Dip detector

When it fires

The dip detector reviews every watchlist item daily and triggers when it detects:

• 1D drop ≥ 4% vs the previous close, or
• 5D drop ≥ 8% vs the close 5 sessions ago

These thresholds filter normal market noise and focus attention on significant moves. You'll get an email with the classification as soon as it fires (Pro plan only).

The 3 classifications

Once triggered, the model classifies the event into one of three types:

tecnica_sin_dano (technical, no damage)

The drop is technical: profit taking, sector rotation, fund portfolio rebalancing, derivatives expiry. The fundamental business case remains intact. Model hypothesis: the price will recover within days or weeks if no negative catalysts appear.

noticia_digerible (digestible news)

There's a specific news item or event behind the drop — weak guidance, analyst downgrade, minor regulatory fine — but the model understands that the underlying business is intact and the valuation adjustment is absorbable. Price may take weeks or months to digest the event.

ruptura_tesis (thesis broken)

The fundamental case is broken. Negative structural change: loss of competitive advantage, fraud uncovered, terminal regulatory risk, irreversible moat erosion. The model says this honestly, even when it "hurts", so as not to shield you from information you need. This is the most important classification: if the model says ruptura_tesis, listen and verify with your own sources.

How to read the result

Each event also includes:

• Conviction (1-10) of the model in the classification. <5 means the model isn't clear — treat as a weak signal.
• entry_opportunity (true/false): whether the context suggests a possible entry. Not a recommendation, it's the model's read on whether there's a coherent technical+fundamental setup.
• Suggested entry and stop-loss price in EUR (optional, only when entry_opportunity=true)
• Estimated recovery horizon: days / weeks / months / n/a
• Cited catalysts and news the model used for classifying — verify them

Remember: the model gets it wrong. We publish its track record on /transparency so you know exactly how often each classification has been correct historically. Don't act without verifying your own judgment.