prediction-market-agent-tooling 0.64.12.dev660__py3-none-any.whl → 0.65.0__py3-none-any.whl

prediction_market_agent_tooling/markets/agent_market.py

@@ -1,25 +1,33 @@
 import typing as t
 from enum import Enum
+from math import prod
 
 from eth_typing import ChecksumAddress
 from pydantic import BaseModel, field_validator, model_validator
 from pydantic_core.core_schema import FieldValidationInfo
 from web3 import Web3
 
+from prediction_market_agent_tooling.benchmark.utils import get_most_probable_outcome
 from prediction_market_agent_tooling.config import APIKeys
+from prediction_market_agent_tooling.deploy.constants import (
+    INVALID_OUTCOME_LOWERCASE_IDENTIFIER,
+    NO_OUTCOME_LOWERCASE_IDENTIFIER,
+    YES_OUTCOME_LOWERCASE_IDENTIFIER,
+)
 from prediction_market_agent_tooling.gtypes import (
-    CollateralToken,
     OutcomeStr,
     OutcomeToken,
+    OutcomeWei,
     Probability,
 )
+from prediction_market_agent_tooling.loggers import logger
 from prediction_market_agent_tooling.markets.data_models import (
     USD,
     Bet,
+    CategoricalProbabilisticAnswer,
     CollateralToken,
     ExistingPosition,
     PlacedTrade,
-    ProbabilisticAnswer,
     Resolution,
     ResolvedBet,
 )
@@ -27,13 +35,12 @@ from prediction_market_agent_tooling.markets.market_fees import MarketFees
 from prediction_market_agent_tooling.tools.utils import (
     DatetimeUTC,
     check_not_none,
-    should_not_happen,
     utcnow,
 )
 
 
 class ProcessedMarket(BaseModel):
-    answer: ProbabilisticAnswer
+    answer: CategoricalProbabilisticAnswer
 
 
 class ProcessedTradedMarket(ProcessedMarket):
@@ -66,15 +73,32 @@ class AgentMarket(BaseModel):
     question: str
     description: str | None
     outcomes: t.Sequence[OutcomeStr]
-    outcome_token_pool: dict[str, OutcomeToken] | None
+    outcome_token_pool: dict[OutcomeStr, OutcomeToken] | None
     resolution: Resolution | None
     created_time: DatetimeUTC | None
     close_time: DatetimeUTC | None
-    current_p_yes: Probability
+
+    probabilities: dict[OutcomeStr, Probability]
     url: str
     volume: CollateralToken | None
     fees: MarketFees
 
+    @field_validator("probabilities")
+    def validate_probabilities(
+        cls,
+        probs: dict[OutcomeStr, Probability],
+        info: FieldValidationInfo,
+    ) -> dict[OutcomeStr, Probability]:
+        outcomes: t.Sequence[OutcomeStr] = check_not_none(info.data.get("outcomes"))
+        if set(probs.keys()) != set(outcomes):
+            raise ValueError("Keys of `probabilities` must match `outcomes` exactly.")
+        total = float(sum(probs.values()))
+        if not 0.999 <= total <= 1.001:
+            # We simply log a warning because for some use-cases (e.g. existing positions), the
+            # markets might be already closed hence no reliable outcome token prices exist anymore.
+            logger.warning(f"Probabilities for market {info.data=} do not sum to 1.")
+        return probs
+
     @field_validator("outcome_token_pool")
     def validate_outcome_token_pool(
         cls,
@@ -91,6 +115,14 @@ class AgentMarket(BaseModel):
                 )
         return outcome_token_pool
 
+    def get_outcome_token_pool_by_outcome(self, outcome: OutcomeStr) -> OutcomeToken:
+        if self.outcome_token_pool is None or not self.outcome_token_pool:
+            return OutcomeToken(0)
+
+        # We look up by index to avoid having to deal with case sensitivity issues.
+        outcome_idx = self.get_outcome_index(outcome)
+        return list(self.outcome_token_pool.values())[outcome_idx]
+
     @model_validator(mode="before")
     def handle_legacy_fee(cls, data: dict[str, t.Any]) -> dict[str, t.Any]:
         # Backward compatibility for older `AgentMarket` without `fees`.
@@ -99,33 +131,51 @@ class AgentMarket(BaseModel):
             del data["fee"]
         return data
 
-    @property
-    def current_p_no(self) -> Probability:
-        return Probability(1 - self.current_p_yes)
+    def market_outcome_for_probability_key(
+        self, probability_key: OutcomeStr
+    ) -> OutcomeStr:
+        for market_outcome in self.outcomes:
+            if market_outcome.lower() == probability_key.lower():
+                return market_outcome
+        raise ValueError(
+            f"Could not find probability for probability key {probability_key}"
+        )
+
+    def probability_for_market_outcome(self, market_outcome: OutcomeStr) -> Probability:
+        for k, v in self.probabilities.items():
+            if k.lower() == market_outcome.lower():
+                return v
+        raise ValueError(
+            f"Could not find probability for market outcome {market_outcome}"
+        )
 
     @property
-    def yes_outcome_price(self) -> CollateralToken:
-        """
-        Price at prediction market is equal to the probability of given outcome.
-        Keep as an extra property, in case it wouldn't be true for some prediction market platform.
-        """
-        return CollateralToken(self.current_p_yes)
+    def is_binary(self) -> bool:
+        # 3 outcomes can also be binary if 3rd outcome is invalid (Seer)
+        if len(self.outcomes) not in [2, 3]:
+            return False
 
-    @property
-    def yes_outcome_price_usd(self) -> USD:
-        return self.get_token_in_usd(self.yes_outcome_price)
+        lowercase_outcomes = [outcome.lower() for outcome in self.outcomes]
+
+        has_yes = YES_OUTCOME_LOWERCASE_IDENTIFIER in lowercase_outcomes
+        has_no = NO_OUTCOME_LOWERCASE_IDENTIFIER in lowercase_outcomes
+
+        if len(lowercase_outcomes) == 3:
+            invalid_outcome = lowercase_outcomes[-1]
+            has_invalid = INVALID_OUTCOME_LOWERCASE_IDENTIFIER in invalid_outcome
+            return has_yes and has_no and has_invalid
+
+        return has_yes and has_no
 
     @property
-    def no_outcome_price(self) -> CollateralToken:
-        """
-        Price at prediction market is equal to the probability of given outcome.
-        Keep as an extra property, in case it wouldn't be true for some prediction market platform.
-        """
-        return CollateralToken(self.current_p_no)
+    def p_yes(self) -> Probability:
+        probs_lowercase = {o.lower(): p for o, p in self.probabilities.items()}
+        return check_not_none(probs_lowercase.get(YES_OUTCOME_LOWERCASE_IDENTIFIER))
 
     @property
-    def no_outcome_price_usd(self) -> USD:
-        return self.get_token_in_usd(self.no_outcome_price)
+    def p_no(self) -> Probability:
+        probs_lowercase = {o.lower(): p for o, p in self.probabilities.items()}
+        return check_not_none(probs_lowercase.get(NO_OUTCOME_LOWERCASE_IDENTIFIER))
 
     @property
     def probable_resolution(self) -> Resolution:
@@ -135,16 +185,8 @@ class AgentMarket(BaseModel):
             else:
                 raise ValueError(f"Unknown resolution: {self.resolution}")
         else:
-            return Resolution.YES if self.current_p_yes > 0.5 else Resolution.NO
-
-    @property
-    def boolean_outcome(self) -> bool:
-        if self.resolution:
-            if self.resolution == Resolution.YES:
-                return True
-            elif self.resolution == Resolution.NO:
-                return False
-        should_not_happen(f"Market {self.id} does not have a successful resolution.")
+            outcome = get_most_probable_outcome(self.probabilities)
+            return Resolution(outcome=outcome, invalid=False)
 
     def get_last_trade_p_yes(self) -> Probability | None:
         """
@@ -201,7 +243,7 @@ class AgentMarket(BaseModel):
         raise NotImplementedError("Subclasses must implement this method")
 
     def get_sell_value_of_outcome_token(
-        self, outcome: str, amount: OutcomeToken
+        self, outcome: OutcomeStr, amount: OutcomeToken
     ) -> CollateralToken:
         """
         When you hold OutcomeToken(s), it's easy to calculate how much you get at the end if you win (1 OutcomeToken will equal to 1 Token).
@@ -225,30 +267,77 @@ class AgentMarket(BaseModel):
         """
         raise NotImplementedError("Subclasses must implement this method")
 
-    def liquidate_existing_positions(self, outcome: bool) -> None:
+    def liquidate_existing_positions(self, outcome: OutcomeStr) -> None:
         raise NotImplementedError("Subclasses must implement this method")
 
-    def place_bet(self, outcome: bool, amount: USD) -> str:
+    def place_bet(self, outcome: OutcomeStr, amount: USD) -> str:
         raise NotImplementedError("Subclasses must implement this method")
 
-    def buy_tokens(self, outcome: bool, amount: USD) -> str:
+    def buy_tokens(self, outcome: OutcomeStr, amount: USD) -> str:
         return self.place_bet(outcome=outcome, amount=amount)
 
     def get_buy_token_amount(
-        self, bet_amount: USD | CollateralToken, direction: bool
+        self, bet_amount: USD | CollateralToken, outcome: OutcomeStr
     ) -> OutcomeToken | None:
         raise NotImplementedError("Subclasses must implement this method")
 
-    def sell_tokens(self, outcome: bool, amount: USD | OutcomeToken) -> str:
+    def sell_tokens(self, outcome: OutcomeStr, amount: USD | OutcomeToken) -> str:
         raise NotImplementedError("Subclasses must implement this method")
 
     @staticmethod
-    def get_binary_markets(
+    def compute_fpmm_probabilities(balances: list[OutcomeWei]) -> list[Probability]:
+        """
+        Compute the implied probabilities in a Fixed Product Market Maker.
+
+        Args:
+            balances (List[float]): Balances of outcome tokens.
+
+        Returns:
+            List[float]: Implied probabilities for each outcome.
+        """
+        if all(x.value == 0 for x in balances):
+            return [Probability(0.0)] * len(balances)
+
+        # converting to standard values for prod compatibility.
+        values_balance = [i.value for i in balances]
+        # Compute product of balances excluding each outcome
+        excluded_products = []
+        for i in range(len(values_balance)):
+            other_balances = values_balance[:i] + values_balance[i + 1 :]
+            excluded_products.append(prod(other_balances))
+
+        # Normalize to sum to 1
+        total = sum(excluded_products)
+        if total == 0:
+            return [Probability(0.0)] * len(balances)
+        probabilities = [Probability(p / total) for p in excluded_products]
+
+        return probabilities
+
+    @staticmethod
+    def build_probability_map_from_p_yes(
+        p_yes: Probability,
+    ) -> dict[OutcomeStr, Probability]:
+        return {
+            OutcomeStr(YES_OUTCOME_LOWERCASE_IDENTIFIER): p_yes,
+            OutcomeStr(NO_OUTCOME_LOWERCASE_IDENTIFIER): Probability(1.0 - p_yes),
+        }
+
+    @staticmethod
+    def build_probability_map(
+        outcome_token_amounts: list[OutcomeWei], outcomes: list[OutcomeStr]
+    ) -> dict[OutcomeStr, Probability]:
+        probs = AgentMarket.compute_fpmm_probabilities(outcome_token_amounts)
+        return {outcome: prob for outcome, prob in zip(outcomes, probs)}
+
+    @staticmethod
+    def get_markets(
         limit: int,
         sort_by: SortBy,
         filter_by: FilterBy = FilterBy.OPEN,
         created_after: t.Optional[DatetimeUTC] = None,
         excluded_questions: set[str] | None = None,
+        fetch_categorical_markets: bool = False,
     ) -> t.Sequence["AgentMarket"]:
         raise NotImplementedError("Subclasses must implement this method")
 
@@ -328,12 +417,17 @@ class AgentMarket(BaseModel):
         return self.get_liquidity() > 0
 
     def has_successful_resolution(self) -> bool:
-        return self.resolution in [Resolution.YES, Resolution.NO]
+        return (
+            self.resolution is not None
+            and self.resolution.outcome is not None
+            and not self.resolution.invalid
+        )
 
     def has_unsuccessful_resolution(self) -> bool:
-        return self.resolution in [Resolution.CANCEL, Resolution.MKT]
+        return self.resolution is not None and self.resolution.invalid
 
-    def get_outcome_str_from_bool(self, outcome: bool) -> OutcomeStr:
+    @staticmethod
+    def get_outcome_str_from_bool(outcome: bool) -> OutcomeStr:
         raise NotImplementedError("Subclasses must implement this method")
 
     def get_outcome_str(self, outcome_index: int) -> OutcomeStr:
@@ -344,13 +438,14 @@ class AgentMarket(BaseModel):
                 f"Outcome index `{outcome_index}` out of range for `{self.outcomes}`: `{self.outcomes}`."
             )
 
-    def get_outcome_index(self, outcome: str) -> int:
+    def get_outcome_index(self, outcome: OutcomeStr) -> int:
+        outcomes_lowercase = [o.lower() for o in self.outcomes]
         try:
-            return self.outcomes.index(outcome)
+            return outcomes_lowercase.index(outcome.lower())
         except ValueError:
             raise ValueError(f"Outcome `{outcome}` not found in `{self.outcomes}`.")
 
-    def get_token_balance(self, user_id: str, outcome: str) -> OutcomeToken:
+    def get_token_balance(self, user_id: str, outcome: OutcomeStr) -> OutcomeToken:
         raise NotImplementedError("Subclasses must implement this method")
 
     def get_position(self, user_id: str) -> ExistingPosition | None:
@@ -385,7 +480,7 @@ class AgentMarket(BaseModel):
     def has_token_pool(self) -> bool:
         return self.outcome_token_pool is not None
 
-    def get_pool_tokens(self, outcome: str) -> OutcomeToken:
+    def get_pool_tokens(self, outcome: OutcomeStr) -> OutcomeToken:
         if not self.outcome_token_pool:
             raise ValueError("Outcome token pool is not available.")
 

prediction_market_agent_tooling/markets/blockchain_utils.py

@@ -16,6 +16,9 @@ from prediction_market_agent_tooling.tools.ipfs.ipfs_handler import IPFSHandler
 from prediction_market_agent_tooling.tools.utils import BPS_CONSTANT
 from prediction_market_agent_tooling.tools.web3_utils import ipfscidv0_to_byte32
 
+# max uint16 for easy prediction identification (if market does not have YES outcome)
+UINT16_MAX = 2**16 - 1  # = 65535
+
 
 def store_trades(
     market_id: str,
@@ -28,6 +31,10 @@ def store_trades(
         logger.warning(f"No prediction for market {market_id}, not storing anything.")
         return None
 
+    yes_probability = traded_market.answer.get_yes_probability()
+    if not yes_probability:
+        logger.info("Skipping this since no yes_probability available.")
+        return None
     reasoning = traded_market.answer.reasoning if traded_market.answer.reasoning else ""
 
     ipfs_hash_decoded = HexBytes(HASH_ZERO)
@@ -43,11 +50,13 @@ def store_trades(
         HexBytes(HexStr(i.id)) for i in traded_market.trades if i.id is not None
     ]
 
+    estimated_probability_bps = int(yes_probability * BPS_CONSTANT)
+
     prediction = ContractPrediction(
         publisher=keys.bet_from_address,
         ipfs_hash=ipfs_hash_decoded,
         tx_hashes=tx_hashes,
-        estimated_probability_bps=int(traded_market.answer.p_yes * BPS_CONSTANT),
+        estimated_probability_bps=estimated_probability_bps,
     )
     tx_receipt = OmenAgentResultMappingContract().add_prediction(
         api_keys=keys,

prediction_market_agent_tooling/markets/data_models.py

@@ -3,6 +3,10 @@ from typing import Annotated
 
 from pydantic import BaseModel, BeforeValidator, computed_field
 
+from prediction_market_agent_tooling.deploy.constants import (
+    NO_OUTCOME_LOWERCASE_IDENTIFIER,
+    YES_OUTCOME_LOWERCASE_IDENTIFIER,
+)
 from prediction_market_agent_tooling.gtypes import (
     USD,
     CollateralToken,
@@ -11,24 +15,22 @@ from prediction_market_agent_tooling.gtypes import (
     Probability,
 )
 from prediction_market_agent_tooling.logprobs_parser import FieldLogprobs
-from prediction_market_agent_tooling.tools.utils import DatetimeUTC
+from prediction_market_agent_tooling.tools.utils import DatetimeUTC, check_not_none
 
 
-class Resolution(str, Enum):
-    YES = "YES"
-    NO = "NO"
-    CANCEL = "CANCEL"
-    MKT = "MKT"
+class Resolution(BaseModel):
+    outcome: OutcomeStr | None
+    invalid: bool
 
     @staticmethod
-    def from_bool(value: bool) -> "Resolution":
-        return Resolution.YES if value else Resolution.NO
+    def from_answer(answer: OutcomeStr) -> "Resolution":
+        return Resolution(outcome=answer, invalid=False)
 
 
 class Bet(BaseModel):
     id: str
     amount: CollateralToken
-    outcome: bool
+    outcome: OutcomeStr
     created_time: DatetimeUTC
     market_question: str
     market_id: str
@@ -38,7 +40,7 @@ class Bet(BaseModel):
 
 
 class ResolvedBet(Bet):
-    market_outcome: bool
+    market_outcome: OutcomeStr
     resolved_time: DatetimeUTC
     profit: CollateralToken
 
@@ -84,6 +86,72 @@ class ProbabilisticAnswer(BaseModel):
     def p_no(self) -> Probability:
         return Probability(1 - self.p_yes)
 
+    @property
+    def probable_resolution(self) -> Resolution:
+        return (
+            Resolution(
+                outcome=OutcomeStr(YES_OUTCOME_LOWERCASE_IDENTIFIER), invalid=False
+            )
+            if self.p_yes > 0.5
+            else Resolution(
+                outcome=OutcomeStr(NO_OUTCOME_LOWERCASE_IDENTIFIER), invalid=False
+            )
+        )
+
+
+class CategoricalProbabilisticAnswer(BaseModel):
+    probabilities: dict[OutcomeStr, Probability]
+    confidence: float
+    reasoning: str | None = None
+
+    @property
+    def probable_resolution(self) -> Resolution:
+        most_likely_outcome = max(
+            self.probabilities.items(),
+            key=lambda item: item[1],
+        )[0]
+        return Resolution(outcome=most_likely_outcome, invalid=False)
+
+    def to_probabilistic_answer(self) -> ProbabilisticAnswer:
+        p_yes = check_not_none(self.get_yes_probability())
+        return ProbabilisticAnswer(
+            p_yes=p_yes,
+            confidence=self.confidence,
+        )
+
+    @staticmethod
+    def from_probabilistic_answer(
+        answer: ProbabilisticAnswer,
+    ) -> "CategoricalProbabilisticAnswer":
+        return CategoricalProbabilisticAnswer(
+            probabilities={
+                OutcomeStr(YES_OUTCOME_LOWERCASE_IDENTIFIER): answer.p_yes,
+                OutcomeStr(NO_OUTCOME_LOWERCASE_IDENTIFIER): Probability(
+                    1 - answer.p_yes
+                ),
+            },
+            confidence=answer.confidence,
+            reasoning=answer.reasoning,
+        )
+
+    def probability_for_market_outcome(self, market_outcome: OutcomeStr) -> Probability:
+        for k, v in self.probabilities.items():
+            if k.lower() == market_outcome.lower():
+                return v
+        raise ValueError(
+            f"Could not find probability for market outcome {market_outcome}"
+        )
+
+    def get_yes_probability(self) -> Probability | None:
+        return next(
+            (
+                p
+                for o, p in self.probabilities.items()
+                if o.lower() == YES_OUTCOME_LOWERCASE_IDENTIFIER
+            ),
+            None,
+        )
+
 
 class Position(BaseModel):
     market_id: str
@@ -124,7 +192,7 @@ class TradeType(str, Enum):
 
 class Trade(BaseModel):
     trade_type: TradeType
-    outcome: bool
+    outcome: OutcomeStr
     amount: USD
 
 
@@ -144,13 +212,13 @@ class PlacedTrade(Trade):
 class SimulatedBetDetail(BaseModel):
     strategy: str
     url: str
-    market_p_yes: float
-    agent_p_yes: float
+    probabilities: dict[OutcomeStr, Probability]
+    agent_prob_multi: dict[OutcomeStr, Probability]
     agent_conf: float
     org_bet: CollateralToken
     sim_bet: CollateralToken
-    org_dir: bool
-    sim_dir: bool
+    org_dir: OutcomeStr
+    sim_dir: OutcomeStr
     org_profit: CollateralToken
     sim_profit: CollateralToken
     timestamp: DatetimeUTC
@@ -170,6 +238,4 @@ class SimulatedLifetimeDetail(BaseModel):
     total_simulated_profit: CollateralToken
     roi: float
     simulated_roi: float
-    sharpe_output_original: SharpeOutput
-    sharpe_output_simulation: SharpeOutput
     maximize: float

prediction_market_agent_tooling/markets/manifold/api.py

@@ -3,9 +3,9 @@ import typing as t
 import requests
 import tenacity
 
-from prediction_market_agent_tooling.gtypes import Mana, SecretStr
+from prediction_market_agent_tooling.gtypes import Mana, OutcomeStr, SecretStr
 from prediction_market_agent_tooling.loggers import logger
-from prediction_market_agent_tooling.markets.data_models import ResolvedBet
+from prediction_market_agent_tooling.markets.data_models import Resolution, ResolvedBet
 from prediction_market_agent_tooling.markets.manifold.data_models import (
     FullManifoldMarket,
     ManifoldBet,
@@ -105,14 +105,13 @@ def get_one_manifold_binary_market() -> ManifoldMarket:
     after=lambda x: logger.debug(f"place_bet failed, {x.attempt_number=}."),
 )
 def place_bet(
-    amount: Mana, market_id: str, outcome: bool, manifold_api_key: SecretStr
+    amount: Mana, market_id: str, outcome: OutcomeStr, manifold_api_key: SecretStr
 ) -> ManifoldBet:
-    outcome_str = "YES" if outcome else "NO"
     url = f"{MANIFOLD_API_BASE_URL}/v0/bet"
     params = {
         "amount": float(amount),  # Convert to float to avoid serialization issues.
         "contractId": market_id,
-        "outcome": outcome_str,
+        "outcome": outcome,
     }
 
     headers = {
@@ -204,11 +203,12 @@ def manifold_to_generic_resolved_bet(
     if not market.resolutionTime:
         raise ValueError(f"Market {market.id} has no resolution time.")
 
-    market_outcome = market.get_resolved_boolean_outcome()
+    market_outcome = market.get_resolved_outcome()
+
     return ResolvedBet(
         id=bet.id,
         amount=bet.amount,
-        outcome=bet.get_resolved_boolean_outcome(),
+        outcome=bet.get_resolved_outcome(),
         created_time=bet.createdTime,
         market_question=market.question,
         market_id=market.id,
@@ -224,3 +224,14 @@ def get_market_positions(market_id: str, user_id: str) -> list[ManifoldContractM
     return response_list_to_model(
         requests.get(url, params=params), ManifoldContractMetric
     )
+
+
+def find_resolution_on_manifold(question: str, n: int = 100) -> Resolution | None:
+    # Even with exact-match search, Manifold doesn't return it as the first result, increase `n` if you can't find market that you know exists.
+    manifold_markets = get_manifold_binary_markets(
+        n, term=question, filter_=None, sort=None
+    )
+    for manifold_market in manifold_markets:
+        if manifold_market.question == question:
+            return manifold_market.resolution
+    return None

prediction_market_agent_tooling/markets/manifold/data_models.py

@@ -1,7 +1,7 @@
 import typing as t
 from enum import Enum
 
-from pydantic import BaseModel
+from pydantic import BaseModel, field_validator
 
 from prediction_market_agent_tooling.gtypes import (
     USD,
@@ -12,6 +12,7 @@ from prediction_market_agent_tooling.gtypes import (
     Probability,
 )
 from prediction_market_agent_tooling.markets.data_models import Resolution
+from prediction_market_agent_tooling.markets.manifold.utils import validate_resolution
 from prediction_market_agent_tooling.tools.utils import DatetimeUTC, should_not_happen
 
 MANIFOLD_BASE_URL = "https://manifold.markets"
@@ -92,21 +93,25 @@ class ManifoldMarket(BaseModel):
     def outcomes(self) -> t.Sequence[OutcomeStr]:
         return [OutcomeStr(o) for o in self.pool.model_fields.keys()]
 
-    def get_resolved_boolean_outcome(self) -> bool:
-        if self.resolution == Resolution.YES:
-            return True
-        elif self.resolution == Resolution.NO:
-            return False
+    def get_resolved_outcome(self) -> OutcomeStr:
+        if self.resolution and self.resolution.outcome:
+            return self.resolution.outcome
         else:
-            should_not_happen(f"Unexpected bet outcome string, '{self.resolution}'.")
+            raise ValueError(f"Market is not resolved. Resolution {self.resolution=}")
 
     def is_resolved_non_cancelled(self) -> bool:
         return (
             self.isResolved
             and self.resolutionTime is not None
-            and self.resolution not in [Resolution.CANCEL, Resolution.MKT]
+            and self.resolution is not None
+            and self.resolution.outcome is not None
+            and not self.resolution.invalid
         )
 
+    @field_validator("resolution", mode="before")
+    def validate_resolution(cls, v: t.Any) -> Resolution:
+        return validate_resolution(v)
+
     def __repr__(self) -> str:
         return f"Manifold's market: {self.question}"
 
@@ -197,18 +202,20 @@ class ManifoldBet(BaseModel):
     createdTime: DatetimeUTC
     outcome: Resolution
 
-    def get_resolved_boolean_outcome(self) -> bool:
-        if self.outcome == Resolution.YES:
-            return True
-        elif self.outcome == Resolution.NO:
-            return False
+    @field_validator("outcome", mode="before")
+    def validate_resolution(cls, v: t.Any) -> Resolution:
+        return validate_resolution(v)
+
+    def get_resolved_outcome(self) -> OutcomeStr:
+        if self.outcome.outcome:
+            return self.outcome.outcome
         else:
-            should_not_happen(f"Unexpected bet outcome string, '{self.outcome.value}'.")
+            raise ValueError(f"Bet {self.id} is not resolved. {self.outcome=}")
 
-    def get_profit(self, market_outcome: bool) -> CollateralToken:
+    def get_profit(self, market_outcome: OutcomeStr) -> CollateralToken:
         profit = (
             self.shares - self.amount
-            if self.get_resolved_boolean_outcome() == market_outcome
+            if self.get_resolved_outcome() == market_outcome
             else -self.amount
         )
         return profit