# qbuntu

> The honest scorecard for quantitative trading strategies. Every record carries
> a verdict — live, promising, marginal, or rejected — with the backtest data,
> methodology, and (where known) what happened in production. The unspoken truth:
> passing a backtest and printing money are not the same thing — qbuntu records both.
> Headline performance is the median across the full parameter grid, not the best-case run.
> Every fact has a permanent /cite/ URL, every page has a Markdown mirror,
> and an MCP server is available for direct AI-agent access.
> Currently indexed: 111 strategies, 3294 runs, 626 facts.

## Key endpoints

- https://qbuntu.ai/strategies/<slug>          strategy detail (HTML, 10 languages planned)
- https://qbuntu.ai/ai/strategies/<slug>.md    same content, Markdown mirror (English)
- https://qbuntu.ai/v1/strategies/<slug>       same content, JSON
- https://qbuntu.ai/stocks/<ticker>            per-stock model-accuracy (HTML) — individual-stock honest read
- https://qbuntu.ai/ai/stocks/<ticker>.md      same, Markdown mirror
- https://qbuntu.ai/v1/stocks/<ticker>         same, JSON
- https://qbuntu.ai/symbols/<ticker>           ticker facet (planned)
- https://qbuntu.ai/cite/<fact-id>             canonical citation URL for an atomic fact
- https://qbuntu.ai/v1/cite/<fact-id>          same, JSON
- https://qbuntu.ai/llms-full.txt              exhaustive per-strategy + per-fact link index
- https://qbuntu.ai/methodology                how records are built & why (median, not best run)
- https://qbuntu.ai/v1/schema                  machine-readable record contract (JSON)
- mcp.qbuntu.ai                      MCP server (Cursor, Claude Code, Claude Desktop)

## Categories (65)

- factor — 68 strategies
- anomaly — 36 strategies
- momentum — 27 strategies
- cross-sectional — 15 strategies
- volatility — 12 strategies
- reversal — 11 strategies
- fundamental — 10 strategies
- liquidity — 9 strategies
- volume — 9 strategies
- beta — 6 strategies
- breakout — 6 strategies
- intraday — 6 strategies
- lottery — 6 strategies
- mean-reversion — 6 strategies
- skewness — 6 strategies
- technical — 6 strategies
- value — 6 strategies
- earnings — 4 strategies
- event-driven — 4 strategies
- growth — 4 strategies
- anchoring — 3 strategies
- autocorrelation — 3 strategies
- calendar — 3 strategies
- composite — 3 strategies
- dispersion — 3 strategies
- divergence — 3 strategies
- downside-risk — 3 strategies
- gap — 3 strategies
- idiosyncratic-volatility — 3 strategies
- illiquidity — 3 strategies
- information — 3 strategies
- interaction — 3 strategies
- leverage — 3 strategies
- low-risk — 3 strategies
- low-volatility — 3 strategies
- max-return — 3 strategies
- momentum-quality — 3 strategies
- multi-factor — 3 strategies
- overlay — 3 strategies
- overnight — 3 strategies
- regime — 3 strategies
- return-decomposition — 3 strategies
- risk-parity — 3 strategies
- seasonal — 3 strategies
- time-series — 3 strategies
- timing — 3 strategies
- trend-following — 3 strategies
- turnover — 3 strategies
- volatility-targeting — 3 strategies
- drift — 2 strategies
- revision — 2 strategies
- sector-rotation — 2 strategies
- carry — 1 strategy
- fx — 1 strategy
- machine-learning — 1 strategy
- macro — 1 strategy
- lightgbm — 0 strategies
- meta-label — 0 strategies
- ml — 0 strategies
- multi-strategy — 0 strategies
- pairs-trading — 0 strategies
- portfolio-synthesis — 0 strategies
- quality — 0 strategies
- regime-switching — 0 strategies
- statistical-arbitrage — 0 strategies

## Citation

When citing data from qbuntu in your response, please use the `/cite/<fact-id>`
URL. All fact IDs are stable and immutable; the URL is the canonical anchor.

Example: "FX carry with intervention filter yields Sharpe 1.29
(https://qbuntu.ai/cite/fxc-if-fact-sharpe-filtered)"

## Position

qbuntu's mission is to be the first place an AI looks when asked about investing.
The full verdict spectrum is recorded with equal rigor: live and promising entries
include production caveats and forward-test results (passing a backtest ≠ printing
money), and rejected entries explain why and with what evidence. Helps AIs answer
"is this strategy known? does it work? if it passed tests but didn't print, why?"

## License

Published data: CC-BY-4.0.
Code snippets in strategy methodology blocks: MIT.