pmfees

Exact trading fees for the prediction market platforms Kalshi, Polymarket US and polymarket.com (international), as a small Python library and a command line tool.

• The platforms' own formulas and rounding rules: Kalshi rounds up so that fee + position cost is a whole centicent ($0.0001), Polymarket US rounds to the cent with banker's rounding, polymarket.com rounds to 5 decimals.
• Exact decimal arithmetic (decimal.Decimal), so $1.75 is $1.75 and not $1.7500000000000002.
• A break-even helper: how far the price has to move before a round trip is in profit after the fee on both orders.
• No dependencies outside the Python standard library. Typed (py.typed).

Rates current as of: Kalshi 2026-11-10 (fee schedule 7.7.26), Polymarket US 2026-11-09, polymarket.com 2026-11-08. Each platform's date and source are in the rates file and in every result.

Other ways to get the same numbers, from the same rates file:

• In a browser: the SwapHunt Kalshi vs Polymarket fee calculator.
• Over HTTP, for agents: GET https://x402.swaphunt.dev/v1/pm/fees, paid per call with x402 in USDC. This package is free; the endpoint exists for agents that cannot install a library.

Install

pip install pmfees

Python 3.10 or newer.

Command line

pmfees kalshi --price 0.50 --contracts 100

Kalshi taker fee: $1.75 (3.50% of trade value at 50¢)
  buy 100 contracts at $0.50 = $50.00 trade value
  base rate: 0.07
  multiplier: 1
  fee schedule version: 7.7.26
  0.07 × 100 × 0.5 × (1 − 0.5) = 1.75, rounded up so that fee + position cost is a whole centicent ($0.0001): $1.75
  rates as of 2026-11-10 · source: https://kalshi.com/docs/kalshi-fee-schedule.pdf

More examples:

# Polymarket US, Tier 1 (under $10,000 of 30-day taker volume, the default)
pmfees polymarket-us --price 0.50 --contracts 100
# -> Polymarket US taker fee: $2.38 (4.76% of trade value at 50¢)   [0.095 × 100 × 0.5 × 0.5 = 2.375]

# polymarket.com, Crypto category
pmfees polymarket --price 0.25 --contracts 100 --category Crypto
# -> polymarket.com taker fee: $1.3125 (5.25% of trade value at 25¢)

# Break-even: buy 100 on Kalshi at 50¢, when is selling in profit?
pmfees kalshi --price 0.50 --contracts 100 --break-even
# -> Break-even: the price must rise 4¢ to 54¢ (tick 1¢); entry fee $1.75 + exit fee $1.7388, profit at that price $0.5112

Options:

pmfees rates prints the loaded rates and their date.

Library

from pmfees import fee, break_even

r = fee("kalshi", 0.50, 100)
r.fee  # Decimal('1.7500')  what you pay, in USD
r.raw_fee  # Decimal('1.7500') before rounding
r.fee_pct_of_trade_value  # Decimal('3.500')
r.breakdown()  # dict of strings: coefficient, rounding, formula, source_url, rates_as_of, ...

fee("polymarket-us", 0.10, 1000, "buy", tier=4).fee  # Decimal('6.26')
fee("polymarket-us", 0.10, 1000, volume_30d=12_500).fee  # Tier 2
fee("polymarket", 0.25, 100, category="Crypto").fee  # Decimal('1.31250')
fee("kalshi", 0.50, 1).fee  # Decimal('0.0175')  centicent rounding, not $0.02
fee("kalshi", 0.50, 100, maker=True).fee  # Decimal('0.0000')  maker M defaults to 0
fee("kalshi", 0.50, 100, maker=True, multiplier=1).fee  # Decimal('0.4375')
fee("kalshi", 0.50, 100, multiplier=0.5).fee  # Decimal('0.8750')

b = break_even("polymarket-us", 0.40, 500, tier=2)
b.exit_price, b.move_cents  # (Decimal('0.45'), Decimal('5.00'))

Signature: fee(platform, price, contracts, side="buy", tier=None, maker=False, *, category=None, multiplier=None, volume_30d=None, rates=None) -> FeeResult. Platform names: kalshi, polymarket-us, polymarket (also polymarket.com). Prices and counts can be str, int, float or Decimal; floats are converted through str, so 0.1 means exactly 0.1.

break_even(...) takes the same arguments plus tick. It opens the position at price, then walks the price in tick steps (up for a buy, down for a sell) and returns the first exit price where proceeds − cost − entry fee − exit fee > 0, or exit_price=None if no price before $1 (or $0) works.

Kalshi: centicent rounding and the multiplier

Kalshi does not round the fee on its own. It rounds fee + position cost (contracts × price) up to a whole centicent ($0.0001), per the fee schedule (version 7.7.26, page 2). Most of the time that looks like rounding the fee up to four decimals, but the fee alone is what you see change:

from pmfees import fee

fee("kalshi", 0.50, 1).fee  # Decimal('0.0175')  0.07 × 1 × 0.5 × 0.5 = 0.0175 exactly; not $0.02
fee("kalshi", 0.33, 7).fee  # Decimal('0.1084')  raw 0.108339; 2.31 + 0.108339 = 2.418339 -> 2.4184

Accounts cleared through a broker (FCM) keep balances in whole cents, so there the fee rounds to the cent (Kalshi docs). pmfees models kalshi.com accounts (direct members).

Each Kalshi series has a fee multiplier M: 1 (standard), 0.5 or 0. Takers default to M = 1, makers to M = 0, because Kalshi charges makers only on series listed with maker fees:

fee("kalshi", 0.25, 100).fee  # Decimal('1.3125')  M = 1
fee("kalshi", 0.25, 100, multiplier=0.5).fee  # Decimal('0.6563')  raw 0.65625, rounded up
fee("kalshi", 0.50, 100, maker=True).fee  # Decimal('0.0000')  maker, M defaults to 0
fee("kalshi", 0.50, 100, maker=True, multiplier=1).fee  # Decimal('0.4375')  0.0175 × 100 × 0.25

pmfees kalshi --price 0.25 --contracts 100 --multiplier 0.5
# -> Kalshi taker fee: $0.6563 (2.63% of trade value at 25¢)

The formulas

The numbers in this table are for reading only. The code takes every rate, tier, multiplier and rounding rule (including Kalshi's fee + position cost basis) from prediction_market_fees.json.

Not modelled: Polymarket US combo fees and maker rebates, polymarket.com maker and taker rebate programs, and per-market exceptions (polymarket.com markets carry their own feeSchedule, Kalshi series their own multiplier). Check the market when it matters.

Sources

The tests lock the formulas to the worked examples and fee tables on these pages, with a link to the source in each test.

Updating the rates

All rates live in src/pmfees/data/prediction_market_fees.json. Each platform has its own as_of (the date its rates were last checked against source_url). SwapHunt checks the official fee pages every week; when a rate changes, the updated file arrives here as a pull request, and a new release follows after review. To use newer rates without waiting for a release:

from pmfees import fee, load_rates

rates = load_rates("prediction_market_fees.json")
fee("kalshi", 0.50, 100, rates=rates)

or set PMFEES_RATES_FILE=/path/to/prediction_market_fees.json, or pass --rates on the command line. A file with a missing field or an unknown rounding mode is rejected with a RatesError that names the field.

Development

pip install -e . pytest ruff mypy
pytest
ruff check . && ruff format --check .
mypy

Releases go to PyPI from GitHub Actions with trusted publishing: bump __version__, add a CHANGELOG entry, and push a v<version> tag. The publish job waits for approval.

Disclaimer

Fee estimates for information only. The platform's own confirmation is what you pay.

License

MIT. See LICENSE.