Under the Hood: The Predictive Attention Engine

If you’ve been following the newsletter, you’ve probably noticed we don’t just look at price charts and moving averages. We track attention.

Why? Because in today’s market, attention precedes capital. Before a stock breaks out, the crowd gathers. Before a token crashes, the retail chatter shifts from high-conviction thesis building to fragmented, random noise. If you can map the attention, you can front-run the liquidity.

But tracking "social volume" isn't enough anymore. Algorithms, bots, and airdrop farmers have completely broken traditional sentiment scanners. If a token's volume spikes by 300% on X (formerly Twitter), you don't know if that's smart money accumulating or a bot farm preparing to dump on you.

To solve this, we built the Predictive Attention Engine. It treats the entire financial ecosystem—both Equities and Crypto—as a single, interconnected matrix. It strips away the bot noise and tells us exactly where the true organic capital is moving.

Here is a quick guide to how the four core models work together to give us our edge.

1. Structural Gravity (The Price Regime)

Before we look at attention, we have to know where the asset currently lives on the price chart. We map every asset into four primary quadrants using Z-scores (standard deviations from the norm):

• Confirmed: Price is up, Sentiment is up. (The golden child).

• Early Belief: Sentiment is up, but Price is still down. (The stealth base).

• Skeptical Rally: Price is up, but Sentiment is down. (The "disbelief" rally).

• Capitulation / Dead Zone: Price is down, Sentiment is down. (The wasteland).

2. Attention Velocity (The Hot Trend)

Price tells us where an asset is. Velocity tells us where it's going. We measure the speed and direction of the attention by comparing the last 7 days of engagement to the prior 7 days.

• ↗ Incoming: Attention is accelerating. The crowd is arriving.

• ↘ Outgoing: Attention is decelerating. The crowd is leaving.

When you combine Regime and Velocity, you get the structure. If an asset is in a "Confirmed" regime but prints a massive "↘ Outgoing" velocity score, that is a Crash Canary. It means the price is high, but the crowd just left the building. The floor is about to fall out.

3. The Conviction Matrix (The Footprint of Capital)

This is where we filter out the bots. We don't just look at how much people are talking; we look at the footprint of the conversation by measuring Breadth (unique participants) vs. Quality (engagement per post). This gives us four types of conviction:

• High-Conviction Organic: Breadth is expanding and engagement is high. This is smart money and true organic breakouts. (This is what we want to buy).

• Inorganic Echo Chamber: Engagement is massive, but unique participants are actually shrinking. This means a tiny group of accounts or bots is yelling loudly to create artificial volume. (This is a trap).

• Broad Retail Noise: Massive reach, but terrible engagement. Mainstream news is talking about it, but no one is actually deploying capital. (Dead air).

• Total Contraction: The crowd has completely abandoned the asset.

4. Semantic Coherence (The NLP Engine)

Finally, we use Natural Language Processing (NLP) to actually read the room. We track the "entropy" (randomness) of the narrative.

• Consolidation: The narrative is tight, focused, and cohesive. The crowd is aligned on a fundamental thesis.

• Decay: The narrative is fracturing. The crowd is confused, talking about random things, and losing the plot.

This model also provides our Ticker Collision Defense. If an asset spikes in volume, the NLP engine checks the actual keywords. If a crypto token spikes but the keywords are all about a traditional finance stock that happens to share the same ticker symbol, the engine automatically flags it and overrides the buy signal.

The Secret Sauce: The Liquidity Cascade

Because we run Crypto and Equities through the exact same engine, we can track cross-asset liquidity. Crypto acts as the ultimate "canary in the coal mine" for global risk appetite. By tracking the time-delay between these markets, we can see if a massive drop in Layer-1 crypto velocity today is going to trigger a breakdown in high-beta Semiconductor stocks 48 hours later.

The TL;DR: We don't guess what the market is feeling. We mathematically map the velocity, conviction, and coherence of the crowd to find the exact moments when algorithmic noise gives way to true, organic capital rotation.