Predictive Strength
unravel
unravel
Predictive Strength
Measures the assets' overall share of all conversations about crypto.
Investor attention demonstrably precedes Bitcoin returns and volatility, acting as a predictive factor. Studies show attention metrics (Google Trends, social volume) Granger-cause price movements, with predictive accuracy improvements of 20% over baseline models in out-of-sample tests. This aligns with behavioral finance principles where attention drives retail investor inflows before institutional actors react.
For smaller-cap assets, attention directly impacts liquidity. A 10% increase in conversation share reduces idiosyncratic risk by improving market depth and narrowing bid-ask spreads. This creates a self-reinforcing cycle: rising attention → improved liquidity → reduced volatility → sustained attention.
The Attention Index sources data from social media platforms, forums, and crypto-related websites to track conversation volumes about specific assets. These inputs are aggregated and normalized against historical averages to determine relative attention levels, with maximum score of 1 indicating heightened interest compared to past trends.
unravel
To understand a predictive factors predictive power, we create a simple long/short strategy and simulate its past performance (with daily rebalancing):
The strategy is rebalanced daily, on a continuous basis. There are 0.5% transaction costs applied on each position adjustment.
Get started by validating the historical performance of the strategy with our transparent code snippets.
Copy and paste the code snippets below into your Python environment or download the files below.