Funding carry at Botty — the first structural edge, and what really defines it live

The idea in one picture

A perpetual future (Perp / Perpetual) is a future without an expiry date. To keep its price glued to the real spot price, there is an hourly balancing payment — the funding. The rule:

If too many people are long (levered on rising prices), the longs pay the shorts. If too many are short, the reverse.

In a bull market there are chronically too many leveraged longs — they must pay, no matter what the price does. So if you hold the unpopular short side in the perp, you collect this funding. To carry no price risk while doing so, you simultaneously buy the same amount in spot:

Buy   1 BTC spot   ─┐
                    ├─►  price movement cancels out (delta-neutral)
Short 1 BTC perp  ─┘     what remains: the funding you collect as the short

If BTC rises, your spot gains exactly what your perp short loses. If BTC falls, the reverse. Net zero price risk — you only keep the funding. No entry signal, no forecast. That is the difference from everything Botty tried before: the edge comes from the counterparty's forced position, not from your market view being right. (Theory: Funding Rate Arbitrage (Delta-Neutral), close relative with an expiry date: Cash-and-Carry / Basis Trade.)

Why this is a big deal at Botty

For months, every TA idea on BTC gave up the ghost under walk-forward (see Out of the dead end - 4 fresh axes to make Botty continuously profitable). The funding carry is the first approach to cleanly pass the cross-asset gate: measured on HL data May 2023–Jun 2026, net of fees, it produced almost identical returns on BTC/ETH/SOL (+6.7 / +6.6 / +5.6% on safe 2x capital). Three assets, same order of magnitude — that is the signature of a structural edge, not a random find fitted to BTC. BTC timing never showed exactly this pattern.

The capital model (important for understanding the return)

Delta-neutral ties up capital in both legs:

The ~11% "APR" you see in the scanner is the return on notional. On the capital actually tied up, the safe variant yields more like ~5%. Doing this math honestly is critical — otherwise you flat-out overestimate the carry by a factor of two.

Lesson 1: The live value lies — look at the realized return

The most important insight came while building. The scanner shows the current funding. But that is a snapshot which often doesn't hold. Computing the actually realized return over the real funding history (trailing 60 days):

Two things jump out: ZEC's spectacular 40% was a spike — realized only 1.5%. And BTC trailing 2.7% is barely more than just letting USDC sit. Whoever follows the loudest live value is chasing flashes in the pan. HYPE, on the other hand, delivers the highest persistent carry at +9.5% with 87% positive hours — and happens to also be the technically simplest variant (see lesson 2). The best and the simplest coincide here.

Lesson 2: For the majors it is necessarily cross-venue

Delta-neutral needs liquid spot AND perp in the same coin. Hyperliquid (Hyperliquid) has perps for everything — but the spot market is its own ecosystem of community tokens. For the majors, only bridged variants exist (UBTC/UETH/USOL), and they are practically dead (24-hour volume near zero). Concretely:

• HYPE has deep native spot and perp on HL → the only clean pure-HL variant (one venue, one USDC collateral). Downside: concentration in HL's own token (platform/depeg risk — delta-neutral only protects against the price, not against a platform event).
• BTC/ETH/SOL need the spot leg elsewhere (Binance/Coinbase) → the perp short runs on HL, the spot long on Binance. That is the standard carry-desk setup, but two venues, two capital pools, transfers and venue-basis risk. And Botty's execution is currently HL-only — a Binance spot trading client would be a project of its own.

What is already built (all dry-run, no live capital)

• Scanner backtesting/funding_carry_scanner.py — ranks HL coins live by net carry on a capital basis.
• Spec execution/funding_carry_spec.md — architecture, state machine, guards, rollout phases.
• Dry-run manager execution/funding_carry/ — the full state machine (SCAN→OPEN→HOLD→REBALANCE→CLOSE), but sends no real orders; the accrual verifier computes against the real HL fundingHistory and proves that the funding math is correct (and that the data is hourly).

Honest verdict

Real and structural — but modest. Directionless ~5–9% depending on coin and timing, cooling off long term. Not a rocket, but a quiet carry sleeve whose real value is its lack of correlation to the directional strategies (a true diversifier). For the majors, the operational effort is barely worth it (~2–5%); the convincing candidate is HYPE pure-HL (~9.5%, simplest execution). This clearly frames the next decision — it just hasn't been made yet.

Capacity & scaling (measured 2026-06-20)

How much fits into the HYPE carry? There are two legs, each with its own limit — the smaller one binds:

• HYPE perp (the short leg): OI $1.45bn, 24h volume $533mm — bottomlessly deep, no limit.
• HYPE spot (the long/hedge leg): the bottleneck. Measured order book: ~$90-100k of depth within 0.5% per side (tight at the mid, barely grows further out).

From that: cleanly deployable ~$50-100k per round trip with <0.5% slippage, more with orders split over time. And because a carry is a hold (enter once, hold for weeks, exit once), the one-time entry/exit slippage is negligible against ~9% funding over weeks.

Why spot is so much thinner than the perp: Hyperliquid is a perps-first exchange. The speculation/leverage runs in the perp; spot is a secondary market, and most HYPE tokens sit in wallets/staked, not as resting spot orders. Spot is thus ~4 orders of magnitude thinner — thin in relative terms, but in absolute terms fine for tens of thousands of $.

For our size ($60 micro): ~100-1000x headroom. The binding factor is not liquidity but concentration prudence — serious money in ONE meme token on ONE exchange is the risk.

Cross-venue basket: what it REALLY delivers

A common fallacy (mine too): that the cross-venue majors basket (HL perp short + Binance spot long, because HL spot is dead for the majors) would deliver higher returns. It does not. The yields are ~equal (HYPE is even among the highest on HL), and with two-venue friction (transfer fees, basis risk) marginally less net. Its value is purely risk: diversification away from the HYPE concentration + unlimited scaling (BTC/ETH spot is bottomless). For anything up to ~$50-100k, HYPE solo is fully sufficient — you only need the basket for larger sums or deliberate de-concentration.

Leverage in the carry: only the perp margin, and capped (2026-06-20)

The delta-neutral carry has two legs, but only one can be levered: the spot leg is full cash (no spot leverage on HL) — the anchor. The leverage sits solely in the perp margin (spot/margin split). Computed on the live HYPE funding (~13.3% notional yield, trailing-30d) via backtesting/carry_leverage_calc.py:

2x -> 1.33x yields +3.1 percentage points (~+50% relative) — the price is perp 3x instead of 1x, i.e. liquidation risk if HYPE rallies ~31% before the hourly margin guard shifts the spot gains into the perp margin. Under delta-neutral, the spot gain economically offsets the perp loss — but the legs sit in separate pots, so the perp can liquidate first (gap risk). More than ~1x capital is impossible without borrowing the spot leg (borrow interest eats the carry) → the ceiling sits at ~13% notional yield, and that is structurally cooling. Live deliberately runs the safe 2x mode. Switching to 1.33x is a risk-return trade, not a free lunch.