Protocol v2 `strategy_grid` Query Formats

This document explains how to query /strategy_grid for Squid, Short Deck, Bombpot, MTT, and PLO spots. It is focused on request fields, response fields, and JSON examples.

The examples use chips, not BB, for stacks and action amounts. With big_blind: 100, a 100 BB stack is 10000.

Output examples are representative snippets. Live strategy probabilities, EVs, bucket ids, and model metadata will depend on the loaded model and spot.

Endpoint

POST <gto-glue-environment-URL>/strategy_grid
Content-Type: application/json

For Holdem-like v2 formats, send protocol_ver: "v2". PLO is selected before the v2 Holdem route by hand.game_type being plo4, plo5, or plo6; keep protocol_ver: "v2" for consistency.

Common Input Fields

All formats use a strategy-act-style top-level request with a required hand object.

Field	Type	Notes
`protocol_ver`	string	Use `"v2"` for the v2 strategy-grid route.
`strategy_min_threshold`	number	Optional. Drops tiny action probabilities and renormalizes. The UI commonly sends `0.05`.
`hand.game_type`	string	`"nlhe"`, `"shortdeck"`, `"plo4"`, `"plo5"`, or `"plo6"`.
`hand.game_mode_code`	string	`"normal"`, `"squid"`, `"bomb"`, or `"spin"`.
`hand.big_blind`	integer	Big blind in chips.
`hand.small_blind`	integer	Small blind in chips. Defaults to `big_blind / 2` if omitted.
`hand.ante`	integer	Per-player ante in chips. For pure Bombpot, this is the bomb deposit. For Short Deck, the backend forces ante to 1 BB.
`hand.dealer_seat`	integer	Button seat, 0-indexed. The UI uses last seat for 3+ players and seat 0 for heads-up.
`hand.sb_seat`	integer	Small blind seat, usually `0`; use `-1` for dead-SB MTT spots.
`hand.bb_seat`	integer	Big blind seat, usually `1`.
`hand.straddle_seat`	integer	Straddle seat, usually `2` for 3-blind cash, otherwise `-1`.
`hand.players[]`	array	One item per table seat: `{ "seat_no": 0, "stack": 10000 }`. Omit `hole_cards` for range-viewer grid requests.
`hand.rake_ratio_pct`	number	Optional cash rake percentage, e.g. `3`. Use `0` for MTT if desired.
`hand.rake_cap_bb`	number	Optional rake cap in BB, e.g. `3`.
`hand.actions.entries[]`	array	Action and board history so far. Empty means first decision.

Action entries use this shape:

{
  "hand": {
    "actions": {
      "entries": [
        { "seat_no": 2, "action": "fold" },
        { "seat_no": 3, "action": "call" },
        { "seat_no": 4, "action": "raise", "amount": 300, "raise_to": 300 },
        { "action": "Ah7d2c" },
        { "action": "Ks" }
      ]
    }
  }
}

Player actions use action: "fold", "call", "check", "raise", "bet", or "allin". amount is raise-by/bet-by in chips. The v2 frontend also sends raise_to, the cumulative raise-to in chips; the backend preserves it and prefers it when present.

Board cards are deal entries: put the card string in action and omit seat_no and amount. Flop uses six characters such as "Ah7d2c"; turn/river use two characters such as "Ks".

Common Holdem-like Output Fields

Squid, Short Deck, Bombpot, and MTT use the v2 Holdem response shape.

Field	Type	Notes
`hands[]`	array	The 169 grid cells. Each item has `handName`, `strategy`, `reach`, `ev`, `actions`, and `combos`.
`hands[].handName`	string	Matrix label such as `"AA"`, `"AKs"`, or `"AKo"`. This replaces v1 top-level `combo`.
`hands[].strategy`	number[]	Strategy vector in action-slot order.
`hands[].reach`	number	Average reach for that grid cell.
`hands[].ev`	number	Aggregated EV for that grid cell.
`hands[].actions[]`	array	Aggregated action rows: `{ "actionName", "probability", "ev" }`.
`hands[].combos[]`	array	Concrete two-card combos in that cell. Each combo still uses `combo`, e.g. `"AsAh"`.
`overallStrategy`	number[]	Overall range strategy in action-slot order.
`overallReach`	number	Average overall reach.
`gridWeightType`	string	`"uniform"` by default, or `"reach"` when requested.
`actionLabels[]`	array	Legal action labels. Each label includes `slot`, `name`, `rawAction`, `amount`, `amountBb`, and `raiseByBb`.
`currentPlayer`	integer	Seat number for the acting player, or `-1` if terminal.
`numBoardCards`	integer	Number of board cards currently dealt.
`pendingDeal`	string	`""`, `"flop"`, `"turn"`, or `"river"` when more board cards are required.
`numCombos`	integer	Number of concrete NLHE combos evaluated.
`tableState`	object	Pot, board, player stacks/commitments, acting context, and equity arrays.
`actionHistory[]`	array	Backend-rendered action history for UI replay.
`stateRender`	string	Debug render of the engine state.
`obsDebug`	object	Debug fields for action mask, action sizes, game tags, raw board, and obs dim.
`model`	string	Selected model name.
`modelMd5`	string	Selected model checksum.
`computeTimeMs`	integer	Total service-side compute time.
`inferenceTimeMs`	integer	Model inference time for v2 Holdem.
`supportScore`	number	Optional support score when provided by the model output.

MTT fills tableState.icmEquityBb and tableState.bountyBb. Squid fills tableState.squidEquityBb.

Squid

Squid is a Holdem-like v2 request with hand.game_mode_code: "squid" and a top-level squid object. Use hand.game_type: "nlhe" for regular Squid and "shortdeck" for Short Deck Squid.

Squid Input Fields

Field	Type	Notes
`hand.game_type`	string	`"nlhe"` or `"shortdeck"`.
`hand.game_mode_code`	string	Must be `"squid"`.
`squid.squid_value_in_bb`	number	Squid value in BB. Legacy alias `val_v` is still normalized, but use this canonical name.
`squid.mode`	string	`"classic"`, `"blood_battle"`, or `"double"`.
`squid.players[]`	array	Per-seat squid counts: `{ "seat_no": 0, "squid_count": 1 }`.
`squid.total_squids`	integer	Total squid markers this round. Classic is commonly `num_players - 1`; blood/double commonly `num_players + 4`.
`squid.remaining_squids`	integer	Markers not yet awarded.
`squid.super_squid_count`	integer	Optional blood/double super-squid count.
`squid.super_squid_remaining`	integer	Optional remaining super squids.
`squid.is_super_squid_hand`	boolean	Optional marker that this hand has a super squid at stake.
`squid.double_thresholds`	integer[]	Optional threshold/multiplier pairs, e.g. `[3, 2, 5, 4]`.

Squid Request Example

{
  "protocol_ver": "v2",
  "strategy_min_threshold": 0.05,
  "hand": {
    "game_type": "nlhe",
    "game_mode_code": "squid",
    "big_blind": 100,
    "small_blind": 50,
    "ante": 0,
    "dealer_seat": 5,
    "sb_seat": 0,
    "bb_seat": 1,
    "straddle_seat": -1,
    "players": [
      { "seat_no": 0, "stack": 10000 },
      { "seat_no": 1, "stack": 10000 },
      { "seat_no": 2, "stack": 10000 },
      { "seat_no": 3, "stack": 10000 },
      { "seat_no": 4, "stack": 10000 },
      { "seat_no": 5, "stack": 10000 }
    ],
    "rake_ratio_pct": 0,
    "rake_cap_bb": 0,
    "actions": { "entries": [] }
  },
  "squid": {
    "squid_value_in_bb": 5,
    "mode": "classic",
    "players": [
      { "seat_no": 0, "squid_count": 0 },
      { "seat_no": 1, "squid_count": 1 },
      { "seat_no": 2, "squid_count": 0 },
      { "seat_no": 3, "squid_count": 2 },
      { "seat_no": 4, "squid_count": 0 },
      { "seat_no": 5, "squid_count": 0 }
    ],
    "total_squids": 5,
    "remaining_squids": 2
  }
}

Squid Output Example

{
  "hands": [
    {
      "handName": "AA",
      "strategy": [0.0, 0.18, 0.0, 0.82, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "reach": 1.0,
      "ev": 14.25,
      "actions": [
        { "actionName": "call", "probability": 0.18, "ev": 11.2 },
        { "actionName": "raise3.0", "probability": 0.82, "ev": 15.1 }
      ],
      "combos": [
        {
          "combo": "AsAh",
          "reach": 1.0,
          "actions": [
            { "actionName": "call", "probability": 0.12, "ev": 10.9 },
            { "actionName": "raise3.0", "probability": 0.88, "ev": 15.4 }
          ]
        }
      ]
    }
  ],
  "overallStrategy": [0.11, 0.31, 0.0, 0.58, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
  "overallReach": 1.0,
  "gridWeightType": "uniform",
  "actionLabels": [
    { "index": 0, "slot": 0, "name": "fold", "rawAction": -1, "amount": null, "amountBb": 0.0, "raiseByBb": 0.0 },
    { "index": 1, "slot": 1, "name": "call 1.0 BB", "rawAction": 0, "amount": 100, "amountBb": 1.0, "raiseByBb": 0.0 },
    { "index": 2, "slot": 3, "name": "raise 3.0 BB", "rawAction": 300, "amount": 300, "amountBb": 3.0, "raiseByBb": 2.0 }
  ],
  "currentPlayer": 2,
  "numBoardCards": 0,
  "pendingDeal": "",
  "tableState": {
    "street": "preflop",
    "pot": 150,
    "chipsPerBb": 100,
    "board": [],
    "actingPlayer": 2,
    "squidEquityBb": [0.0, 5.2, 0.0, 10.1, 0.0, 0.0]
  },
  "model": "example-model.onnx",
  "modelMd5": "example-md5",
  "computeTimeMs": 18,
  "inferenceTimeMs": 7
}

Short Deck

Short Deck is selected by hand.game_type: "shortdeck". Use hand.game_mode_code: "normal" for regular Short Deck. The backend forces ante = big_blind for all Short Deck variants, matching the v1 behavior; send that value explicitly for readability.

Short Deck Input Fields

Field	Type	Notes
`hand.game_type`	string	Must be `"shortdeck"`.
`hand.game_mode_code`	string	`"normal"` for regular Short Deck. Use `"squid"` or `"bomb"` only for combined variants.
`hand.ante`	integer	Backend treats this as 1 BB regardless of request value. Send `big_blind`.
`hand.players[]`	array	Two to six players are typical for Short Deck configs.
`hand.actions.entries[]`	array	Same action/deal entry format as NLHE.

Short Deck Request Example

{
  "protocol_ver": "v2",
  "strategy_min_threshold": 0.05,
  "hand": {
    "game_type": "shortdeck",
    "game_mode_code": "normal",
    "big_blind": 100,
    "small_blind": 50,
    "ante": 100,
    "dealer_seat": 5,
    "sb_seat": 0,
    "bb_seat": 1,
    "straddle_seat": -1,
    "players": [
      { "seat_no": 0, "stack": 10000 },
      { "seat_no": 1, "stack": 10000 },
      { "seat_no": 2, "stack": 10000 },
      { "seat_no": 3, "stack": 10000 },
      { "seat_no": 4, "stack": 10000 },
      { "seat_no": 5, "stack": 10000 }
    ],
    "rake_ratio_pct": 3,
    "rake_cap_bb": 3,
    "actions": { "entries": [] }
  }
}

Short Deck Output Example

{
  "hands": [
    {
      "handName": "AKs",
      "strategy": [0.08, 0.52, 0.0, 0.40, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "reach": 1.0,
      "ev": 5.2,
      "actions": [
        { "actionName": "fold", "probability": 0.08, "ev": 0.0 },
        { "actionName": "call", "probability": 0.52, "ev": 4.7 },
        { "actionName": "raise4.0", "probability": 0.40, "ev": 6.1 }
      ],
      "combos": [
        {
          "combo": "AsKs",
          "reach": 1.0,
          "actions": [
            { "actionName": "fold", "probability": 0.06, "ev": 0.0 },
            { "actionName": "call", "probability": 0.54, "ev": 4.8 },
            { "actionName": "raise4.0", "probability": 0.40, "ev": 6.2 }
          ]
        }
      ]
    }
  ],
  "actionLabels": [
    { "index": 0, "slot": 0, "name": "fold", "rawAction": -1, "amount": null, "amountBb": 0.0, "raiseByBb": 0.0 },
    { "index": 1, "slot": 1, "name": "call 1.0 BB", "rawAction": 0, "amount": 100, "amountBb": 1.0, "raiseByBb": 0.0 },
    { "index": 2, "slot": 3, "name": "raise 4.0 BB", "rawAction": 400, "amount": 400, "amountBb": 4.0, "raiseByBb": 3.0 }
  ],
  "currentPlayer": 0,
  "numBoardCards": 0,
  "pendingDeal": "",
  "tableState": {
    "street": "preflop",
    "pot": 700,
    "chipsPerBb": 100,
    "board": [],
    "actingPlayer": 0,
    "players": [
      { "position": "SB", "stack": 9900, "startStack": 10000, "committed": 100, "folded": false, "allIn": false }
    ]
  },
  "model": "example-model.onnx",
  "computeTimeMs": 15,
  "inferenceTimeMs": 6
}

Bombpot

Bombpot is selected by hand.game_mode_code: "bomb". Use hand.game_type: "nlhe" for regular Bombpot and "shortdeck" for Short Deck Bombpot.

For pure non-shortdeck Bombpot, hand.ante is the bomb deposit in chips. For Short Deck Bombpot, the backend follows Short Deck behavior and forces the effective ante to 1 BB, then uses a 2 BB bomb deposit internally to match v1.

Bombpot Input Fields

Field	Type	Notes
`hand.game_type`	string	`"nlhe"` or `"shortdeck"`.
`hand.game_mode_code`	string	Must be `"bomb"`.
`hand.ante`	integer	Bomb deposit in chips for regular Bombpot.
`hand.actions.entries[]`	array	Include board card deal entries to query a board-specific spot.

Bombpot Request Example

{
  "protocol_ver": "v2",
  "strategy_min_threshold": 0.05,
  "hand": {
    "game_type": "nlhe",
    "game_mode_code": "bomb",
    "big_blind": 100,
    "small_blind": 50,
    "ante": 500,
    "dealer_seat": 5,
    "sb_seat": 0,
    "bb_seat": 1,
    "straddle_seat": -1,
    "players": [
      { "seat_no": 0, "stack": 10000 },
      { "seat_no": 1, "stack": 10000 },
      { "seat_no": 2, "stack": 10000 },
      { "seat_no": 3, "stack": 10000 },
      { "seat_no": 4, "stack": 10000 },
      { "seat_no": 5, "stack": 10000 }
    ],
    "rake_ratio_pct": 3,
    "rake_cap_bb": 3,
    "actions": {
      "entries": [
        { "action": "Ah7d2c" }
      ]
    }
  }
}

Bombpot Output Example

{
  "hands": [
    {
      "handName": "KQo",
      "strategy": [0.34, 0.46, 0.0, 0.20, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "reach": 1.0,
      "ev": 2.15,
      "actions": [
        { "actionName": "fold", "probability": 0.34, "ev": 0.0 },
        { "actionName": "call", "probability": 0.46, "ev": 1.7 },
        { "actionName": "bet5.0", "probability": 0.20, "ev": 3.9 }
      ],
      "combos": [
        {
          "combo": "KhQc",
          "reach": 1.0,
          "actions": [
            { "actionName": "fold", "probability": 0.31, "ev": 0.0 },
            { "actionName": "call", "probability": 0.49, "ev": 1.8 },
            { "actionName": "bet5.0", "probability": 0.20, "ev": 4.0 }
          ]
        }
      ]
    }
  ],
  "currentPlayer": 0,
  "numBoardCards": 3,
  "pendingDeal": "",
  "tableState": {
    "street": "flop",
    "pot": 3150,
    "chipsPerBb": 100,
    "board": ["Ah", "7d", "2c"],
    "actingPlayer": 0
  },
  "actionHistory": [
    { "isDeal": true, "label": "Flop: Ah 7d 2c" }
  ],
  "modelMd5": "example-md5",
  "computeTimeMs": 16,
  "inferenceTimeMs": 7
}

MTT

MTT is selected by a top-level mtt object with tournament data. Use hand.game_type: "nlhe" and hand.game_mode_code: "normal" for regular MTT. Use hand.game_type: "shortdeck" with the same mtt object for Short Deck MTT.

MTT Input Fields

Field	Type	Notes
`hand.game_type`	string	`"nlhe"` or `"shortdeck"`.
`hand.game_mode_code`	string	Usually `"normal"`.
`mtt.total_entries`	integer	Total tournament entries, including rebuys.
`mtt.total_alive_players`	integer	Players still alive. Defaults to `total_entries` if omitted.
`mtt.prize_pool`	number	Total prize pool in dollars.
`mtt.initial_chips`	number	Starting stack in chips.
`mtt.payout_structure_limit_top10[]`	array	Top payout entries: `{ "position": 1, "prize_value": 2500 }`. Legacy alias `payout_structure` is accepted and truncated to top 10.
`mtt.current_table_players_mttinfo[]`	array	Per-seat tournament info: `seat_no`, `stack`, `rank`, optional `head_bounty`. Legacy alias `players` is accepted.
`mtt.bounty_structure`	object	Required for bounty formats; use `{ "bounty_type": "normal" }` for no bounty.
`mtt.prize_fee`	number	Optional prize component of buy-in.
`mtt.bounty_fee`	number	Optional bounty component of buy-in.
`mtt.registration_fee`	number	Optional total buy-in.
`mtt.current_level`	integer	Optional current tournament level.
`mtt.current_blind_level`	integer	Optional blind level id.
`mtt.blind_structure[]`	array	Optional levels with `level`, `ante`, `small_blind`, and `big_blind`.

bounty_structure.bounty_type can be "normal", "progressive_bounty_pko", "normal_knockout_ko", "mystery_bounty", or "spin".

MTT Request Example

{
  "protocol_ver": "v2",
  "strategy_min_threshold": 0.05,
  "hand": {
    "game_type": "nlhe",
    "game_mode_code": "normal",
    "big_blind": 1000,
    "small_blind": 500,
    "ante": 100,
    "dealer_seat": 5,
    "sb_seat": 0,
    "bb_seat": 1,
    "straddle_seat": -1,
    "players": [
      { "seat_no": 0, "stack": 42000 },
      { "seat_no": 1, "stack": 38000 },
      { "seat_no": 2, "stack": 61000 },
      { "seat_no": 3, "stack": 27000 },
      { "seat_no": 4, "stack": 51000 },
      { "seat_no": 5, "stack": 33000 }
    ],
    "rake_ratio_pct": 0,
    "rake_cap_bb": 0,
    "actions": {
      "entries": [
        { "seat_no": 2, "action": "raise", "amount": 2400, "raise_to": 2500 },
        { "seat_no": 3, "action": "fold" },
        { "seat_no": 4, "action": "call" }
      ]
    }
  },
  "mtt": {
    "total_entries": 100,
    "total_alive_players": 54,
    "prize_pool": 9000,
    "initial_chips": 30000,
    "payout_structure_limit_top10": [
      { "position": 1, "prize_value": 2500 },
      { "position": 2, "prize_value": 1600 },
      { "position": 3, "prize_value": 1100 },
      { "position": 4, "prize_value": 800 },
      { "position": 5, "prize_value": 600 }
    ],
    "bounty_structure": {
      "bounty_type": "normal",
      "initial_bounty": 0,
      "bounty_proportion": 0
    },
    "current_table_players_mttinfo": [
      { "seat_no": 0, "stack": 42000, "rank": 18 },
      { "seat_no": 1, "stack": 38000, "rank": 24 },
      { "seat_no": 2, "stack": 61000, "rank": 8 },
      { "seat_no": 3, "stack": 27000, "rank": 39 },
      { "seat_no": 4, "stack": 51000, "rank": 12 },
      { "seat_no": 5, "stack": 33000, "rank": 31 }
    ],
    "prize_fee": 90,
    "bounty_fee": 0,
    "registration_fee": 100,
    "current_level": 10,
    "current_blind_level": 10,
    "blind_structure": [
      { "level": 10, "ante": 100, "small_blind": 500, "big_blind": 1000 }
    ]
  }
}

MTT Output Example

{
  "hands": [
    {
      "handName": "AQs",
      "strategy": [0.22, 0.48, 0.0, 0.30, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "reach": 0.77,
      "ev": 6.4,
      "actions": [
        { "actionName": "fold", "probability": 0.22, "ev": 0.0 },
        { "actionName": "call", "probability": 0.48, "ev": 5.8 },
        { "actionName": "raise7.5", "probability": 0.30, "ev": 7.9 }
      ],
      "combos": [
        {
          "combo": "AsQs",
          "reach": 0.8,
          "actions": [
            { "actionName": "fold", "probability": 0.18, "ev": 0.0 },
            { "actionName": "call", "probability": 0.52, "ev": 5.7 },
            { "actionName": "raise7.5", "probability": 0.30, "ev": 8.1 }
          ]
        }
      ]
    }
  ],
  "tableState": {
    "street": "preflop",
    "pot": 9500,
    "chipsPerBb": 1000,
    "board": [],
    "actingPlayer": 5,
    "icmEquityBb": [38.2, 34.7, 55.1, 24.3, 46.5, 30.2],
    "bountyBb": []
  },
  "actionLabels": [
    { "index": 0, "slot": 0, "name": "fold", "rawAction": -1, "amount": null, "amountBb": 0.0, "raiseByBb": 0.0 },
    { "index": 1, "slot": 1, "name": "call 2.5 BB", "rawAction": 0, "amount": 2500, "amountBb": 2.5, "raiseByBb": 0.0 },
    { "index": 2, "slot": 3, "name": "raise 7.5 BB", "rawAction": 7500, "amount": 7500, "amountBb": 7.5, "raiseByBb": 5.0 }
  ],
  "currentPlayer": 5,
  "numBoardCards": 0,
  "pendingDeal": "",
  "model": "example-model.onnx",
  "computeTimeMs": 20,
  "inferenceTimeMs": 9
}

PLO

PLO uses the PLO-specific bucket taxonomy pipeline, not the Holdem hands[] grid. Select it with hand.game_type: "plo4", "plo5", or "plo6".

Do not expect the v2 Holdem hands[] array to contain strategy data for PLO. PLO strategy is returned in ploStrategy, keyed by concrete PLO combo string.

PLO Input Fields

Field	Type	Notes
`hand.game_type`	string	`"plo4"`, `"plo5"`, or `"plo6"`. Controls hole-card count.
`hand.game_mode_code`	string	Usually `"normal"`.
`hand.players[]`	array	Omit `hole_cards`; the grid enumerates/samples PLO combos.
`hand.actions.entries[]`	array	Same action/deal format. For normal PLO, only add board cards after a valid completed preflop sequence.
`postProcess.ploSampling.perGroup`	integer	Current JSON route reads camelCase `postProcess.ploSampling`; default is `100`.
`postProcess.ploSampling.groupParentId`	string	Optional taxonomy parent to sample within.
`postProcess.ploSampling.fullId`	string	Optional taxonomy group id to return in full, paged.
`postProcess.ploSampling.fullOffset`	integer	Optional page offset for `fullId`; page size is capped internally.
`reach_query.target_seat_no`	integer	Optional. Seat whose replayed range reach should be reported. Defaults to current player.
`reach_query.normalize`	string	Optional. `"conditional"` or `"absolute"`.
`reach_query.group_by`	string	Optional. `"bucket"` or `"combo"`; current response builder returns bucketed data.
`reach_query.max_buckets`	integer	Optional cap on returned reach buckets.
`reach_query.max_examples_per_bucket`	integer	Optional cap on example combos per bucket.

The proto schema names post-processing as post_process.plo_sampling, but the current /strategy_grid PLO route reads the UI JSON spelling postProcess.ploSampling. Use the camelCase spelling for PLO sampling on this endpoint.

PLO Output Fields

Field	Type	Notes
`taxonomyOnly`	boolean	Present because the response includes taxonomy data. Check `ploStrategy` to see strategy results.
`ploBucketTree[]`	array	Hierarchical taxonomy buckets.
`ploBucketMap`	object	Concrete combo string to bucket id.
`ploMultiplicity`	object	Combo string to suit-isomorphic multiplicity/weight. Sampling may scale this value.
`ploBucketCounts`	object	True combo count per leaf bucket when sampling is used.
`sampledMode`	boolean	Present when sampling was applied.
`fullTotalCount`	integer	Present when `fullId` full-mode paging is used.
`holeCardCount`	integer	`4`, `5`, or `6`.
`hands[]`	array	Empty for PLO; do not consume this for PLO strategy.
`ploStrategy`	object	Strategy keyed by combo string. Each value has `strategy`, `evs`, `ev`, and optional `reach`.
`ploReach`	object	Combo-level reach map when reach computation runs.
`reachResult`	object	Bucketed reach summary when `reach_query` is sent.
`actionLabels[]`	array	Legal action labels for interpreting the 11 strategy slots.
`tableState`	object	Pot, board, acting player, players, and acting context.
`pendingDeal`	string	`""`, `"flop"`, `"turn"`, or `"river"` when board cards are required.
`currentPlayer`	integer	Acting seat, or `-1` if terminal/pending.
`numBoardCards`	integer	Count of board cards in the request/session.
`model`	string	Selected PLO model.
`modelMd5`	string	Selected model checksum.
`computeTimeMs`	integer	Total service-side compute time.

PLO Request Example

{
  "protocol_ver": "v2",
  "strategy_min_threshold": 0.05,
  "hand": {
    "game_type": "plo4",
    "game_mode_code": "normal",
    "big_blind": 100,
    "small_blind": 50,
    "ante": 0,
    "dealer_seat": 5,
    "sb_seat": 0,
    "bb_seat": 1,
    "straddle_seat": -1,
    "players": [
      { "seat_no": 0, "stack": 10000 },
      { "seat_no": 1, "stack": 10000 },
      { "seat_no": 2, "stack": 10000 },
      { "seat_no": 3, "stack": 10000 },
      { "seat_no": 4, "stack": 10000 },
      { "seat_no": 5, "stack": 10000 }
    ],
    "rake_ratio_pct": 3,
    "rake_cap_bb": 3,
    "actions": { "entries": [] }
  },
  "postProcess": {
    "ploSampling": {
      "perGroup": 100
    }
  },
  "reach_query": {
    "target_seat_no": 2,
    "normalize": "conditional",
    "group_by": "bucket",
    "max_buckets": 10,
    "max_examples_per_bucket": 3
  }
}

PLO Output Example

{
  "taxonomyOnly": true,
  "holeCardCount": 4,
  "sampledMode": true,
  "ploBucketTree": [
    {
      "id": "made_hand",
      "label": "Made hand",
      "children": [
        { "id": "one_pair", "label": "One pair", "children": [] }
      ]
    }
  ],
  "ploBucketMap": {
    "AsAhKdQc": "one_pair"
  },
  "ploMultiplicity": {
    "AsAhKdQc": 124.0
  },
  "ploBucketCounts": {
    "one_pair": 8420
  },
  "hands": [],
  "ploStrategy": {
    "AsAhKdQc": {
      "strategy": [0.0, 0.42, 0.0, 0.58, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "evs": [-1.0, 2.4, 0.0, 3.1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
      "ev": 2.806,
      "reach": 1.0
    }
  },
  "ploReach": {
    "AsAhKdQc": 1.0
  },
  "reachResult": {
    "normalization": "conditional",
    "range_buckets": [
      {
        "bucket": "one_pair",
        "mass": 1.0,
        "examples": [
          { "hand": "AsAhKdQc", "reach_prob": 1.0, "multiplicity": 124.0 }
        ]
      }
    ],
    "covered_mass": 1.0
  },
  "actionLabels": [
    { "index": 0, "slot": 0, "name": "fold", "rawAction": -1, "amount": null, "amountBb": 0.0, "raiseByBb": 0.0 },
    { "index": 1, "slot": 1, "name": "call 1.0 BB", "rawAction": 0, "amount": 100, "amountBb": 1.0, "raiseByBb": 0.0 },
    { "index": 2, "slot": 3, "name": "raise 3.0 BB", "rawAction": 300, "amount": 300, "amountBb": 3.0, "raiseByBb": 2.0 }
  ],
  "tableState": {
    "street": "preflop",
    "pot": 150,
    "chipsPerBb": 100,
    "board": [],
    "actingPlayer": 2
  },
  "currentPlayer": 2,
  "numBoardCards": 0,
  "pendingDeal": "",
  "model": "example-plo-model.onnx",
  "modelMd5": "example-md5",
  "computeTimeMs": 32
}

Protocol v2 strategy_grid Query Formats

On this page

Protocol v2 `strategy_grid` Query Formats