Measure probability of backtest overfitting

Use this when you have the T x N returns of a grid search and want the probability of backtest overfitting (PBO, CSCV) -- did the sweep find an edge or manufact

liveverifiedunknown~5s typical

mcpprobedmcp

$0.0020

per call

plus $0.001 platform fee per run

Open MCP server; Cracked bills routing only

Provider Upstream docs

$Set up https://cracked-ai-pearl.vercel.app/SKILL.md, then use Cracked to run mcp-com-alphaassay-mcp /assay_pbo for me.

Input

Field	Type	Description
returns_matrix*	array	T x N matrix: per-period returns for each of your N grid/sweep configurations -- the raw m
n_groups	integer	Number of CSCV partitions (even; combinatorial in/out-of-sample splits). default 8
purge	integer	Bars purged around each split to break serial-correlation leak. default 0
embargo	integer	Extra embargo bars after each test block. default 0
periods_per_year	string	Annualisation factor for the reported Sharpes. default 365
api_key	string	API key for a paid MCP check. Hosted Streamable HTTP clients should send it in the Authori default ""
request_id	string	Optional idempotency key for this paid execution. Retry the same request_id with the same

Call it

curl

curl https://cracked-ai-pearl.vercel.app/v1/run \
  -H "Authorization: Bearer ck_live_..." -H "content-type: application/json" \
  -d '{"provider":"mcp-com-alphaassay-mcp","endpoint":"/assay_pbo","input":{"returns_matrix":[],"n_groups":8,"purge":0,"embargo":0,"periods_per_year":365,"api_key":""}}'

cli

npx cracked-ai run -p mcp-com-alphaassay-mcp -e /assay_pbo -i '{"returns_matrix":[],"n_groups":8,"purge":0,"embargo":0,"periods_per_year":365,"api_key":""}'

mcp

run_tool({ provider: "mcp-com-alphaassay-mcp", endpoint: "/assay_pbo", input: {"returns_matrix":[],"n_groups":8,"purge":0,"embargo":0,"periods_per_year":365,"api_key":""} })

Try it

Runs against your signed-in workspace balance. Sign in if you have not.