How to read the board

Every trader on the board carries two returns. GROSS is what the wallet made. COPY is what a copier would have netted after realistic execution cost: 5% slippage and a 1% fee on the way in, and again on the way out. The board leads with COPY.

The cost lands twice because copying is two trades. You buy in behind the trader and you sell out behind them, and you pay slippage and fee on both legs.

Measured across the full roster over 30 days, 56% of these wallets are profitable as traders, but only 24% are profitable to follow. 116 of them are genuine winners who would have lost a copier money.

That is the whole reason visionscan exists. A leaderboard that advertises the leader's return is showing a number you cannot have, and the wallets where the two numbers disagree most are exactly the ones that look best on every other board.

The three sort buttons above the roster change the order, not the maths.

COPY

The follower return at the 5% slippage tier: what copying would have netted after costs. This is the default sort and the number in the row.

GROSS

The wallet's own return with no slippage and no fee. Shown so the distance from COPY is visible, not because it is a number you could have had.

LIVE

Most recently active first. This orders by time, not by performance. A wallet at the top of LIVE has traded recently and nothing more.

Scores are computed separately at 24H, 7D and 30D. The windows are never blended, and every figure on the terminal names the window it was measured over. Switching the window re-reads a different score, it does not rescale the one on screen.

A position counts only once it is fully closed. Open bags never inflate a return, so a trader still holding a position that is up 10x gets nothing for it until they sell, and a trader who never sells a loser never books it either.

Below 5 closed positions in a window there is no return at all. It renders as unavailable rather than as a number, because a percentage off four trades is not a measurement.

The label under the window buttons is a live filter, not a note. The roster is restricted to wallets with at least 30 closed positions inside the selected window.

Without it the top of the board is small samples. A large percentage on a dozen closed trades sorts above a trader with a real record, and it does not survive the next dozen. Turning Vision Picks on replaces this filter with the full gate set, which already demands 30 closed positions of its own.

Vision Picks is the only quality signal on the terminal and it is deliberately hard to earn. A wallet has to clear all nine gates on a complete scan. Over 30 days, 7 of 553 do.

The scan itself is a gate before any of the nine. If reconstruction did not reach the start of the window, the history is partial and the wallet cannot qualify at any score. Partial data reads as unknown here, never as a pass.

30 closed

At least 30 positions fully closed inside the window.

PF 1.20

Profit factor of at least 1.2: gross wins over gross losses.

COPY > 0

The follower return at the 5% slippage tier has to be positive. A trader who only clears the bar before costs does not qualify.

hold 60s

Median hold of at least 60 seconds. A four second arbitrage bot can be profitable and still be impossible to follow.

transfers 20%

At most 20% of positions may start with a token transferred in rather than bought. Tokens that arrive for free are not a trade a copier can repeat.

closed 25%

At least a quarter of the tokens the wallet touched were actually closed. A wallet that touches 800 tokens and cleanly exits 43 is being scored on the fraction it chose to sell.

14 days

Active on at least 14 separate days. Sixty closed positions inside one session is a session, not a record.

5 SOL

At least 5 SOL deployed. Returns on tiny positions in thin pools do not survive a follower's real size.

top win 25%

No single winning position may account for more than 25% of total profit. One lottery ticket is not a method.

Vision Picks is unavailable at 24H and 7D. One gate asks for 14 active days, so no shorter window can satisfy it, and the toggle is disabled there rather than returning an empty list that would read as an absence of good traders.

The gates are not loosened to produce a longer list. 7 of 553 is the finding.

Roster

553 wallets seeded from kolscan, all 553 carrying an X handle. The roster is a seed list, not an endorsement of anyone on it.

Trades

Swaps observed on-chain through Helius. Over one million swaps across all 553 wallets: a 30-day backfill with no failed wallets, growing live as new trades land.

Scoring

copier-sim-v1, run at 24H, 7D and 30D. Every scan in the current pass reached the window start, so no score on the board is computed on truncated history.

Holdings

Read live from chain per request and cached for 60 seconds.

Webhook

Registered against production for all 553 wallets. New swaps are delivered per transaction and the board advances as they are observed; the backfill only covers history from before the webhook went live.

There is no automated copy execution. Following a trader notifies you. It does not place an order, and nothing on this product will place one for you.

There is no custody and no delegated signing. Trades are built unsigned on the server and signed by your own wallet, deliberately, one at a time. Your keys are never held here and no session token can move funds.

There is no order ticket in the app, and a Buy or Sell control is out of scope. The terminal is a measurement tool. Any real trade you make on what you see here is placed from your own wallet, deliberately, by you.