Two products, one pool

Retail buys Curve Long, a lottery ticket on BTC that can't be liquidated. The only short is a Hedged Vault: it sells Curve Short and holds BTC to cancel the price exposure, so what's left is the funding the longs pay. This page runs that setup on real BTC prices and on random paths, and checks whether the vault beats ETH staking.

Where the money goes

What $1,000 in the vault becomes

One depositor who joins on day one and stays. Hedged vault against the same vault without a hedge and against benchmark staking. BTC price for context (grey, right axis).

Hedged vaultSame vault, no hedgeBenchmark stakingBTC

Vault size and the yield depositors see

Vault sizeRetail long claimsTrailing 30-day yield (right axis)Target yield

Vault PnL, piece by piece

Funding is the income. Rebalancing the hedge daily costs money whenever BTC moves (the short curve's convexity). That cost is what the premium has to beat.

Retail longs

The other side of the same run: what the lottery ticket did for the people buying it.

How individual longs ended

Every closed Curve Long in this run, grouped by what it returned (payout ÷ ticket, after fees).

Search: which settings make both sides want in

750 settings (players, short premium, highest curvature, steep-curve surcharge), each run on 32 random years: calm, wild, bull and bear BTC. Vault $1M, $400 tickets, 30-day holds. A setting passes if the vault's median year beats 7% and 9 of 10 years beat 3% staking. Among those, the table shows the one that pays retail back the most.

Open playersPremiumHighest pSurchargeVault medianVault bad yearRetail paid back per $1Chance to winChance of 2×+
500%10+200%+7.3%+5.6%$0.7224%2.9%
100+100%4+200%+7.9%+6.3%$0.8431%1.1%
200+50%4+50%+7.5%+5.1%$0.9135%2.1%
400+25%3+50%+7.5%+3.3%$0.9337%1.2%
800+100%20%+14.7%+5.4%$0.9438%0.3%
200, today's contract−50%100%−11.7%−21.6%$1.0036%9.5%
  • Scale is the lever. With more players per dollar of vault, the same yield takes a smaller cut of each ticket: retail payback rises from $0.72 at 50 players to $0.94 at 800.
  • Steep curves are the vault's income and retail's lottery. At 200 players and +25% premium, raising the cap from p = 4 to p = 10 lifts the vault from 3.4% to 9.2% and the chance of 2× from 2.5% to 6.3%, but lowers payback from $0.94 to $0.86 and doubles the vault's drawdown (−2.2%).
  • Today's multiplier can't work: retail gets every dollar back on average, and the vault loses 1% to 12% a year. No depositor stays.
  • Payback below $1 is the price of the ticket. A casino slot pays back about $0.90 to $0.95 and a state lottery about $0.50. Around $0.90 is a fair target for a long that can't be liquidated and can't lose more than the ticket.

Each preset uses 200 open players and resets everything else to the defaults. Script: sim/search.js, full output in sim/search-results.txt.

How many players does the vault need?

Vault yield against the number of retail longs open at any time, with everything else as set on the right. Each point is the median of 30 random one-year paths.

Vault APY, median10th to 90th percentileBenchmarkUtilization (right axis)

Monte Carlo: one year, 200 random paths

Random BTC paths with the volatility and drift set on the right. Shows how often the vault beats the benchmark, and how bad the bad years are.

Does the two-product setup work?

How the vault works, step by step

A long claim is what the pool owes a Curve Long right now: ticket × (BTC now ÷ BTC at entry)p, minus funding paid so far. A $100 long at p = 3 is owed $133 after a 10% rally (1.1³ = 1.331) and $73 after a 10% drop. The pool pays every claim out of the same USDC, and whatever is left belongs to the short side.

Why p × claims of BTC. When BTC moves 1%, a claim at curvature p moves about p%. A $1,000 claim at p = 10 grows by about $100 for a 1% rally, the same as holding $10,000 of BTC. So the vault cancels that move by holding $10,000 of BTC, ten times the claim. At p = 2 it needs only $2,000. This is why steep curves tie up the most capital.

Daily rebalancing. The amount needed changes as BTC moves: after a 10% rally that $1,000 claim at p = 10 is worth about $2,594, so the vault now needs about $25,900 of BTC, not $10,000. Once a day the keeper reads the oracle, computes p × claims for every curve, and buys or sells BTC to match. It always buys after a rise and sells after a fall. That small loss on every move is the cost of being short convexity, and the funding has to cover it.

Deposits in USD only. A depositor sends USDC (or any token through NEAR Intents, swapped to USDC). The vault itself splits the money: part goes into the pool as Curve Short, part buys BTC for the hedge. Depositors own vault shares, not BTC. When BTC rises, the BTC in the vault gains about what the short side loses, so the share price stays roughly flat and moves up by the funding earned. The depositor never sees the BTC leg.

Capital and yield find a balance. Yield ≈ net spread × claims ÷ vault size. A small vault next to many players earns a lot, which draws in deposits; deposits grow the vault, and the same funding is split over more capital, so yield falls until it is just above what depositors can get elsewhere. Switch on "Capital follows yield" to watch this: the vault settles at a size set by retail activity, not by its starting capital. A small starting fund is fine; it starts with a high yield.

Why the hedge isn't free

Curve Short holds −(claims), and claims grow like Pp. The vault cancels the first-order move by holding p × claims worth of BTC, rebalanced daily. What's left each day is:

vault PnL per day ≈ claims × p(p−1)/2 × (funding variance × multiplier − realized variance) × dt

So the vault earns only if funding is priced above the volatility that actually happens. Two consequences: the funding multiplier must stay at or above 1 even at low utilization (today's contract uses 0.5 + u, which pays shorts less than fair below 50%), and funding should follow realized volatility, not a fixed setting. The premium slider on the right is that markup.

The second limit is size: a vault can only earn on long claims that exist. Yield on vault capital ≈ net spread × claims ÷ vault capital. More players or a smaller vault raise it, and utilization rises with it, which raises funding again.