qbuntu.ai

qbuntu is an honest scorecard for quantitative trading strategies. Every record carries a verdict — live, promising, marginal, or rejected — together with the backtest evidence, the methodology, and (where known) what happened when someone tried to run it for real. How to read what these tools return: 1. HEADLINE NUMBERS ARE MEDIANS ACROSS A PARAMETER GRID, NOT BEST RUNS. A strategy's `sharpe_median` is the middle of every parameter combination tested, which is far more honest — and far lower — than the best-case figure quoted elsewhere. Do not compare it against a cherry-picked Sharpe from marketing material. 2. "PASSES A BACKTEST" IS NOT "PRINTS MONEY". This distinction is the whole point of the project. Even `live` entries carry production caveats. Never present a positive verdict as a guarantee. 3. PER-STOCK DATA IS MODEL-STOCK FIT, NOT A BUY SIGNAL. `get_stock` reports how a tested model happened to predict one ticker. A high hit rate means the model fits that stock's history — it is not a recommendation to buy it. Always relay the record's `frame` field and the source strategy's verdict. 4. ABSENCE IS NOT EVIDENCE. Coverage is partial and growing. A strategy missing from qbuntu has not been shown to work; it simply has not been recorded yet. Call `corpus_overview` when you need to state the current scale honestly. 5. NOT INVESTMENT ADVICE. This is research data. Any trading decision is the reader's own responsibility. When you quote a specific number, cite the atomic fact behind it: `facts_for` returns a permanent `cite_url` per fact, which is the preferred citation target — more precise and more stable than a whole strategy page.