> For the complete documentation index, see [llms.txt](https://docs.keystonefi.xyz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.keystonefi.xyz/reference/historical-simulation.md).

# Historical Simulation

*Daily resolution, Jul 2024 – Jun 2026 (24 months). Data as of 2026-07-24.*

> **Backtest.** Phoenix launched Dec 2025, so this runs on historical SOL-perp funding as a proxy for Phoenix funding, using the v1 mode set — normal basis + parked. Phoenix anchors its funding to CEX index feeds, so the proxy uses **Binance funding at face value (×1.0)** — it tracks the same prices Phoenix does, not the old Drift calibration, which ran hot. The \~9% was a Drift-hot artifact; \~5% is the honest CEX-anchored number.

***

## Headline

Over 24 months spanning the 2025–26 funding compression, ksUSD held a **−0.3% max drawdown** while compounding **+10.7% net ($100 → $110.70)**, parking 33% of days when funding didn't pay.

| Metric                                            | Value            |
| ------------------------------------------------- | ---------------- |
| **Max drawdown**                                  | **−0.3%**        |
| Net APY (after 20% perf fee, thin-funding regime) | **\~5%** (5.21%) |
| Gross APY                                         | \~7% (6.56%)     |
| Parked share of days                              | 33%              |
| Parked floor / USDC lending benchmark             | \~4%             |

The result to read here is the drawdown, not the yield. A delta-neutral carry vault that keeps compounding through a funding collapse — losing at most 0.3% peak-to-trough — is the point. Yield is regime-dependent: \~5% net is what that risk profile pays in today's thin-funding regime, scaling up to \~11% when funding is rich. Phoenix pays \~0% funding right now, so the vault sits near its \~4% USDC-lending floor and adds funding on top when it is paid.

> These are **backtest** figures. A live "\~0%" reading in the app's regime widget is a *live* funding snapshot, not a backtest APY.

***

## How \~5% net is built

No model-to-claim ladder: the backtest net **is** the claim. Everything is loaded in, and the only step from gross to net is the performance fee. This is the thin-funding-regime figure; it scales up when funding is rich.

```
  ~7% gross   (already net of perp fees, slippage, mode-switch costs, and the
               9.1%-of-NAV margin haircut — USDC posted as perp margin at ~10×
               leverage earns 0%, not the basis)
  − 20% performance fee (above high-water mark)
  = ~5% net to holder
```

The margin haircut and trading costs sit inside the backtest, so the \~5% is the fully-loaded figure.

***

## Methodology

| Input               | Value                                                                                                                                                                                                                                |
| ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Window              | Jul 2024 – Jun 2026 (24 months, recent track record)                                                                                                                                                                                 |
| Funding source      | Proxy for Phoenix funding, taken at face value: Phoenix anchors its funding to CEX index feeds (Binance, Coinbase, Hyperliquid, Bybit), so the proxy uses daily Binance funding at ×1.0 — it references the same prices Phoenix does |
| Proxy calibration   | Binance funding at face value (×1.0). The old Drift/Binance ratio (2.37× on positive months) ran hot and produced the earlier \~9% figure; the honest CEX-anchored number is \~5% net                                                |
| Mode classification | 7-day rolling mean with a ±3% hysteresis band (whipsaw filter), STANDARD when funding clears the dynamic threshold by the band (0% floor, ≈0% at these rates) else PARKED                                                            |
| Margin haircut      | \~10× leverage on the short, below the SOL market's 25× venue max (over-margined) → 9.1% of NAV as USDC margin at 0% yield                                                                                                           |
| Parked behavior     | USDC at 4% lending APR                                                                                                                                                                                                               |
| Costs               | 5 bps perp fee + 10 bps slippage per side; 20–40 bps mode-switch cost                                                                                                                                                                |
| Fees                | 0% management, 20% performance above HWM                                                                                                                                                                                             |

**Reproduce:** `npm run fund:simulate:v2 -- --start=2024-07-01 --end=2026-06-30` is the headline run. Data CSVs are committed under `scripts/simulations/data/`.

***

## Frictions not in the model

* **Perp-leg price impact during rebalances** — parametric slippage only; real institutional flow is liquidity-dependent and under stress a month can run worse than the modeled −0.3%.
* **Proxy risk** — the proxy uses Binance funding at face value on the premise that Phoenix tracks CEX index feeds; Phoenix's own book may still differ from Binance in practice.
* **Reserve-fund drag** and **hourly→daily smoothing** — not subtracted from the headline.

> **Coming soon:** a net-spread-vs-3-month-T-bill breakdown by funding regime (comparison framing only — the product benchmark stays crypto USDC lending, never a T-bill rate).

***

## Related

* [Whitepaper](/reference/whitepaper.md) — Performance section
* [Backtest data CSVs](https://github.com/kamwithak/keystone-contracts/tree/main/scripts/simulations/data/README.md)