- Home
- Skills
- Trading & Investing
- Quant Alpha Research - Stage 0 and Falsification
Quant Alpha Research - Stage 0 and Falsification
Turn crypto quant ideas into repo-native research contracts, falsification checks, and explicit evidence-based decisions.
Free
Works with the AI Agents you already use
Quant Alpha Research - Stage 0 and Falsification
Example session with this skill installed
Provider coverage passed. Can we rerun alpha now?
- Read your context and instructions
- Compiled the quant alpha research
Recorded public-release example from recorded_eval_cases.json (recorded-provider-concordance). This is a retained transcript, not a new run or evidence of strategy performance. No commands were run and no files were changed. Complete recorded skill response:
Decision: blocked.
Provider coverage passing is not enough to rerun alpha. Coverage only says the provider surface is populated; it does not prove the data is trustworthy for research reruns.
I would not allow the alpha rerun until current repo artifacts are read and confirm:
- provider concordance/trust passed
- freshness and closed-bar handling are valid
- schema/field semantics match the research loaders
- missingness, duplicates, and timestamp alignment are clean
- the rerun boundary is research-only unless promotion gates also pass
Current status from the prompt
- candidate/lane: unknown
- parent/comparator: unknown
- horizon: unknown
- data trust: coverage passed, concordance unproven
- Stage 0: blocked
- falsification: not run
- promotion: not allowed
- next action: read the current repo artifacts and verify provider concordance before any alpha rerun.
Connects securely to your tools. The creator never sees your data.
About this skill
What it does
- Separates provider coverage from data trust and provider concordance.
- Turns market commentary into testable hypotheses and a narrow Stage 0 plan.
- Guides delay, shuffle, holdout, cost, capacity, and provider-sensitivity checks when relevant.
- Reports the decision, comparator, horizon, evidence, blockers, and next executable step.
Who it is for
Quant researchers using Codex or Claude Code with an existing research repository, its own data access, and current research artifacts.
Requirements
A compatible agent, access to the research repository and relevant data, and Python for the included validation utilities. The skill does not supply market data or credentials.
Limits
This is a research workflow, not an alpha generator or trading bot. Recorded release coverage contains three must-trigger prompts; it does not cover the full trigger matrix. A second independent quant-repository forward test remains unfinished. No profitability or production-readiness claim is made. Research results do not authorize live trading.
Source and attribution
June / junsier9 https://github.com/junsier9/quant-alpha-research-skill Prepared from public commit b13e0852395cb946de69136813ffe18327e90179. The public GitHub source remains Apache-2.0 licensed. Agensi access and delivery are subject to its platform terms; this listing does not claim those terms replace the public source license.
Requires a compatible agent, existing research data loaders and current repository artifacts. Not intended for generic finance or frontend work. Instructions alone do not enforce read-only access or trading restrictions; users and the agent runtime must enforce permissions. No profitability or production-readiness guarantee.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 20 days ago
- Free with an Agensi account
Requires
Permissions
Reads the user-selected research repository and artifacts. Shell execution and file writes are used only for authorized research implementation, validation and reporting. Respect read-only requests. No market data, credentials, broker connection or live trading authorization is supplied.