Compute event weights
use openquant::sample_weights::get_weights_by_time_decay;
let w = get_weights_by_time_decay(&returns, 0.5);sample_weights
Event-Driven Data and Labeling
Sample weighting utilities for overlapping event structure.
Why This Module Exists
Adjusts training influence to avoid overcounting dense overlapping labels.
Key Public APIs
get_weights_by_returnget_weights_by_time_decay
Core Math
Uniqueness Weight
\[w_i=\sum_t\frac{I_{t,i}}{\sum_j I_{t,j}}\]
Time Decay
\[w_i=(\frac{i}{T})^\delta\]
Code Examples
Implementation Notes
- Pair with sequential bootstrap for robust label sampling.
- Time-decay controls recency bias explicitly.