| 2S 9D KC AS | 0.2 | 0 | 0 |
| 2S 9D TH AS | 0.2 | 1 | 0.2 |
| 2S 9D TH KC | 0.2 | 1 | 0.2 |
Engine 0.1.0; rules balatro-001a-rules-v0.1; interface balatro-web-v1; status completed. Snapshot originated on aarch64-apple-darwin and is regenerated after validated engine releases.
Machine-readable request and native response · Load the state · Permanent example page
Why conditional continuation matters
The action value is the sum of each draw probability multiplied by the optimal continuation value for that successor state. The explanation uses the same hand/discard transitions as the solver, and its weighted sum must reconcile within 1e-10. Drawing a visually stronger hand alone is not the objective.
Scope and trust
Supported model: 20 validated Jokers, 150 catalog names; exact analysis accepts 1–8 hand cards, 0–12 draw cards, 1–2 hands and 0–2 discards. Uniform replacement draws and optimal continuation are assumed. Unsupported boss, consumable, shop, history and scaling effects are rejected. No standalone sampled full-deck advice. Card order is optimized; Joker slot order is preserved.
Mathematical action values are conditional on the stated game model, not estimates of real-game success. Equal values are ties. Incomplete searches provide no completed recommendation. This independent tool is not affiliated with the game publisher.