Each row converts the current basis into an annualized return after funding. Gross carry is locked at entry by hold-to-expiry; funding is a projection of the current rate. Sorted by net carry %.
basis_bps = (perp_mark − cme_last) / cme_last × 10 000 — how rich (positive) or cheap (negative) the perp is vs CME.
gross carry % = |basis_bps| / 10 000 × 365 / days_to_expiry × 100 — what the basis is worth annualized, assuming it converges to zero at CME expiry. This is locked if you execute both legs at quoted prices.
funding % p.a. = funding_rate × 24 / interval_h × 365 × 100, sign-flipped if you're long the perp (longs pay positive funding, shorts collect it). Funding interval: Binance 4h, OKX 8h, Hyperliquid 1h. This is a projection — actual realised funding is the integral over the holding period.
net carry % = gross + funding — total annualized return.
Days-to-expiry is the approximate front-month CME settle (GC ≈ 28th of month, CL ≈ 21st of month). The 24h Vol ($) column tells you whether the venue is deep enough to actually trade.
Gross carry (locked) is guaranteed if you (1) execute both legs at current quoted prices and (2) hold to CME expiry. CME futures must settle to spot at expiry; the perp tracks spot continuously via funding. So the basis spread you enter at converges to zero at expiry — that P&L is structurally locked.
Funding % p.a. (variable) is an extrapolation of today's funding rate. Over a 21-day hold, you'll realise whatever funding actually accrues hour-by-hour. Commodity perps tend to keep funding rates more persistent than crypto pairs (no spot-arb path back), but there is no guarantee — funding can flip sign sharply.
What still has to be true for net carry to be realisable:
• Both legs filled at quoted prices — slippage on entry/exit erodes the locked basis
• You have capital + a CME futures broker for the short leg (1 CL contract = 1,000 barrels ≈ $105k notional; 1 GC contract = 100 oz ≈ $460k)
• You can post and maintain initial + maintenance margin on both sides without a margin call forcing an early exit
• Counterparty risk on each leg (Hyperliquid HIP-3 markets are newer than native venues)
| Venue | Product | Basis (bps) | Days to CME | Gross carry %locked at entry | Funding % p.a.current snapshot | Net carry %locked + projected | 24h Vol ($) | Setup hover for detail |
|---|---|---|---|---|---|---|---|---|
| Loading… | ||||||||
The total — basis convergence + funding. Same math as the Net column in section 1, evaluated at every snapshot. Compare with section 3 below: the gap between net and gross IS the funding contribution.
Basis-only annualized carry — what's structurally locked in at entry by hold-to-expiry. This is the "guaranteed floor" component. If this line is high and stable, the trade has a durable structural edge regardless of funding.
All three pairwise basis lines between the crypto venues. Lives through weekends and CME closures. Sign: (A − B) / B × 10 000.
Top 10 largest |basis| events in the selected range, ranked by absolute basis. Gross carry annualizes the dislocation against time-to-CME-expiry — a 30 bps basis 60 days from expiry is a much bigger trade than 30 bps with 1 day left.
| Loading… |
Each dot is one snapshot for one venue. X = funding rate annualized (% p.a.), Y = basis vs CME (bps). Tests whether funding is correcting the basis.
The mechanism. Funding rate is designed to pull the perp price toward the underlying. When perp trades rich (positive basis), too many longs → funding goes positive → longs pay shorts → being long becomes expensive → basis closes. When perp trades cheap, the opposite happens with negative funding. So if funding is working, dots should fall along an upward-sloping diagonal (top-right and bottom-left).
Steep, tight upward slope → market is efficient, basis closes fast, carry trades are short-lived (hours, not days).
Flat / scattered cloud → funding isn't tracking basis. The mechanism is broken or laggy. This is where opportunities live — basis can stay open for days because nothing is forcing convergence.
Practical use. Before entering a carry trade from section 1, look at the cloud for that venue: a flat shape says the carry is durable, a steep shape says move fast.
With only a few snapshots so far, the chart is sparse. After a few days of cron data it becomes a real regime read.
Rolling 24h USD volume per venue, over time. Tells you how deep the market actually is — a 50% net carry on a venue trading $10k/day isn't a real trade. Use this to sanity-check section 1.