Docs / The trade-off
The trade-off
Protection is not free. Floor charges for it in upside, not in fees we hide. We measured the trade.
You give, you get
| You give | You get |
|---|---|
| About 58% of the gain in an up year. At a 90% floor the vault kept about 42% of a three-stock basket's gain (roughly 4 × (100 − floor)%). | Bad years cut to −8.6% where holding lost 17.2% (NVDA + TSLA + QQQ basket, typical bad year). |
| NVDA: kept about 45% of the gain. QQQ: about 32%. | NVDA: −9.9% where holding lost 36.5% (typical bad year). QQQ: −7.4% where holding lost 19.4%. |
| Trading costs on each rebalance. Live aggregator quotes on a $10k round trip were 0.7 to 6.6 basis points for QQQB, NVDAB, SPCXB and SPYB (Thursday, 2026-10-02), if the vault uses the aggregator route. | A floor you chose, written in a public contract you can read. |
“Typical bad year” is the median of the windows where holding lost more than 10%. It is not the single worst year. Details on the backtest page. Past data, not a prediction.
Upside kept
Share of the gain you keep
42%
Backtest, past data, not a prediction.
Source: docs/data/gap_backtest.csv (NVDA+TSLA+QQQ, open_close, m = 4, capture_up)
The 42% holds at a 90% floor only (about 20% at 95%) and is a pooled average of gains in up years, not a typical year. Across 1,581 one-year periods from 1928 to 2026 the vault's median year was +1.6% against +13.7% for holding: in a normal year you give most of it up. See Evidence.
TSLA is shown on purpose. Choppy price swings made the vault give up the whole gain and lose a little: whipsaw.
Backtest, past data, not a prediction.
Source: docs/data/gap_backtest.csv (open_close, m = 4, capture_up)
Best window, same rule
Strong trends keep more of the gain than the typical 42% (90% floor). NVDA, 8 Mar 2023 to 7 Mar 2024: holding +298.1%, with Floor +240.6%. Past data, not a prediction.
Choppy stocks cost more
Warning
For TSLA, the vault's median year was −6.8%, where holding returned +22.4%. We tell you this before you deposit. Choppy price swings make the vault sell low and buy back high: whipsaw.
Costs
There is no protocol fee in v1. The cost is the upside you give up, plus small trading costs on each rebalance. 1 basis point (bp) = 0.01%. So 0.7 bps = 0.007%, and 100 bps = 1%. The full cost table is on the backtest page. The vault may route directly through PancakeSwap, where we measured about 49 bps for NVDAB and SPCXB at 100 USDT. Weekend and crash-time costs are not yet measured.
Next: how we tested it, on Backtest.