How Fast Are Copied Trades Executed on Polymarket?

June 21, 2026

When you copy a trader, you are not getting their fill. You are getting a fill, placed some number of seconds after they acted, at whatever price the market has moved to in the meantime. That gap is copy trade latency, and it quietly determines whether mirroring a profitable wallet actually leaves you profitable. This post explains what causes mirror delay on Polymarket, what realistic copy trading speed looks like, and why the smartest defense is choosing traders whose edge does not live or die on sub-second timing.

Why latency moves your fill price

Prediction-market prices are probabilities between 0 and 1. A contract at $0.40 implies a 40% chance. When a trader you follow buys, their order consumes liquidity and nudges the price up. If your mirror lands two minutes later, you are often buying after that move — paying $0.42 or $0.45 for the same position the source got at $0.40.

On a single trade that sounds trivial. Across hundreds of mirrored fills it compounds into a structural drag. The source's reported edge assumes their entry price; you inherit a worse one on every leg. The faster the source's style, the more the price has drifted by the time your order clears — and the more of their edge you lose to the clock.

Two costs hide inside "delay"

• Adverse price movement — the headline cost. The market moves toward fair value while you wait.
• Missed trades entirely — if a source buys and exits inside a window shorter than your detection-plus-execution time, you may never get a clean mirror at all, or you enter just as they leave.

What actually causes mirror delay

Mirror delay is not one number. It is a stack of independent costs, and most platforms quietly absorb the largest one. Here is the real breakdown, measured on live mirrored orders.

1. Data indexing lag (the big one)

Most copy systems detect a source trade by polling Polymarket's public data API for that wallet's recent trades. The problem: the data API lags the chain. A fill that already happened on-chain can take roughly 10 to 60 seconds to appear in the indexed feed. You cannot copy what you cannot yet see, so this delay is irreducible as long as detection runs through the data API.

2. Poll cadence

Polling every N seconds adds, on average, half of N. A 15-second loop adds about 7.5 seconds on average and up to 15 in the worst case. Tightening the loop to 3 seconds cuts that to roughly 1.5 seconds average — cheap to fix, and worth fixing.

3. Execution overhead

Once a trade is detected, the mirror has to fetch market metadata, size the order against your guardrails, sign it, and post it. Caching metadata removes most of this; signing and posting a fill-or-kill order is around 0.5 to 1.5 seconds and is largely irreducible without co-located infrastructure.

4. Watermark and gap bugs

The nastiest latency is infinite latency — a trade that never gets mirrored. Naive "jump to now" detection logic can drop a burst of trades that follows a quiet period, producing mysterious missing fills and a fat tail of apparent slowness. Robust detection always re-scans a trailing window rather than skipping ahead.

Realistic expectations: poll vs on-chain

Be skeptical of any platform claiming consistent sub-second mirroring. Here is what the numbers actually support:

• Data-API polling, untuned: on the order of 120+ seconds average, with a long tail past 240 seconds when gap bugs are present.
• Data-API polling, tuned (tight loop, cached metadata, windowed detection): roughly 15 to 65 seconds, now dominated almost entirely by the indexing lag you cannot control.
• On-chain detection (subscribing to fill events over a Polygon WebSocket instead of polling the data API): roughly 2 to 4 seconds on public nodes, faster on paid infrastructure.

True sub-second mirroring requires co-located or paid nodes and only matters for fast markets. For politics-horizon trades that play out over days, the difference between 2 seconds and 4 seconds is competitively irrelevant. The honest takeaway: copy trading speed on Polymarket is "seconds, sometimes tens of seconds," not "instant."

The real fix: copy edges that survive latency

You can shave latency, but you cannot win the speed race against a trader's own client. So the durable answer is selection, not pure speed. Some edges are simply uncopyable at retail size and speed, no matter how fast your mirror is.

Favourite-scalping is uncopyable

Consider a wallet that buys heavy favourites at $0.97 to $0.99 in high volume and wins about 99% of the time. The math works for them at scale because their per-trade cost is near zero and absolute dollar gains compound with size. At copy size, the gross upside on a $0.97 buy is about 3% — smaller than the round trip of fees, spread, and the slippage a delayed mirror eats. Even on a 100% win rate, copying these trades is net negative. That is why our pipeline skips source buys above roughly $0.952: the edge does not exist at the price you can actually get.

What does survive latency

• Slower-horizon theses — politics and event markets that resolve over days, where a 30-second entry difference barely registers.
• Meaningful per-trade edge — entries with enough gross upside to absorb spread, fees, and a few seconds of drift.
• Position-building, not scalping — traders who accumulate and hold, not those flipping longshots in minutes.

This is exactly what to look for when you read a wallet on our trader leaderboard and study its profile on the metrics that predict profit. A high win rate built on $0.98 favourites is a latency trap; a steady edge on day-horizon markets is copyable.

How to protect yourself in practice

1. Paper-trade first. Mirror a wallet with virtual money in virtual copy trading and watch the equity curve include real mirror timing before you risk a cent.
2. Use slippage guardrails. In live copy trading, set per-trade slippage limits so a stale mirror skips rather than chases a runaway price.
3. Prefer slower edges. Use trader comparison to favor wallets whose returns do not depend on millisecond entries.

Latency is a risk to manage, not a feature to oversell. No mirror is instant, every entry can fill worse than the source, and past performance does not guarantee future results — copying can lose money even when the source is winning.

Want to see mirror timing for yourself without risk? Create a free account, start a paper wallet on a politics-horizon trader, and watch how real copy trade latency plays out on the equity curve. When you are ready for live mirroring with stop-loss, take-profit, and slippage caps, the same guardrails carry over — read how copy trading works first so you go in clear-eyed.