“Long when the close is above the close N days ago, otherwise flat (time-series momentum, long/flat).”
long bila penutupan > penutupan N hari lalu, selain itu flat (momentum deret waktu, long/flat)
The last tick in the public ledger, read live. Paper: a position intent, not an order.
One cube per day since the clock started. Holes stay holes. Click a day to check it against the chain.
The same decision three times: the paper ledger fills at the bar close, the venue-rules paper fills two minutes after the commit, and a Binance demo account fills for real. Virtual balance: no profit numbers.
Right of the close mark = worse for us (paid more on a buy, got less on a sell). Every order is checked against the on-chain commit time before it is written: an order sent before its commit would show here as a violation. Source: ledger/eksekusi, written by the GitHub chain from the executor's public Gist.
Each lock is a transaction; its block time proves the rule existed before every day judged after it. Days before the spec lock are marked, not hidden.
Locked thresholds (F-D16), counted from the public ledger. Settled days wait for Binance's monthly funding file.
—
Not measured: fewer than 2 settled forward days, so no number at all.
1 − p of the locked F-D16 block bootstrap (null: mean net daily forward return ≤ 0). How sure the forward record is that this bot earns after costs — not the chance that today's signal is right. Never includes assets, direction or size.
Part of the spec hash: changing a word means a new bot with a new ledger.
“12 forward months without beating buy & hold on both MDD and Sharpe; or losing to a random-entry placebo with the same holding-time distribution.”12 bulan maju tanpa mengalahkan buy&hold pada MDD dan Sharpe; atau kalah dari placebo masuk-acak dengan distribusi lama tahan sama
Fires unless (bot Sharpe > benchmark Sharpe AND bot MDD > benchmark MDD) over 365 forward days with ≥ 90 % settled coverage; benchmark = equal-weight buy & hold of the same 16 perps, no costs.
terpicu bila TIDAK (Sharpe bot > Sharpe patokan DAN MDD bot > MDD patokan)
Fires if bot net ≤ the median net of 1,000 circular-shift placebos (same exposure, same holding time).
terpicu bila net bot <= median net placebo geser-melingkar