Our strength lies not in a single algorithm, but in a complex, multi-layered system of capital allocation and risk management.
Hierarchical Portfolio Construction
We employ a three-tier capital allocation model:
Tier 1: Allocation of capital among individual trading algorithms.
Tier 2: Allocation among pools of strategies based on the same mathematical model.
Tier 3: Allocation among different tradable instruments (e.g., BTC futures, ETH futures).
Capital is dynamically reallocated towards more efficient strategies, with rebalancing frequency ranging from weekly to quarterly.
The Power of Ensembling
Our final trading portfolio is an "ensemble" of over 1000 individual, weak yet practically uncorrelated (β1%) algorithms. This diversification allows for a high overall Sharpe ratio (risk-adjusted return) even if each individual algorithm yields a modest result. This approach is analogous to ensemble learning methods in machine learning.
Capital is automatically redistributed in favor of strategies showing better performance, with a frequency ranging from one week to three months.