Most traders lose money on Polymarket without being wrong. The culprit is execution: spread, depth, and fee curves that turn the midpoint into a comforting fiction. This guide shows how to estimate execution cost from order book snapshots and how to choose maker vs taker in a way you can reproduce.
Explore structured research pillars and internal link paths.
Visit Research SeriesIf you only remember one thing about trading a Central Limit Order Book (CLOB), make it this:
Mid price is not a tradable price. It’s a statistic.
On Polymarket, the difference between a great thesis and a profitable trade is often a few basis points of execution. That gap comes from:
This post gives you a reproducible way to quantify those costs from order book snapshots, then shows how to convert that into a practical decision: maker vs taker.
Internal links for later:
Most UI surfaces show:
Midpoint is useful for charting, but it hides the two things that determine your realized entry:
In a deep book, midpoint and VWAP are close. In a thin book, midpoint can be a lie you tell yourself before paying the spread plus impact.
Let:
p_mid = midpoint at time of orderp_vwap = volume-weighted average execution price for your sizep_ref = your reference price (midpoint, or a “fair” model price)A clean definition of execution slippage (for buys) is:
slippage_bps = 10,000 × (p_vwap - p_ref) / p_ref
For sells, flip the sign:
slippage_bps = 10,000 × (p_ref - p_vwap) / p_ref
Then add fees when they apply:
total_cost_bps ≈ slippage_bps + fee_bps(p, market_type)
Polymarket exposes public APIs for order books and pricing, and public APIs for market metadata (including CLOB token IDs). The docs list the base URLs for Gamma (market metadata) and CLOB (order books) APIs.
Source: Polymarket API endpoints overview (Gamma / CLOB / Data API). https://docs.polymarket.com/quickstart/reference/endpoints
Polymarket’s Gamma Markets API returns market objects with clobTokenIds (YES/NO token IDs) when order book trading is enabled.
Source: Gamma “Get markets” docs (includes CLOB token IDs and enableOrderBook). https://docs.polymarket.com/developers/gamma-markets-api/get-markets
Example (replace slug with any market/event slug you care about):
curl "https://gamma-api.polymarket.com/markets?slug=fed-decision-in-october"
From the response:
enableOrderBook is trueclobTokenIds for the outcome you tradePolymarket’s CLOB API exposes a public GET /book endpoint for a token’s order book (best bids/asks with price levels).
Source: CLOB API endpoints reference (includes /book, /midpoint, /spread). https://docs.polymarket.com/quickstart/reference/endpoints
Example:
curl "https://clob.polymarket.com/book?token_id=YOUR_TOKEN_ID"
For a buy:
Here’s a minimal Python script (standard library only) that:
/bookimport json
import urllib.request
TOKEN_ID = "YOUR_TOKEN_ID"
NOTIONAL_USD = 5000 # change to your size
def get_json(url: str):
with urllib.request.urlopen(url) as r:
return json.loads(r.read().decode("utf-8"))
book = get_json(f"https://clob.polymarket.com/book?token_id={TOKEN_ID}")
mid = get_json(f"https://clob.polymarket.com/midpoint?token_id={TOKEN_ID}")
asks = book["asks"] # [{ "price": "...", "size": "..." }, ...]
p_mid = float(mid["midpoint"])
remaining = NOTIONAL_USD
filled_shares = 0.0
paid = 0.0
for level in asks:
p = float(level["price"])
s = float(level["size"])
level_notional = p * s
take = min(remaining, level_notional)
take_shares = take / p
paid += take
filled_shares += take_shares
remaining -= take
if remaining <= 1e-9:
break
if remaining > 0:
raise SystemExit("Not enough depth to fill this size from the snapshot.")
p_vwap = paid / filled_shares
slippage_bps = 10000 * (p_vwap - p_mid) / p_mid
print({"p_mid": p_mid, "p_vwap": p_vwap, "slippage_bps": slippage_bps, "shares": filled_shares})
What makes this verifiable:
For binary markets, depth often differs between YES and NO. Two common causes:
So your execution model should be side-specific and outcome-specific:
That last point matters even more once taker fees are enabled, because fees peak around 50% probability for certain market types.
Source: Polymarket “Fees” documentation (fee curve; markets with fees; fee-rate endpoint). https://docs.polymarket.com/trading/fees
Think in expected value, not ideology.
You cross the spread, accept price impact, and (sometimes) pay the taker fee.
Taker is optimal when:
You post a limit order and try to earn the spread (or at least avoid paying it).
Maker is optimal when:
Here is a simple, usable model:
expected_cost_taker = spread_bps + impact_bps(size) + fee_bps
expected_cost_maker = miss_prob × move_bps - fill_prob × rebate_bps
You don’t need to estimate everything perfectly. You need the sign.
If you want to formalize this in a way that survives overfitting, build it into a backtest:
This is exactly the kind of tooling we keep in the product stack:
When we analyze “smart money” wallets, the best ones are not just right—they are cheap.
They consistently:
That’s why we treat execution as alpha, not implementation detail:
Use this pre-trade checklist:
If you want alerts that surface only the trades worth copying—and filter out noise—upgrade for real-time delivery:
Research Series
Follow related research articles or jump to the full pillar library.