Funding carry: a funding rate strategy, backtested on Hyperliquid
Funding carry earns the funding that crowded traders pay on Hyperliquid perps: each day it shorts the coins paying the richest funding and buys the ones with negative funding, balanced long and short. From 2024-08-15 to 2026-09-29 it returned +49.2% with a Sharpe of 1.39 and a 10.5% worst drawdown, net of fees, slippage and funding. It is a perp-only carry across many coins, not the spot-against-perp cash and carry most funding rate arbitrage guides describe.
2024-08-15 → 2026-09-29 · net of fees, slippage, and funding · price-only Sharpe 0.33 · these are the default settings, not a hand-picked best case.
At $1,000, this strategy’s worst historical dip was about $105. Free account, your own Hyperliquid keys. Keel only trades.
Arbitrage on one venue
Crypto arbitrage means buying a coin where it is cheap and selling it where it is dear at the same moment, so the price gap is the profit and the market's direction doesn't matter. The usual forms are the same coin on two exchanges, a coin's spot price against its perp, or a loop through three trading pairs.
Keel can't do any of these. It trades one venue, Hyperliquid, with perps only and no spot leg, and a strategy acts at the close of each bar, so it can't catch a price gap between exchanges or hold a hedged spot position.
Funding carry is what Keel runs instead, and it is a funding strategy, not an arbitrage. It collects the funding gap between coins on one venue, and because its long and short sides are different coins, it carries real market risk.
Funding rate arbitrage, in plain words
A perpetual future never expires, so the exchange keeps its price close to the coin's spot price with a funding payment between traders. When longs crowd in, longs pay shorts. When shorts crowd in, shorts pay longs. On Hyperliquid this is paid every hour.
Funding rate arbitrage usually means cash and carry on one coin: buy the coin on spot, short the same coin's perp, and collect the funding while the two legs cancel each other's price moves. Some traders do the same across two exchanges, long the perp where funding is low and short it where funding is high.
What this strategy does instead
This strategy holds no spot and uses one venue. Each day it ranks the largest Hyperliquid perps by funding rate, shorts the ones paying the richest funding and goes long the ones where shorts are paying, sized so the long side and the short side roughly balance, which still leaves real market risk.
The price legs do not cancel exactly, because they are different coins. The return comes from the funding gap between them, spread across many coins instead of one hedged pair. Price moves alone returned +8.2% over the test. With funding included, the total was +49.2%, after fees and slippage. Funding is what carries this strategy.
It is not risk free. When funding falls toward zero there is little to collect. When a crowded short squeezes, the coins it is long can fall while the coins it is short keep rising.
How it works
How it works
Perpetual funding is a crowding fee: when longs are over-leveraged they pay
shorts to hold the position, and when shorts are crowded the payment flips. This
strategy collects that fee. Each day it ranks the top-100 Hyperliquid perpetuals
by funding rate, shorts the names paying the richest funding and longs the names
paying negative funding, then standardizes the signal across all names so the
book is balanced dollar-for-dollar long and short at 1x gross. The published
top100_launch preset retains its name but explicitly excludes PURR, leaving 99
resolved members so the advertised window uses only authoritative coverage. It
rides the funding spread rather than the market's direction. A 10-bar smoothing
calms the signal and a 15% drift buffer holds turnover down.
When it works best
Crowded, over-leveraged markets where funding is actually being paid and the spreads between names are wide — most of the 2024-2026 window. Because the book is balanced long and short, the return comes from the funding gap rather than market direction. Over 2024-08-15 to 2026-09-29 the rerun default returned +49.2% with a Sharpe of 1.39 and a maximum drawdown of −10.5%, net of fees, slippage, and funding.
When it struggles
When funding compresses toward zero there is little left to collect. It also loses when crowded shorts squeeze — the negative-funding names it is long get sold off while the high-funding names it is short keep running. The price side of the book is only mildly positive on its own; funding is what carries the result. The default placed 17,680 trades in 3,210 positions over the window, and the funding collected more than covers the fees.
How it’s built
The exact strategy behind this backtest, with no black box. Fork it to change it and run your own backtest.
Configuration
Globals(target_timeframe="1d")Universe(mode="top_volume", top_n=100, market="perp", lookback="90d", exclusions=["PURR"], resolved=["AAVE", "ADA", "AERO", "ALGO", "APE", "APT", "AR", "ARB", "ASTER", "AVAX", "AXS", "BCH", "BIO", "BLUR", "BNB", "BTC", "CAKE", "CC", "CHIP", "CRV", "DASH", "DOGE", "DOT", "DYDX", "EIGEN", "ENA", "ETC", "ETH", "ETHFI", "FARTCOIN", "FET", "FIL", "GRAM", "GRASS", "HBAR", "HYPE", "ICP", "INIT", "INJ", "IP", "JTO", "JUP", "KAITO", "LDO", "LINK", "LIT", "LTC", "MEGA", "MEME", "MET", "MON", "MORPHO", "NEAR", "NIL", "ONDO", "OP", "ORDI", "PAXG", "PENDLE", "PENGU", "PNUT", "POL", "POPCAT", "PUMP", "PYTH", "RENDER", "RUNE", "SEI", "SKY", "SOL", "SPX", "STABLE", "STRK", "SUI", "SYRUP", "TAO", "TIA", "TON", "TRUMP", "TRX", "UNI", "VIRTUAL", "VVV", "W", "WIF", "WLD", "WLFI", "XLM", "XMR", "XPL", "XRP", "ZEC", "ZEN", "ZORA", "ZRO", "kBONK", "kNEIRO", "kPEPE", "kSHIB"], resolved_at="2026-07-07T19:46:36.522366+00:00")Execution(rebalance="buffered", buffer_threshold=0.15, buffer_mode="relative", rebalance_method="to_edge", min_trade_size=0)Pipeline([ PriceDataLoader(), FundingDataLoader(), NegateTransform(), EWMATransform(window=10), CrossSectionalZScore(), ForecastScaler(avg_abs_target=10), ForecastCapper(limit=20), ForecastWeightNormalizer(target_leverage=1),], name="funding_carry")Explore the settings
precomputed · updates instantlyAdjust a setting to see the exact backtested result, including the ones that lost money.
Complete 18-cell public cross-product on explicit resolved universes.
At $1,000, these settings' worst historical dip was about $105. You land in the editor with this exact setup. Free account, your own keys.
Compare all 18 settings
| Funding smoothing | Rebalance buffer | Pinned universe | Sharpe | Return | Worst DD | Trades |
|---|---|---|---|---|---|---|
| 5 | 0.1 | Top 60 coins | 1.32 | 56.2% | −9.3% | 14,539 |
| 5 | 0.1 | Top 100 coins | 1.18 | 40.3% | −10.0% | 23,230 |
| 5 | 0.15 | Top 60 coins | 1.36 | 57.3% | −9.0% | 12,820 |
| 5 | 0.15 | Top 100 coins | 1.24 | 41.9% | −9.7% | 20,501 |
| 5 | 0.2 | Top 60 coins | 1.44 | 60.0% | −8.3% | 11,508 |
| 5 | 0.2 | Top 100 coins | 1.28 | 42.6% | −9.2% | 18,321 |
| 10 | 0.1 | Top 60 coins | 1.43 | 62.8% | −10.3% | 12,699 |
| 10 | 0.1 | Top 100 coins | 1.31 | 46.3% | −11.1% | 20,552 |
| 10 | 0.15 | Top 60 coins | 1.50 | 65.8% | −10.3% | 10,891 |
| 10 | 0.15 | Top 100 coinsdefault | 1.39 | 49.2% | −10.5% | 17,680 |
| 10 | 0.2 | Top 60 coins | 1.58 | 68.6% | −10.4% | 9,566 |
| 10 | 0.2 | Top 100 coins | 1.48 | 51.7% | −9.7% | 15,490 |
| 20 | 0.1 | Top 60 coins | 1.52 | 70.7% | −12.4% | 11,222 |
| 20 | 0.1 | Top 100 coins | 1.45 | 53.2% | −12.1% | 18,351 |
| 20 | 0.15 | Top 60 coins | 1.57 | 73.1% | −12.4% | 9,515 |
| 20 | 0.15 | Top 100 coins | 1.53 | 56.3% | −11.9% | 15,454 |
| 20 | 0.2 | Top 60 coins | 1.62 | 75.5% | −12.5% | 8,277 |
| 20 | 0.2 | Top 100 coins | 1.61 | 58.9% | −11.8% | 13,349 |
The data
Monthly returns
| Month | Return |
|---|---|
| 2024-08 | 0.0% |
| 2024-09 | -1.5% |
| 2024-10 | -1.1% |
| 2024-11 | +18.4% |
| 2024-12 | +7.9% |
| 2025-01 | +5.6% |
| 2025-02 | +11.8% |
| 2025-03 | +3.0% |
| 2025-04 | -0.8% |
| 2025-05 | -5.6% |
| 2025-06 | +2.0% |
| 2025-07 | +1.7% |
| 2025-08 | +6.1% |
| 2025-09 | -3.5% |
| 2025-10 | -0.3% |
| 2025-11 | -0.2% |
| 2025-12 | +1.1% |
| 2026-01 | +4.5% |
| 2026-02 | +3.0% |
| 2026-03 | -0.4% |
| 2026-04 | +1.3% |
| 2026-05 | -4.3% |
| 2026-06 | +3.6% |
| 2026-07 | +0.1% |
| 2026-08 | -4.5% |
| 2026-09 | -0.7% |
Which assets it traded
| Avg allocation | Days held | % of time held | Asset |
|---|---|---|---|
| -0.0135 | 774 | 99.87 | LDO |
| 0.0082 | 773 | 99.74 | APT |
| -0.0089 | 773 | 99.74 | NEAR |
| -0.0135 | 773 | 99.74 | POPCAT |
| -0.0063 | 773 | 99.74 | WIF |
| 0.0024 | 772 | 99.61 | FET |
| -0.003 | 771 | 99.48 | CRV |
| 0.0078 | 771 | 99.48 | JUP |
| 0.0089 | 771 | 99.48 | MEME |
| -0.0099 | 771 | 99.48 | PENDLE |
| 0.0132 | 771 | 99.48 | SEI |
| 0.0088 | 771 | 99.48 | kSHIB |
| -0.0127 | 770 | 99.35 | AAVE |
| 0.0048 | 770 | 99.35 | APE |
| -0.004 | 770 | 99.35 | CAKE |
| 0.0068 | 770 | 99.35 | DOT |
| -0.0008 | 770 | 99.35 | ETHFI |
| -0.0046 | 770 | 99.35 | HBAR |
| 0.0088 | 770 | 99.35 | OP |
| -0.0055 | 770 | 99.35 | RUNE |
| 0.0023 | 770 | 99.35 | SUI |
| 0.0073 | 770 | 99.35 | TRX |
| -0.0101 | 770 | 99.35 | UNI |
| -0.0035 | 770 | 99.35 | kBONK |
| -0.0097 | 770 | 99.35 | kPEPE |
| -0.005 | 769 | 99.23 | BTC |
| -0.0009 | 769 | 99.23 | EIGEN |
| 0.0071 | 769 | 99.23 | ENA |
| -0.0106 | 769 | 99.23 | LINK |
| 0.0059 | 769 | 99.23 | TIA |
| -0.0012 | 769 | 99.23 | WLD |
| -0.0041 | 769 | 99.23 | ZRO |
| 0.0089 | 768 | 99.1 | BCH |
| -0.0007 | 768 | 99.1 | ETH |
| 0.0023 | 768 | 99.1 | FIL |
| 0.0003 | 768 | 99.1 | ONDO |
| 0.0077 | 768 | 99.1 | PYTH |
| -0.0102 | 768 | 99.1 | TAO |
| 0.0092 | 768 | 99.1 | W |
| -0.0078 | 767 | 98.97 | AR |
| 0.0053 | 767 | 98.97 | INJ |
| 0.0039 | 767 | 98.97 | JTO |
| 0.0027 | 767 | 98.97 | SOL |
| -0.0009 | 767 | 98.97 | XRP |
| 0.009 | 766 | 98.84 | ADA |
| -0.0018 | 766 | 98.84 | DYDX |
| 0.0052 | 766 | 98.84 | ETC |
| -0.0051 | 766 | 98.84 | ORDI |
| 0.0035 | 765 | 98.71 | ARB |
| 0.005 | 765 | 98.71 | BNB |
| -0.0046 | 765 | 98.71 | DOGE |
| -0.0079 | 765 | 98.71 | LTC |
| -0.0048 | 765 | 98.71 | RENDER |
| 0.0029 | 764 | 98.58 | AVAX |
| 0.0065 | 764 | 98.58 | BLUR |
| -0.0013 | 762 | 98.32 | ZEN |
| 0.0068 | 761 | 98.19 | STRK |
| 0.0042 | 732 | 94.45 | POL |
| 0.0026 | 701 | 90.45 | kNEIRO |
| -0.0008 | 694 | 89.55 | GRASS |
| 0.0043 | 677 | 87.35 | PNUT |
| 0.0099 | 673 | 86.84 | XLM |
| -0.0024 | 661 | 85.29 | TON |
| -0.0154 | 658 | 84.9 | HYPE |
| 0.0007 | 652 | 84.13 | ALGO |
| 0.0021 | 648 | 83.61 | VIRTUAL |
| -0.0204 | 647 | 83.48 | FARTCOIN |
| 0.0021 | 646 | 83.35 | PENGU |
| 0.0094 | 630 | 81.29 | BIO |
| -0.005 | 624 | 80.52 | SPX |
| -0.0014 | 616 | 79.48 | MORPHO |
| 0.0137 | 616 | 79.48 | TRUMP |
| 0.0045 | 604 | 77.94 | VVV |
| 0.0159 | 582 | 75.1 | KAITO |
| -0.0009 | 551 | 71.1 | NIL |
| -0.0037 | 537 | 69.29 | PAXG |
| 0.0098 | 522 | 67.35 | ZORA |
| 0.006 | 518 | 66.84 | INIT |
| 0.0145 | 501 | 64.65 | IP |
| -0.0086 | 457 | 58.97 | SYRUP |
| -0.0133 | 445 | 57.42 | PUMP |
| -0.0129 | 403 | 52 | XPL |
| 0.0071 | 401 | 51.74 | WLFI |
| -0.0109 | 372 | 48 | ASTER |
| 0.0021 | 372 | 48 | SKY |
| -0.0043 | 358 | 46.19 | ZEC |
| -0.0034 | 355 | 45.81 | MON |
| -0.009 | 351 | 45.29 | MET |
| -0.0107 | 342 | 44.13 | MEGA |
| -0.0061 | 331 | 42.71 | CC |
| 0.0057 | 322 | 41.55 | ICP |
| -0.0008 | 320 | 41.29 | AERO |
| 0.0137 | 295 | 38.06 | STABLE |
| -0.0116 | 281 | 36.26 | LIT |
| -0.0182 | 257 | 33.16 | XMR |
| 0.0145 | 254 | 32.77 | AXS |
| 0.0013 | 251 | 32.39 | DASH |
| 0.0056 | 158 | 20.39 | CHIP |
| -0.0165 | 89 | 11.48 | GRAM |
What this is: a historical backtest on real Hyperliquid market data, net of fees, slippage, and funding. Its worst historical drawdown was 10.5%. Expect drawdowns of that order or worse. Past performance does not predict future results, and this is not investment advice. Size your account so a full drawdown is survivable.