Executive Summary

Delta-Neutral Spot+Perp Funding Arbitrage on Hyperliquid

8.079
Sharpe
5c carry, 2.7yr
10.065
Sortino
38.0%
CAGR
140.7%
Total Return
-4.0%
Max Drawdown
9.39
Calmar
100.0%
MC P(>1)
10k bootstrap
0.04
Corr BTC
near zero
83.4%
Win Rate
daily
1.278
BTC Sharpe
Buy&Hold ref
CarryArb delivers a Sharpe of 8.079 over 2.7 years — pure funding income from delta-neutral spot+perp positions. MaxDD of only -4.0% vs BTC buy-and-hold (-32.0%).

Performance

Production config: 5 coins, carry selection, monthly rebalance

Consistent growth from $10,000 to $24,070. Near-zero drawdown (-4.0%) — delta-neutral eliminates price risk.
Near-zero correlation with BTC (0.04) — pure alpha from funding, not crypto beta.
MetricValue
Sharpe8.079
Sortino10.065
CAGR38.0%
Total Return140.7%
Max Drawdown-4.0%
Max DD Duration96d
Calmar9.39
Win Rate83.4%
Profit Factor7.43
Max Consec. Losses17d
Corr BTC0.04
Beta BTC0.0

Portfolio Size Comparison

More coins = higher Sharpe, lower return. Fewer = concentrated, higher return.

More coins = higher Sharpe (more diversification, lower DD) but lower absolute return. Fewer coins = concentrated bets with higher return but more volatility. 5 coins is the sweet spot for the live deployment.
ConfigSharpeReturnMaxDDCalmar
3c_carry6.581171.1%-7.3%6.05
5c_carry8.079140.7%-4.0%9.39
7c_carry8.749127.8%-3.4%10.3
10c_carry9.647111.6%-3.0%10.63

Selection Method

Carry (mean funding) vs Sharpe (risk-adjusted funding)

Carry selection (rank by mean funding rate) generates 3x more absolute return than Sharpe selection (rank by funding rate / volatility). Carry selection picks the highest-paying coins regardless of volatility — this works because delta-neutral positions eliminate price risk.

Fee Sensitivity

Sharpe at different fee levels — Nautilus backtests (5c carry)

FeeSharpeReturnMaxDD
0 bps8.142+142.0%-3.9%
~1.5 bps (maker)8.124+141.6%-4.0%
2 bps8.106+141.2%-4.0%
3.5 bps8.079+140.7%-4.0%
~4.5 bps (taker)8.05+140.2%-4.1%
7 bps8.011+139.4%-4.1%
CarryArb is fee-insensitive — monthly rebalancing means fees are negligible compared to funding income. Even at 7bps per leg, Sharpe barely changes. This is a major structural advantage over high-frequency strategies.

Risk Protection

Funding floor and exit guards from VBT validation (Script 06)

ConfigSharpeReturnMaxDDOOS/IS
baseline8.21141.7%-3.0%1.07
floor=08.29142.2%-3.0%1.06
floor=1e-58.42140.8%-2.5%1.05
floor=2e-58.45138.8%-2.4%0.83
floor=3e-58.32136.2%-2.4%0.67
exit_neg_24h8.69129.0%-0.5%1.76
floor=0+exit_24h8.52127.6%-0.5%1.81
floor=2e-5+exit_24h8.22119.7%-0.6%1.62
exit_neg_48h5.98185.8%-1.1%2.47
floor=0+exit_48h5.89184.3%-1.1%2.51
10c_baseline9.1852.8%-2.0%1.17
10c_floor=09.5646.6%-3.0%1.15
10c_floor=2e-510.3061.1%-0.4%1.22
10c_floor=0+exit_24h12.2854.4%-0.1%2.31
The exit-on-negative-24h guard closes positions when trailing 24h funding turns negative. This reduces MaxDD 6x (from -3.0% to -0.5%) and improves OOS robustness (ratio 1.81). The strategy is robust with or without the guard — it's an optional safety layer.

Monte Carlo Analysis

Block bootstrap on Nautilus equity curves

Block bootstrap (block=20 days, 10k samples). P(Sharpe>0)=100.0%, P(Sharpe>1)=100.0%. 95% CI: [6.452, 9.95]. The funding carry edge is highly statistically significant.