What a copy trading bot actually copies
The input is an observed trade or position change. The bot maps it to a market and outcome, applies your filters, calculates a size, and decides whether to submit your order. That is a new transaction at the price and liquidity available when your request reaches the venue.
A useful distinction is trade copying versus portfolio mirroring. Trade copying reacts to individual actions. Portfolio mirroring tries to maintain exposure similar to a target account. A tool can support one without supporting the other. Ask what happens when you follow an account that already holds positions, and when that account sells only part of a position.
A leaderboard is a starting point for investigation. It does not show every reason behind a trade, activity on another account, or an off-platform hedge. A trader can also choose a size that is reasonable for their account and unsuitable for yours.
Three approaches to compare
| Approach | Useful when | What to verify |
|---|---|---|
| Dedicated copy product | You want a trader feed and copying controls in one interface | Entry and exit handling, price limits, permissions and all fees |
| Your own API service | You need custom matching, sizing and reconciliation | Data source completeness, order state, idempotency and maintenance |
| Research and alert workflow | You want to examine activity before deciding what to do | Data provenance, observation time and separation from order execution |
Polycopy currently advertises manual copying, automated trader and strategy bots, and a paper mode. That is a documented product fit for a copy-specific evaluation. We have not audited its execution quality or reproduced its performance calculations. Confirm the current feature set and fees before using it.
A custom implementation uses the venue's APIs and your own state store. Polymarket's order-management documentation explains how to inspect orders and trades. You still need a reliable source for the activity you want to follow; an order endpoint alone is not a copying engine.
NickAI fits the research and workflow path. Its Polymarket Data node supports market discovery and price reads, with authenticated execution in a separate order node. We are not presenting it as a ready-made wallet-mirroring service. A wallet-copying design would first need a verified activity source, persistent state and an explicit exit policy. See the broader prediction-market tools comparison for where each approach fits.
The controls that determine whether copying is practical
- Maximum entry price. Define how far your order may move from the observed price. Skipping a trade is a valid result when the quote has already moved.
- Position limits. Cap each order, market and combined set of related markets. Several similar questions can concentrate the same underlying exposure.
- Exit handling. Specify partial sales, closed markets and what happens when copying starts after the original entry. Never sell more than your own filled position.
- Duplicate protection. Persist a source trade identifier and the corresponding decision. A restart or retry must not turn one signal into several orders.
- Freshness limits. Record when the source acted, when your service observed it, and when it made a decision. An old fill can be new to your polling job.
- Pause and reconciliation. Stop new entries on inconsistent state. A pause button should have a documented policy for outstanding orders and existing positions.
Why matching a wallet does not match its results
Consider an illustrative trade observed at $0.40. By the time your bot checks the book, the best available ask is $0.44. Buying 100 shares costs $4 more before fees. You have copied the direction, but you have made a different purchase. The example is arithmetic, not a measured result from any tool.
Compare fills at your size, not only the best displayed quote. A shallow book can offer a few shares near the observed price and the rest at higher prices. Polymarket's pricing and order-book documentation distinguishes the available market reads. The Polymarket API guide explains how to move from a market question to the identifier used for pricing.
Account for the venue's current fee rules, the copying service's charges and any costs of your own infrastructure. Do not subtract a single assumed percentage from every historical trade and call it a live execution test.
A test log you can review before enabling orders
Choose a small, defined sample of activity. Record every observed action, including the ones your filters reject. For each one, save the source identifier, market and outcome, original price, observation time, available quote and size, proposed amount, decision and reason. Avoid selecting only examples that were easy to copy.
Use a paper or shadow mode that models the market being tested. A generic equities paper account does not reproduce Polymarket liquidity or settlement. If the product offers a simulator, ask how it treats delay, partial fills and unavailable depth. If those are omitted, use the test to check logic, not to estimate live execution.
Then review three failure cases: the source exits before your entry; your first request times out after reaching the venue; and your service restarts while an order is unresolved. A system that cannot explain its state in those cases is not ready to act unattended.
Build the research stage with Nick
A concrete first workflow is a market monitor. Ask Nick to find a chosen group of Polymarket markets, record their identifiers and current quotes, and prepare a report with source links and timestamps. Keep the workflow inactive for inspection and leave out order nodes during the first test.
Use the Polymarket workflow walkthrough to check the data path. If you prefer to start with a plain-English task, the vibe-trading page carries an editable example into NickAI. Add a copying-specific activity source only after verifying its coverage and output.