Reference desk — hypothetical, no capital, measured

A public paper desk with zero capital behind it: the same discipline engine that scores real-position advisories also scores a fixed reference watchlist (BTC/ETH/SOL/XRP/DOGE/BNB/ADA/AVAX/LTC/NEAR/DOT/TRX/XLM/AAVE/POL) as hypothetical advice — a synthetic 1x long per symbol per 4-hour bucket, entered at the emission mark, graded against the real tape. Every counter is public from unit zero. Descriptive history, not a forecast.

Hypothetical · no capital · measured Reference-desk accrual — watch it fill

58.6% of graded reference units proved correct, Wilson 95% lower bound 55.9%, over 1,349 graded units in the trailing 30 days.

1,479 hypothetical units emitted so far

15 still inside their 4h grading window

1,349 graded units (trailing 30d window)

58.6% graded correct — floored, never rounded up

55.9% the cautious number — Wilson 95% lower bound

$3,677.63net after execution costs — 1,313 units, floored

Net after execution costs prices every graded unit as the desk's synthetic 1x long traded at market (taker in + taker out) on a fixed $1,000 reference notional, at measured Bybit fees (taker 5.5bp/side, fills-TCA verified) plus a 1bp/side slippage allowance — floored to cents, never rounded up. Correct calls can still lose money after costs; that is the point of showing this number.

Hypothetical · no capital · measured P&L money grade — desk units priced after execution costs

Expectancy per unit (descriptive, not a promotion claim): positive — DESCRIPTIVE expectancy of the per-unit net over the trailing 30d window: mean net per unit > 0 AND day-cluster ci95_low > 0. Descriptive history of hypothetical units — NEVER a promotion claim, never a forecast; promotion stays behind the kill-gates.

$3,677.63 cumulative net after costs — 1,313 units, trailing 30d, floored

$2.80 mean net per unit — floored to cents

$1.69 per-unit mean CI95 low — UTC-day clustered

16 independent UTC-day clusters behind the bound

The money grade prices every graded hypothetical unit as the desk's synthetic 1x long traded at market (taker in + taker out) on a fixed $1,000 reference notional at measured Bybit taker fees (5.5bp/side, fills-TCA verified) plus a 1bp/side standing slippage allowance — floored to cents, never rounded up. The per-unit interval is the UTC-day cluster bootstrap — the same honest interval the promotion kill-gates use, because units within a day share one market regime; below 10 day-clusters no bound is served at all. Hypothetical, no capital, measured — descriptive history, never a promotion claim, never a forecast.

Per reference symbol

Symbol	Units emitted	Graded	Correct (floored)	Wilson 95% floor	Net after costs
BTC	109	95	67.4%	57.4%	$470.78
ETH	109	97	61.9%	51.9%	$335.53
SOL	109	96	58.3%	48.3%	$167.80
XRP	96	86	55.8%	45.2%	$277.82
DOGE	96	87	62.1%	51.5%	$318.23
BNB	96	86	62.8%	52.2%	$366.10
ADA	96	92	50.0%	39.9%	$123.06
AVAX	96	90	53.3%	43.1%	$168.26
LTC	96	86	59.3%	48.7%	$277.48
NEAR	96	91	48.4%	38.3%	-$51.26
DOT	96	91	57.1%	46.8%	$139.84
TRX	96	81	79.0%	68.9%	$761.72
XLM	96	89	51.7%	41.4%	$67.45
AAVE	96	92	51.1%	41.0%	$27.49
POL	96	90	62.2%	51.8%	$227.25

A dash means that symbol has no graded units yet — accruing, never estimated.

The grading rule — identical to the position advisories

The rule, verbatim from the measurement artifact: “reference-desk units — no open position; hypothetical advice (synthetic 1x long on the reference watchlist — crypto majors of the governed pool — entered at the emission mark), graded by the identical |move|>=0.3%/4h rule as position advisories; NEVER blended into the position-advisory headline”

In plain English: a unit grades once its 4-hour window closes. Only moves of |move| ≥ 0.3% against the emission mark count — the identical |move|≥0.3%/4h rule as the position advisories, run by the same resolver over the same kline source. Sub-0.3% (sideways) windows are skip-accounted: disclosed, never counted either way. A hold unit is correct when the 4h move was with the trade.

Never blended. This desk has its own journal, its own state file, and its own count. It is never blended into the position-advisory record or any real-capital claim — hypothetical units prove the machinery measures honestly; they are not trading results and not a forecast.

The headline measured record: /accuracy · metric definitions: /methodology · the full scored ledger: /proof. Everything on this page is descriptive history, not a forecast, and not financial advice.

Measurement stamps — graded state generated 2026-08-16 20:50 UTC · newest graded unit 2026-08-16 12:28 UTC · latest emission 2026-08-16 20:28 UTC · the resolver has not graded new units recently (its own stale flag) — disclosed, never hidden