Transparency

Honest metrics

The /transparency page publishes, unfiltered, the historical performance of the house simulator and the narratives:

• Narrative hit rate: % of short theses that called direction correctly 5 sessions later
• Portfolio drawdowns: worst drop from a peak
• Annualized Sharpe ratio (standard risk-adjusted return measure)
• Best and worst week of the house simulator
• Score distribution vs actual outcomes — whether scores 8-10 actually outperformed 1-3
• Dip detector track record by classification: how often tecnica_sin_dano was truly no-damage, how often ruptura_tesis was confirmed, etc.

Each metric comes with its period, observation count, and methodological caveat. If a metric has <30 observations, we mark it as "small sample".

Why we publish bad results too

Because survivorship bias is the biggest lie in the financial industry. Any investing blog, YouTube channel, or signals service shows their wins with screenshots and hides the failures. Here we publish both.

If one week the house simulator rotates badly and loses 3%, we publish it. If a stock narrative predicts a bounce and the stock keeps falling, we publish it. If the dip detector had 8/10 conviction and the classification was wrong, we publish it.

Why?

1. It protects you — you know the real quality of the tool before making decisions with it
2. It disciplines us — knowing each error gets published forces us not to brag and to iterate the model with humility
3. It differentiates us — 99% of the industry doesn't do this. This page is one of the reasons paying for Pro makes sense: you're paying for information about the quality of the information

This page is probably the most important in the product. Read it before deciding whether to pay for Pro, and revisit periodically to detect if the model is degrading.