Lingmirage Quant

Resources

LINGMIRAGE QUANT / RESOURCES

Your strategy skill. From download to first report.

Free community skills, data requirements, and a transparent framework for comparing strategies.

Free open-source skill library

AGIPro Trading Skills

Browse the original publisher’s collection of trading and quantitative research Agent Skills. This is a free third-party resource, not a Lingmirage-authored product. Its MIT license permits reuse subject to retaining the copyright and license notices.

Open free skill library → · License and attribution

Review each skill’s code, permissions, data dependencies and current documentation before use. No library-level return claim is made.

Run a Lingmirage skill

  1. Buy and download the ZIP from your order or account.
  2. Extract it and open SKILL.md. Optionally ask your AI coding assistant to explain the instructions.
  3. Prepare licensed daily data with Date and AdjustedClose columns. Require at least 300 rows; prefer ten years.
  4. Run the included Python script. No third-party Python packages or broker credentials are required.
  5. Inspect the JSON report, equity curve, assumptions and dataset hash.

AI assistant subscriptions and market data are separate. The packages do not place orders.

Run your first backtest

Install Python 3.9+ and open a terminal in the extracted package folder. Save your licensed data as prices.csv with the exact headers Date,AdjustedClose. Dates must be unique, ascending and formatted YYYY-MM-DD; adjusted prices must be positive. At least 300 daily rows are required.

For the trend package:

python3 scripts/quant_backtest.py prices.csv --strategy trend --cost-bps 10 --output report.json

For the momentum or reversion package, replace trend with momentum or reversion. On Windows, your Python command may be python instead of python3.

The JSON report contains performance metrics, the equity curve, assumptions and a dataset fingerprint. It is a research report, not a trade instruction.

Download or setup trouble?

  • No download listed: sign in with the account used for the purchase and check whether the order is confirmed.
  • ZIP is not a program: extract it first, then read SKILL.md.
  • Python not found: check that Python 3.9+ is installed in your working environment.
  • CSV rejected: check column names, date order, duplicate dates, positive adjusted prices and the minimum row count.

Open My Account → Downloads

How returns may be compared

Rank eligible historical results by net CAGR, highest first, only within the same instrument, dataset, period, execution timing, leverage and trading-cost assumptions. Show total return, maximum drawdown, zero-risk-free Sharpe and entries beside the rank. Unverified or incomparable results remain unranked.

The three original products are ordered by historical SPY CAGR after modeled costs: 2017-06-21 to 2026-09-04, after a 200-session warm-up; next-close execution, long/cash, no leverage, zero cash yield and 10 bps each way. These are retrospective backtests, not live or independently out-of-sample results. Same-period gross SPY buy-and-hold returned approximately +264.75%, higher than all three strategies. Synthetic software tests check calculation behavior, not future profitability.

Minimum evidence record

Instrument · data vendor · adjustment method · source hash · baseline and end dates · code version · signal and execution timing · fees and slippage · cash rate · total return · CAGR · drawdown · Sharpe · entries · holdout status · limitations.

Historical returns are hypothetical and do not guarantee future results.

Browse AI quant strategy skills →