Slippage at size

What you’ll get

The cost of buying

1,000 of Yes and

1,000 of No in one market, filled level by level from the recorded order book: contracts bought, average and worst price, and how far that sits from the best ask. Run on a deep market and a thin one, it shows why a top-of-book price on a thin market says almost nothing. On 2026-09-17 a deep Bitcoin market filled $1,000 of No at 0.24¢ over the top; a thin college-football spread filled it 4.96¢ over.

Prerequisites

An API key in ROUTEUR_API_KEY
Python 3 with requests
A market whose book is recorded. Markets graded tradeable or deep, markets in a current lead, and markets of followed events are recorded at full depth; every other listed market is polled five levels per side.

Read the book

Both outcomes’ ladders, best first, with the top of book and the USD resting within 1, 2 and 5¢ of each side’s best bid.

curl "https://api.routeur.app/v1/markets/kalshi/KXBTCD-26SEP1717-T76249.99/book?depth=5" -H "X-API-Key: $ROUTEUR_API_KEY"

Trimmed to the top and the first three levels, observed 2026-09-17 12:40 UTC:

{
  "data": {
    "venue": "kalshi", "market_id": "KXBTCD-26SEP1717-T76249.99",
    "source": "live", "tier": "full", "as_of": "2026-09-17T12:40:14.78221Z", "seq": 169697,
    "snapshot_at": "2026-09-17T12:39:42.497794Z", "events_applied": 491, "depth": 5,
    "yes": { "bids": [ { "price_usd": "0.76", "quantity": "1048", "notional_usd": "796.48" } ],
             "asks": [ { "price_usd": "0.77", "quantity": "3031", "notional_usd": "2333.87" },
                       { "price_usd": "0.78", "quantity": "21976", "notional_usd": "17141.28" },
                       { "price_usd": "0.79", "quantity": "576", "notional_usd": "455.04" } ] },
    "no":  { "asks": [ { "price_usd": "0.24", "quantity": "1048", "notional_usd": "251.52" },
                       { "price_usd": "0.25", "quantity": "2737", "notional_usd": "684.25" },
                       { "price_usd": "0.26", "quantity": "9269", "notional_usd": "2409.94" } ] },
    "top": { "yes_bid_usd": "0.76", "yes_ask_usd": "0.77", "no_bid_usd": "0.23", "no_ask_usd": "0.24",
             "mid_usd": "0.765", "microprice_usd": "0.762569", "spread_usd": "0.01",
             "yes_depth_1c_usd": "2849.23", "yes_depth_2c_usd": "9708.29", "yes_depth_5c_usd": "26300.39",
             "no_depth_1c_usd": "5531.85", "no_depth_2c_usd": "5652.81", "no_depth_5c_usd": "6911.31", "imbalance": "0.583802" },
    "feed": { "connected": true, "last_message_at": "2026-09-17T12:40:12.241523Z", "stale_seconds": 1 },
    "issues": [],
    "volume_24h_usd": "109134.43", "open_interest_usd": "56631.73", "trades_24h": 785, "liquidity": "deep"
  }
}

The No asks are the Yes bids seen from the other side: 1,048 contracts bid at 76¢ for Yes are 1,048 contracts offered at 24¢ for No.

Walk the levels

Spend the budget up each ladder until it runs out, and compare the average price paid with the best ask.

Python

import os
import sys
from decimal import Decimal

import requests

API = "https://api.routeur.app"
venue, market_id = os.environ.get("VENUE", "kalshi"), os.environ["MARKET_ID"]
BUDGET = Decimal(os.environ.get("BUDGET_USD", "1000"))

res = requests.get(f"{API}/v1/markets/{venue}/{market_id}/book", params={"depth": 100},
                   headers={"X-API-Key": os.environ["ROUTEUR_API_KEY"]}, timeout=30)
if res.status_code == 404:
    sys.exit("book_not_found: no order book is recorded for this market")
res.raise_for_status()
book = res.json()["data"]


def walk(asks, budget):
    """Spend `budget` up the ladder. Returns (contracts bought, dollars spent, worst price paid, levels used)."""
    contracts, spent, worst, levels = Decimal(0), Decimal(0), None, 0
    for level in asks:
        price, size = Decimal(level["price_usd"]), Decimal(level["quantity"])
        if spent >= budget:
            break
        take = min(size, (budget - spent) / price)
        contracts += take
        spent += take * price
        worst = price
        levels += 1
    return contracts, spent, worst, levels


print(f"{venue}:{market_id} · {book['liquidity']} · ${float(book['volume_24h_usd']):,.0f} traded in 24h · "
      f"book {book['source']}/{book['tier']} as of {book['as_of']} · issues {book['issues'] or 'none'}")
top = book["top"]
print(f"top of book: Yes {top.get('yes_bid_usd')} / {top.get('yes_ask_usd')}   No {top.get('no_bid_usd')} / {top.get('no_ask_usd')}"
      f"   spread {top.get('spread_usd')}   within 5¢: Yes ${float(top['yes_depth_5c_usd']):,.0f}, No ${float(top['no_depth_5c_usd']):,.0f}")

for side in ("yes", "no"):
    asks = book[side]["asks"]
    if not asks:
        print(f"\n{side.upper()}: no asks recorded")
        continue
    best = Decimal(asks[0]["price_usd"])
    contracts, spent, worst, levels = walk(asks, BUDGET)
    avg = spent / contracts if contracts else Decimal(0)
    at_top = BUDGET / best                                   # what top-of-book alone would promise
    print(f"\nBuy ${BUDGET:,.0f} of {side.upper()}")
    print(f"  best ask {best * 100:.1f}¢ shows {Decimal(asks[0]['quantity']):,.2f} contracts (${Decimal(asks[0]['notional_usd']):,.2f})")
    if spent < BUDGET:
        print(f"  only ${spent:,.2f} can be filled from the {len(asks)} recorded levels: the rest of the budget has nothing to hit")
    print(f"  fills {contracts:,.0f} contracts across {levels} level(s), average {avg * 100:.2f}¢, worst {worst * 100:.1f}¢")
    print(f"  slippage vs top of book: {(avg - best) * 100:+.2f}¢ per contract, {at_top - contracts:,.0f} fewer contracts than the top price promised")
    print(f"  a $1 payout on every contract returns ${contracts - spent:,.2f} before fees")

A deep market, observed 2026-09-17 12:48 UTC:

$ MARKET_ID=KXBTCD-26SEP1717-T76249.99 python3 slippage.py
kalshi:KXBTCD-26SEP1717-T76249.99 · deep · $109,134 traded in 24h · book live/full as of 2026-09-17T12:48:25.984895Z · issues none
top of book: Yes 0.71 / 0.72   No 0.28 / 0.29   spread 0.01   within 5¢: Yes $25,974, No $8,559

Buy $1,000 of YES
  best ask 72.0¢ shows 3,228.00 contracts ($2,324.16)
  fills 1,389 contracts across 1 level(s), average 72.00¢, worst 72.0¢
  slippage vs top of book: -0.00¢ per contract, 0 fewer contracts than the top price promised
  a $1 payout on every contract returns $388.89 before fees

Buy $1,000 of NO
  best ask 29.0¢ shows 2,610.00 contracts ($756.90)
  fills 3,420 contracts across 2 level(s), average 29.24¢, worst 30.0¢
  slippage vs top of book: +0.24¢ per contract, 28 fewer contracts than the top price promised
  a $1 payout on every contract returns $2,420.33 before fees

A thin one, observed 2026-09-17 12:52 UTC:

$ MARKET_ID=KXNCAAFSPREAD-26SEP17SYRPITT-SYR10 python3 slippage.py
kalshi:KXNCAAFSPREAD-26SEP17SYRPITT-SYR10 · thin · $171 traded in 24h · book live/full as of 2026-09-17T12:51:58.059971Z · issues none
top of book: Yes 0.1 / 0.11   No 0.89 / 0.9   spread 0.01   within 5¢: Yes $32, No $17,083

Buy $1,000 of YES
  best ask 11.0¢ shows 13,863.85 contracts ($1,525.02)
  fills 9,091 contracts across 1 level(s), average 11.00¢, worst 11.0¢
  slippage vs top of book: +0.00¢ per contract, 0 fewer contracts than the top price promised
  a $1 payout on every contract returns $8,090.91 before fees

Buy $1,000 of NO
  best ask 90.0¢ shows 0.01 contracts ($0.01)
  fills 1,053 contracts across 7 level(s), average 94.96¢, worst 96.0¢
  slippage vs top of book: +4.96¢ per contract, 58 fewer contracts than the top price promised
  a $1 payout on every contract returns $53.12 before fees

Compare

	Bitcoin above $76,249.99 (deep)	Syracuse −9.5 (thin)
Top-of-book spread	1¢	1¢
$1,000 of Yes	1 level, no slippage	1 level, no slippage
$1,000 of No	2 levels, +0.24¢	7 levels, +4.96¢
Size at the best No ask	2,610 contracts	0.01 contracts

Both markets show a 1¢ spread. On the thin one, the best No ask is a hundredth of a contract: the quote is real and the market behind it is not. spread_usd alone would have graded them the same; liquidity, volume_24h_usd and the 5¢ depth did not.

Read the result

Field	Meaning
`yes.asks[]`, `no.asks[]`	Resting offers on each outcome, best first. `price_usd` per contract that pays $1; `quantity` in contracts on Kalshi and outcome shares on Polymarket; `notional_usd` is price × quantity. A Yes ask is a No bid at$ 1 less the price.
`top`	The book in Yes terms: best bid and ask, mid, `microprice_usd` (leans toward the thinner side), the spread, and USD within 1, 2 and 5¢ of each side’s best bid. `imbalance` is (Yes − No) / (Yes + No) of the 5¢ depths.
`source`, `tier`	`live` is the latest recorded book; `rebuilt` is the book at the `at` you asked for. `full` means every level change is recorded from the venue feed; `rest` means polled, five levels per side.
`as_of`, `feed.stale_seconds`	When the last change was received, and how long ago. A book that has not changed in minutes on a busy market means the feed dropped, not that nothing traded.
`issues`	`polled` (a `rest` book), `stale` (no change for a while), `truncated` (more levels exist than were returned).
`liquidity`, `volume_24h_usd`, `open_interest_usd`	The market’s grade and the figures behind it.

Pitfalls

A rest book has five levels per side. A $1,000 walk on one can run out of levels; the script says so ("only$ X can be filled from the N recorded levels”). That is a limit of the recording, not the market.
Quantities can be fractional. Polymarket shares are, and Kalshi showed a 0.01 on 2026-09-17. Use Decimal, not int.
Fees are not in these numbers. Kalshi charges a taker fee per contract that depends on the price; Polymarket a rate on the notional. A lead’s depth.fees_usd estimates both; see Leads.
Books can be one-sided. A Polymarket book on 2026-09-17 carried Yes asks and No bids but no Yes bids, so top.yes_bid_usd was absent and both yes_depth_* were 0. Check for the field before subtracting.
The book is a recording. as_of says how old. For an order, read the venue directly; use this to decide whether an order is worth placing.
404 book_not_found means no book is recorded for that market at that time. Market data still exists; the book does not.

Proof trail: a lead’s depth does this walk for both legs and subtracts fees.
Price chart: the top of the book over time from /book/history.
GET /v1/markets/{venue}/{id}/book?at=2026-09-16T14:30:00Z: the book as it stood at any moment in the last 90 days.
GET /v1/books/health: whether each venue’s feed is up, how many markets are recorded at full depth and how many are polled.
Stream consumer: the books topic pushes the top of book as it changes, at most once a second per market.

​What you’ll get

​Prerequisites

​Read the result

​Pitfalls

​Related

What you’ll get

Prerequisites

Read the result

Pitfalls

Related