> For the complete documentation index, see [llms.txt](https://delpho.gitbook.io/delpho-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://delpho.gitbook.io/delpho-docs/risk/stress-test-results.md).
# Stress test results
Delpho's quantitative risk advisor, StableWatch, backtested the protocol across the 12 months from April 2025 through 16 April 2026. The dataset covers every meaningful funding-rate stress event Hyperliquid produced in that window, with live position sizing, historical funding and price data, and the treasury reserve rebalanced as the live system would have rebalanced it.
## **Behavior across normal periods**
Across the trailing 12 months, hourly HYPE perp funding was positive in roughly 90% of intervals. In that environment, sUSDV's exchange rate compounded steadily and the treasury reserve grew toward its \~5% of TVL target without external subsidy. The mechanism described in [The yield engine: funding capture](/delpho-docs/how-delpho-works/yield-engine-delta-neutral.md) operated as designed.
## **The 10 October 2025 ADL crash**
**Date:** 10 October 2025. The worst event in the dataset.
A forced liquidation cascade on Hyperliquid drove HYPE perp funding to **-1,650% annualized in a single hour**, while **HYPE itself fell roughly 28%** across the cascade. The perp price detached so far from spot during the forced deleveraging that the funding mechanism overcorrected violently. Duration was short. Magnitude was unprecedented in the venue's history.
## **How the architecture behaved**
The shock was absorbed before it could reach USDV holders.
* **USDV was insulated by design.** Its peg is defended by overcollateralization and the auto-deleverage system, not by HYPE staying flat, so a drop of that magnitude was taken on the collateral side. See [Peg and redemption risk](/delpho-docs/risk/peg-and-redemption-risk.md) and [Liquidation and collateral risk](/delpho-docs/risk/liquidation-and-collateral-risk.md).
* **The funding-capture hedge earns from the funding rate, not HYPE's direction.** It held at constant position size while the treasury reserve absorbed the short-term funding gap. See [Funding rate risk](/delpho-docs/risk/funding-rate-risk.md).
* **The strain surfaced in sUSDV's value**, the layer built to absorb it.
* Conditions normalized within hours.
## **What the backtest establishes (and does not)**
The dataset is finite and the venue has matured since the worst event in it, so the next stress event will not look like the last one. What the backtest establishes is that the architecture has already been tested against the largest single-hour funding move Hyperliquid has produced, and the loss-absorbing layers behaved as designed.
{% hint style="info" %}
See [Risk advisor reports](/delpho-docs/audits-and-security/risk-advisor-reports.md).
{% endhint %}