Source of truth
The canonical source is the verified Bybit closed-PnL ledger — the exchange’s own record of what closed and for how much — used whenever it is available. The public tape or a generated snapshot is used only as a degraded fallback when the ledger cannot be read, and that state is labeled rather than hidden.
The v2 reset policy
On 14 June 2026 (00:00 UTC) Glimpse shipped a new version and reset the public record to a clean $1,000 account. Trades closed before that instant are excluded, so the record starts fresh and retired v1 history never leaks into the v2 curve. The start date is fixed in code, not recomputed, so the curve cannot silently reset again.
Account model — $1,000, 100% equity, no leverage
The published curve models a $1,000 account that commits 100% of current equity to each trade and compounds between trades, with no leverage: a 1% move on the position is a 1% move on the account, never amplified. It is a deliberately conservative, spot-style view of the underlying edge — the bot may trade leveraged on the exchange, but the public simulation strips leverage out so the curve is honest about edge, not amplification.
Open trades are redacted
A live position may appear on the tape, but its direction, entry, targets, sizing, and reasoning stay hidden until it closes. Only closed outcomes are scored. This prevents the record from being read as a live signal feed and keeps the private logic private.
Fees, funding, and slippage
Fees, funding, and slippage are part of real futures trading and are part of the closed outcomes shown. Partial exits are counted as part of the outcome too: if the desk banks a partial take-profit and the runner later closes red, the row says so rather than pretending the final leg tells the whole story.
Sample size and how to read it
The headline must be read as a package: net result, max drawdown, date range, sample size, and source. Win rate is not the same as expectancy — many small wins can still lose money if the losses are larger. The sample size is the canonical close count, not open positions. See the glossary for the terms.