Funding carry

funding.whalestamp.com · Kraken perps · market-neutral

Predicting price is a game retail loses. Funding is different — a cash flow you get paid to hold, no forecast required. This harvests it across the whole perp board: find the handful of contracts paying the most to hold right now, hold them delta-neutral, rotate as funding shifts. Every number here is net of realistic execution costs — spread, fees, market impact, and basis risk — and only counts funding you could actually capture; a positive number still has to survive weeks out-of-sample, spanning a funding regime change, before any capital is risked.

Readiness gate

Evaluating…

Phase 1 · the universe

Live cross-section — who pays the most right now

Every tracked perp ranked by annualized funding magnitude. This is available instantly; the higher and more one-sided, the more there is to harvest.

Loading the board…
Phase 2 · the engine

Cross-sectional carry backtest

Hold the top-k highest-funding perps delta-neutral, rotate with hysteresis so fees don't churn it away, net of real fees. Compared against harvesting LTC alone — the whole point of picking from many.

Basket size Min funding Fee
Running…
Phase 3 · the discipline

Config lab — in-sample vs out-of-sample

Every config is scored on the first half of history (in-sample) and the second half it never saw (out-of-sample). Only out-of-sample Sharpe counts — a config that shines in-sample and dies out-of-sample is noise, and this is where most backtests lie to you.

Scoring configs…
Phase 4 · the gate

Prove, then risk

Capital only goes in when a config has held up out-of-sample with a real Sharpe over a meaningful window (≥14 days). Until then it's research, not a trade — the gate at the top of the page is the honest verdict. Basis risk and margin costs aren't modelled here; this sizes the opportunity, it isn't a live P&L. Part of the whalestamp research stack. Not investment advice.

$10,000 paper run

What $10,000 would have done

A paper run of the strategy above — your selected basket and fee, on a $10,000 stake, over the recorded funding window. Gross of slippage and basis risk, so an upper bound, not a live account.

Waiting for the backtest…
Refined run · cost-aware, fresh $10,000

Refined run — the mitigations, from a fresh $10,000

A second paper run testing whether disciplined execution makes this profitable: cost-aware rotation (hold ≥24h, only switch when funding beats the round-trip cost), a tight liquidity tier, and both-sides capture (positive funding = short perp, negative = long perp — the negative side assumes a feasible spot short). Switch the execution assumption: taker crosses the spread; maker posts limit orders and doesn't pay it. Cross-venue funding (mitigation 4) is excluded for now. This is a true forward run: it starts from zero the moment live bid/ask began recording and only ever trades on real spreads — no hindsight replay of the pre-deploy history. Every trade is logged below in pages of 50.

Execution
Running refined…