We’ve been testing whether our on-chain dataset can be used not only to describe what happened on ICP, but to extract signals with potential forward-looking value.
One of the strongest results so far comes from a proprietary downside-risk model.
We tested it as a simple long-only spot implementation: the strategy remains exposed to ICP under normal conditions and moves out of the asset when the model identifies an elevated downside-risk regime.
No shorting. No leverage.
Historical backtest — Jan 2025 to Aug 2026
Risk-managed ICP: +118%
ICP buy & hold: −76%
Maximum drawdown:
−32% vs −84%
The model also outperformed in 2026:
+44% vs −16%
What matters to us is not simply the backtested return.
The result suggests that the structure of on-chain activity can contain information about subsequent downside risk— and that sufficiently detailed blockchain data can potentially be transformed into systematic risk-management and research signals.
The signal methodology and construction are proprietary.
The historical backtest uses a one-day execution lag, assumes 10 bps per exposure change, and credits no return while out of ICP. It is a research backtest, not live performance.
We provide on-chain data, research and custom analytics for professional and institutional use cases: https://icterminal.com/custom-inquires