scientia
← the cortex

VERGE-0

1d · spot · 36 products · coinbase · epoch 1

rejected

failed a gate; retained so the failure stays visible

Cross-sectional momentum over 10 days, rebalanced every 1 day, long the top and short the bottom 2 names by inverse-volatility weight at a 35% volatility target. Gross of costs it returns 18.8% out of sample. Net of the spread, the fee and its own market impact it returns -73.8%. It was never wrong about direction — it was wrong about arithmetic.

equity, in-sample then out

in-sample · fitted out-of-sample · not

OOS sharpe

-3.69

in-sample -2.85 → deflated 0.00

PBO

0.65

above the reject line

OOS return

−73.8%

max drawdown

-76.5%

the gauntlet

01

lookahead

pass

02

survivorship

pass

03

overfit

fail

04

fills

pass

05

regime

fail

06

costs

fail

This configuration held MPL-USD, BADGER-USD — products Coinbase has since delisted. They were traded all the way down and marked out at the last price they printed, not removed from the record.

This agent is retained on the board with its failure intact. It is not deployable and never will be — but deleting it would make the roster look like a process that produces winners, when it is a process that produces almost entirely losers and reports that honestly.

the file

in-sample sharpe-2.85
in-sample return+-93.7%
out-of-sample sharpe-3.69
deflated0.00
trials searched630
capacity$2,000,000
gross of costs+18.8%
survivors-only universe−68.6%
cost drag7737 bps / yr
turnover253.9x

forward window · never used for selection

days90
this agent−16.4%
buy and hold BTC+17.2%

The last 90 days were held back from the search entirely. Whatever this agent did here, it did without having been chosen for it.

the attestation