Catalyst Event Impact Analysis

THE TWO HALVES. Events are extracted from news by a language model - the right instrument for reading text - and then classified by RULE-BASED HEURISTICS keyed on event TYPE: unlocks lean bearish and high-volatility, upgrades and ETF decisions lean bullish, and so on.

WHY THE SECOND HALF IS WEAKER THAN THE FIRST. A type-keyed rule assigns the same direction to every instance of a category, and the direction of an event depends on its content, not its label. Three examples that break the mapping:

• AN UNLOCK'S IMPACT DEPENDS ON SIZE, RECIPIENT AND ANTICIPATION. A 1% unlock to a long-term team on a schedule published two years ago is not the same event as a 20% unlock to early investors. Unlock dates are also typically KNOWN IN ADVANCE, and under semi-strong efficiency a known future supply increase should be substantially priced before it occurs - the empirical evidence on unlock effects is mixed for exactly this reason.
• AN UPGRADE CAN BE BEARISH. A delayed, contested, or risky upgrade is a risk event; a smooth one may be priced already.
• A REGULATORY DECISION HAS A DIRECTION ONLY WHEN IT ARRIVES. The category is neutral; the outcome is everything.

SO THE RULE TABLE ENCODES A PRIOR, and priors are legitimate - they are how you reason before evidence. What matters is not to confuse a prior with a reading. In this analysis the table sets the direction: a model reads the article to extract the event and writes the note, but the direction is the category's default, moved only by a few words in the event's description such as a delay, a lawsuit or an approval. So it is a prior about the category, not a judgement of this story.

THE VOLATILITY CLAIM IS THE STRONGER ONE. Whether an event moves price UP or DOWN is a directional forecast and hard. Whether it raises volatility is far more defensible: a scheduled resolution of uncertainty reliably widens the distribution of outcomes around its date, which is exactly what options markets price as an event premium in the term structure. Reading this analysis for WHEN rather than for WHICH WAY plays to what it can actually support.

AND THE PRICING-IN POINT THAT GOVERNS ALL OF IT. An event on a public calendar is known to everyone. The tradeable content is in the surprise relative to expectations, not in the event's occurrence - which is why 'buy the rumour, sell the news' is a description of efficient pricing rather than a paradox.