“Always use limit orders” is bad advice on a fast CLOB. Sometimes crossing the spread is cheaper than waiting. This post builds an execution-cost model that combines spread, depth, fee curves, fill probability, and opportunity cost—then shows how to validate it with Polymarket’s public APIs.
Explore structured research pillars and internal link paths.
Visit Research SeriesTrading culture loves slogans:
On Polymarket’s CLOB, those slogans will cost you money.
Sometimes the cheapest trade is the one that crosses the spread immediately.
This post explains when, using a model you can reproduce from public data:
Internal links:
On a CLOB:
The naive view:
The correct view:
maker trades price for time. taker trades time for price.
Your goal is not to be “maker.” Your goal is to minimize total expected cost.
Start with two costs:
If you cross the spread, you pay:
cost_taker ≈ half_spread + impact(size) + taker_fee
If you post a limit order, you pay:
cost_maker ≈ miss_prob × adverse_move - fill_prob × (spread_capture + maker_rebate)
The important term is miss probability. In fast markets, miss risk dominates.
Polymarket is fee-free on most markets, but some market types have taker fees enabled and the effective fee rate varies with share price, peaking near 50% probability.
Source: Polymarket “Fees” documentation (fee curve; fee-rate endpoint). https://docs.polymarket.com/trading/fees
Separately, Polymarket documents a maker rebates program on eligible market types where taker fees fund rebates paid to liquidity providers.
Source: Maker Rebates Program documentation. https://docs.polymarket.com/developers/market-makers/maker-rebates-program
Implication:
This pushes you toward a conditional policy:
Here are the scenarios where taker is not just defensible, but optimal.
If you believe a market will reprice quickly and one-sidedly (e.g., a breaking headline), your “limit price” is often just a hope.
If the price gaps through your level:
In this regime, the opportunity cost of waiting is larger than half-spread.
Many Polymarket books have:
If your order size clears the top level anyway, the incremental cost of taking the whole stack now can be small relative to the risk of missing.
If you trade markets where the “signal” is short-lived (crypto minute markets, sports in-play), the correct comparator is not midpoint.
It is:
the expected future price if you wait.
If that expected future price is worse than paying the spread now, you take.
Polymarket documents public APIs for:
At minimum, you can estimate:
spread: GET /spread?token_id=...midpoint: GET /midpoint?token_id=...depth/impact: GET /book?token_id=... and compute snapshot VWAP for your sizefee: GET /fee-rate?token_id=... on fee-enabled markets (or use the documented fee curve)That’s enough to compare policies.
If you want a template, the slippage/VWAP method is laid out here:
Most traders optimize a number they can see (spread) and ignore the number that matters (miss rate).
You can measure miss rate empirically:
T minutes?Then compare:
This is a backtest, not an opinion.
If you want a policy that behaves like a professional executor:
“Happy to miss” is not sarcasm. It’s discipline.
If you are not happy to miss, you are not really making—you’re just delaying the taker decision.
The best wallets aren’t just right; they’re efficient. They:
We surface those behaviors so you can trade like an executor, not a spectator.
Research Series
Follow related research articles or jump to the full pillar library.