BasisYield collects the hourly payments leveraged traders make on perpetual-futures exchanges — while cancelling out the price risk. This page explains exactly how, what returns to expect, and what can go wrong. No jargon without explanation, no promises without data.
Not familiar with crypto? This page explains the strategy honestly and completely. Some terms are specific to perpetual-futures markets — each is defined the first time it appears. If you have never used a decentralized exchange and want a simpler starting point, the plain-English guide at /learn covers all the foundational concepts before any of the mechanics below.
Honest framing: BasisYield is an automated yield strategy on Hyperliquid, a decentralized exchange that launched in 2024. It is not regulated, not FDIC insured, and carries real platform risk that a savings account does not. The strategy's edge is real and documented — and so are the risks. Both matter before you commit capital.
BasisYield trades on Hyperliquid — a decentralized perpetual-futures protocol — specifically on two of its builder sub-DEXs: markets.xyz (ticker prefix km:) and trade.xyz (prefix xyz:). Both list stocks, gold, oil and currencies as 24/7 perpetual contracts. Your collateral never leaves your Hyperliquid wallet.
To keep each perpetual contract glued to the real-world price, the exchange uses a mechanism called funding: every hour, the heavier side of the market (determined by open-interest imbalance) pays a small fee to the lighter side. The rate is set algorithmically — if there are more longs than shorts, longs pay shorts; if shorts dominate, shorts pay longs. Think of it as rent — impatient traders pay it; whoever warehouses their imbalance collects it.
The key fact our data shows: this rent is persistently one-sided. On many assets the same side paid the fee 84–96% of all hours for a month straight. Someone gets to be the landlord. The bot makes sure it's you.
markets.xyz and trade.xyz both list the same underlying assets — e.g. km:EUR and xyz:EUR — but with independent order books and independent funding rates. The core trade:
If EUR moves, one leg wins exactly what the other loses. What remains is the funding-rate gap between the two venues, collected every hour. The bot scans 11 such pairs continuously, holds the 4 best — only when the gap has been wide and consistent for a trailing week — and rotates out when a gap decays.
Leverage and margin structure. Each leg runs at 5× leverage, held in a separate margin account on its own exchange. The two DEXs do not cross-net margin — if the underlying price trends strongly, the losing leg's collateral drains even as the winning leg's grows. The engine raises a rebalance alert at 10% leg-price drift (giving ~1.8× safety headroom before forced exit at ~18% drift). At 5× leverage, available exchange maximums range from 10× to 50× per asset — the chosen 5× is the highest level where the alert still fires early enough. The backtest Sharpe is 2.80 and is constant regardless of which leverage you pick; the tradeoff is purely about the margin-drain risk window, not yield efficiency.
The strategy earns by collecting funding fees from the crowded side. But funding rates can reverse: if the market becomes balanced, or if short sellers dominate, the rate can go to zero or negative — meaning positions would pay rather than receive.
When this happens, the bot does not force positions to stay open. Each pair has a minimum gap threshold: if the funding spread between the two venues falls below the cost of holding (fees, slippage, rebalance cost), the bot exits that pair and does not re-enter until the spread recovers. On weeks where all pairs simultaneously compressed toward zero, the strategy simply holds no positions and earns nothing. This is visible on the dashboard — weeks with near-zero activity are shown, not hidden.
Historically on these venues, the same direction of funding has persisted 84–96% of hours in a month on many assets. That persistence is the edge. It is a durable structural feature of 24/7 leveraged markets — but it is not permanent. If broader market structure changes (for example, if professional arbitrage capital fully equalizes funding rates across venues), the edge compresses. The dashboard's live opportunity table shows current conditions; the historical backtest shows what the pattern has looked like across all 142 days these venues have existed.
Real markets close Friday 5pm New York; these exchanges don't. With the rowdy crowd alone all weekend, rent runs 3–10× weekday levels. The bot enters Friday evening at 3× leverage (single-leg, liquidation at ~31% price move — far above any observed weekend gap), collects all weekend, leaves at the Sunday reopen.
All weekend the 24/7 price drifts on guesses. Sunday 6pm ET the real market reopens and prints the truth — and for about an hour the 24/7 price visibly converges to it. The bot trades that gap when it exceeds 0.3%. No prediction needed.
A circuit breaker watches two inputs: Bitcoin's 24h move (triggers if ≥5%, evergreen) and Polymarket prediction markets on live geopolitical crises (triggers if odds drop below 25%, or move more than 5 points in 24h). Either trigger → weekend trades cancelled, all new positions halved. It only blocks opening new positions; closing always works.
Important: the Sunday snap-back trade is intentionally directional. The core spread-carry strategy (§02) is market-neutral — it holds offsetting positions that cancel price risk. The snap-back trade is a directional bet on price convergence, not a hedged carry trade. It carries exposure: if convergence takes longer than expected, or if the market moves further away before converging, the position can lose money. The bot sizes it at 3× leverage on a single unhedged leg — liquidation would require a ~31% overnight price move, far above anything observed historically. But it is not market-neutral, and calling it otherwise would be inaccurate.
The dashboard shows three kinds of number and they mean different things:
| Number | What it is | Trust level |
|---|---|---|
| Projected | Today's rates extended for a year | A snapshot — rates change hourly |
| Backtest band | Our exact rules replayed over every day the venues have existed | Real data, simplified execution; past ≠ future |
| Actual | The live track record | The only number that ultimately matters |
The backtest is drawn as a band because execution style changes everything: the bottom edge assumes impatiently crossing the spread every trade (taker), the top edge assumes idealized patient orders (maker), and the solid middle line is our actual plan — hybrid execution: rest a patient order on the quiet venue, and the moment it fills, instantly hedge on the busy one.
Current numbers (142-day backtest, combined strategy, hybrid execution): the spread-carry sleeve at 5× leverage contributes ~15.2% APR on equity; the weekend harvest sleeve at 3× leverage adds ~6.8% APR. Combined: ~22% APR / ~24% APY. The annualised Sharpe ratio across the spread-carry component is 2.80 — and it is constant regardless of which leverage you choose, because both the return and the volatility scale proportionally. The lever that changes is not the Sharpe but how much margin headroom you hold before a rebalance is needed.
Why we can't show a 1-year backtest: these exchanges launched in January 2026. The backtest covers every single day since — no longer history exists anywhere, and we never cherry-pick a window. If someone shows you a longer backtest of this strategy, ask them where the data came from.
BasisYield is non-custodial. There are no deposits and no platform balance — your collateral sits in your own wallet on a decentralized exchange the entire time. We never take possession of it, even for a second. This is an architectural fact, not a policy promise.
You authorize a trade-only agent key — a native permission that can place and cancel orders on your account but cannot withdraw, transfer, or touch anything outside it. That restriction is enforced by the exchange protocol itself. Revoke it anytime with one click and the system is fully disconnected.
Our only revenue is a small builder fee on each executed trade, capped at a rate you approve in advance and recorded on-chain with every fill. No deposits, no assets under management, no performance cut, no withdrawal fees. If the strategy doesn't trade, we don't earn.
Non-custodial means we cannot lose your funds the way a failed exchange does — it does not make the strategy riskless. Markets can move against positions, these venues are young, and smart-contract risk is real. The custody model removes counterparty risk to us; section 06 covers everything else.
MetaMask, Rabby or WalletConnect; funds stay in your wallet on Hyperliquid (USDC).
One signature creates an agent key that can place orders but can never withdraw or transfer. Revoke anytime.
A small, capped, on-chain builder fee per trade. It is the only way the platform earns — there are no deposits to us, ever.
Conservative / Standard / Micro (works from ~$100). The engine trades; the dashboard becomes your account view; one click pauses everything.
Not financial advice. This is an experimental automated strategy on young venues. Past and simulated performance do not guarantee future results. Only allocate what you can afford to lose.
The dashboard is live right now on practice money — every decision, every skip, every fee, explained.
Open the dashboard →