A call is marked a loss when the token never reached a profitable exit and is no longer realistically tradable. In practice one of these applies:

• Gone quiet: no trading activity for more than 2 hours. The token has stalled and you could not have exited into real liquidity.
• Left for dead: the market cap has fallen below $2,000.
• Rug or dead token: flagged as rugged or dead.
• Faded out: the call never reached your take-profit and then died.

In each of these cases there was no real liquidity left to sell into, so the whole stake is gone: the loss is booked as the full -100% of what you invested (a $2 entry books a $2 loss).

A stop-loss exit is booked at its capped result, so a stop that triggers near your entry shows as a small loss or roughly breakeven, not the token's full drawdown.

Calls on tokens whose peak happened before our entry (already run up and dumped) are not counted at all: they are filtered out, so they never score as a win or a loss.

• PnL: profit or loss across the source's calls in the selected window, in SOL and USD.
• W/L: winning versus losing calls.
• W/L Ratio: wins divided by losses.
• Win Rate (2X+): share of calls that reached at least 2x.
• Signals / Day: average number of calls per day.
• ROI % and ROI $: return on the calls.
• Followers: audience size for social sources.

Every number is computed from real, timestamped on-chain outcomes, not from anything a source claims. In effect it is a continuous live backtest: when a source calls a token we lock the entry point and then measure what the token actually did afterwards, its peak market cap, the multiple, and the ROI. Because only tokens that passed our filters are counted, the stats reflect calls on tokens we genuinely stood behind.

First we use the realistic entry a copier could actually have gotten:

• If a source called before our signal, we use our own signal entry, because you could not have entered earlier than the signal.
• If a source called at or after our signal, we use the source's own entry, since that is the point you would have copied them.

From that entry, each trade runs through your take-profit ladder: at every target you hit, that share of the position is sold at that multiple, and any remainder is valued at the token's final multiple (or zero if it died). Your P&L is the total proceeds minus what you put in.

Two figures are kept fully separate so paper and live never mix:

• Realized P&L is real money: only the trades you actually executed live, at your real stake and real fill.
• Paper projection is a what-if: every filtered signal taken at your strategy's own stake. It keeps running in the background even after you go live, so you can always compare. On a live strategy card, use the history icon to flip to it.

The projection always uses your strategy's stake. Set it to $0.25 per signal and the projected numbers scale to $0.25, not some other amount.

The trailing stop is the percentage drop from the highest price reached after entry. It follows the price up as it climbs and sells automatically on the way down, so you lock in profit instead of giving it all back.

Example: a trailing stop of 70% sells once the price has fallen 70% below its peak (so at 30% of the high). If a token peaks at $100K, the stop sits at $30K and sells if it drops there. Every new high moves the stop up with it, but it never moves back down.

So 70% means a 70% drop from the top, not "price still at 70% of the top".

We began tracking on-chain wallets in June 2026, and X (Twitter) accounts and Telegram groups from early July 2026. Every track record only counts calls from the moment we started tracking that source, and the dataset grows every day as new calls come in.

Market caps, multiples and outcomes refresh continuously, so the rankings reflect the latest results rather than a stale snapshot.