A trader can make money while a follower loses money copying the same visible activity. Different entry prices, delay, fees, sizing and missed exits can change the result. The first useful question is not “Who has the largest PnL?” but “Which parts of this behavior can I observe and reproduce?”
This guide is a research workflow, not a recommendation to follow a wallet or authorize a bot. Use the tools directory to distinguish analytics platforms from execution services.
Establish what the target's result actually means
Start with the trader PnL checklist. Identify the exact wallet, observation period, realized/open distinction, dominant category and whether one event accounts for most of the result.
A public account is not necessarily the trader's whole portfolio. Positions in other wallets or venues may change the economics of a visible trade. Do not infer a person's identity or complete strategy from a single address.
The Polymarket interview with Stand's co-founder discusses this partial-view problem alongside accumulation, merges and high-frequency strategies. Those are practitioner observations, not an independently reproduced estimate of any platform's future returns.
Separate the trigger from the fill
For each candidate trade, log when the target executed, when your data source exposed it and when your proposed order could have arrived. Then inspect the available price and size at that later time.
An alert timestamp is not a trade timestamp. A displayed quote is not a guaranteed fill. A backtest that assigns the target's earlier execution price to a later follower is testing a different process from the one you can operate.
For example, a target could buy at 40 cents while the first visible follower ask is 44 cents. Buying 100 contracts would then cost $44 rather than $40 before fees. The four-dollar difference in this hypothetical example reduces the follower's upside and changes its break-even point.
Use the payout calculator to inspect that arithmetic. Use the order-book guide to understand the available size rather than assuming the full order fills at the best displayed ask.
Inspect the exit path as carefully as the entry
A target can reduce exposure through actions that a simple buy/sell feed does not fully represent. On-chain merges, redemptions and transfers may matter. An executor that recognizes only ordinary sales can miss an economically important exit.
Check what happens when the target sells only part of a position, enters another related outcome, becomes inactive, or changes wallets. Also distinguish the service's pause action from closing open positions. The PolyCop guide gives a concrete example of documented pause and limit behavior.
Model sizing after all the controls
Write down the amount the follower would actually submit after percentage sizing, fixed amounts, minimums, caps and existing exposure. A minimum-order rule can turn a tiny target action into a significant fraction of a smaller follower account.
Check concentration across related markets as well as per-market caps. Five market identifiers may still depend on the same underlying event. A category filter can constrain the topic, but it does not guarantee independent positions or a better price.
Add service fees and venue costs to the spread and slippage analysis. Do not evaluate the economics with a target's historical fee schedule if the follower operates under different conditions.
Use a paper log with explicit missing outcomes
Before granting execution authority, record a sequence of observed triggers and realistic proposed actions. Include skipped trades, unavailable quotes, partial fills, rejected orders, delays and open positions. Do not label an unknown fill as a successful one.
Keep one row per trigger with source time, observed time, proposed size, price limit, available depth, estimated costs and eventual status. At review time, retain the unsuccessful cases. Dropping them can make an unusable copying process look profitable.
The paper-trading guide explains how to separate a useful simulation from a claim of live performance.
Define a rejection condition
Decide in advance what would disqualify the process: unavailable accounting, unobservable exits, insufficient liquidity at your observation time, unexplained permissions or a strategy whose apparent edge disappears after costs.
A good outcome can be deciding to keep watching instead of copying. If you proceed further, the service must fit your location and account permissions, and its results must be measured from your actual fills. A green target chart is not a substitute for that evidence.