Models: Strike + fee-multiplier fields use bare Decimal/float, bypass #225 coercion #259

Summary

A handful of money-adjacent fields are typed as bare Decimal or float, skipping the _coerce_decimal invariant set by #225:

• Market.floor_strike: Decimal | None, cap_strike: Decimal | None
• Event.fee_multiplier_override: Decimal | None
• MarketLifecyclePayload.floor_strike: Decimal | None
• Series.fee_multiplier: float, SeriesFeeChange.fee_multiplier: float

Bare Decimal skips _coerce_decimal — Pydantic will accept a JSON-number 1.23 by routing through float first, exactly the precision-loss path the SDK paid an audit to eliminate. fee_multiplier: float is worse: the value is committed to a binary float on parse (0.65 -> 0.6500000000000000222...) which then propagates into any cost computation. None of these fields reject bool (the #225 invariant). Strikes and multipliers are load-bearing in payout math.

Location

• kalshi/models/markets.py:117-118
• kalshi/models/events.py:39
• kalshi/ws/models/market_lifecycle.py:33
• kalshi/models/series.py:30
• kalshi/models/series.py:48

Evidence

# markets.py Market
floor_strike: Decimal | None = None
cap_strike: Decimal | None = None
# events.py Event
fee_multiplier_override: Decimal | None = None
# market_lifecycle.py MarketLifecyclePayload
floor_strike: Decimal | None = None
# series.py Series / SeriesFeeChange
fee_multiplier: float

Recommended fix

Switch floor_strike/cap_strike to DollarDecimal (they are dollar values per spec). Switch fee_multiplier/fee_multiplier_override to Decimal with BeforeValidator(_coerce_decimal) (or introduce a MultiplierDecimal alias) so the parse path matches every other numeric SDK field.

Severity & category

medium / correctness