Is funding-rate farming profitable?

Yes, but modestly — a low single-digit, market-neutral edge before costs, and only if you rebalance rarely enough that trading costs don’t eat it. No annualising to flatter it.

The short answer

Funding-rate farming — shorting a perpetual and hedging it with equal spot to collect funding with near-zero price exposure — has a real but small edge. Across five liquid pairs on real Binance history (2022–2026), gross funding accrual was +19.3% total, about ~4%/yr. Gross is not what you keep: after realistic two-leg costs, the honest net at a weekly rebalance was +2.8% total, Sharpe 0.5, -5.3% max drawdown. The realistic answer is a modest, market-neutral ~3–5% before costs — not the passive double-digit yield most bots advertise.

Why costs and rebalance cadence decide everything

The edge is small per period, so the biggest lever is how often you rebalance — every rebalance pays fees and slippage on both legs. The same edge, same window, three cadences:

Gross funding accrual (before costs)+19.3%~4%/yr — the edge in isolation, NOT achievable net
Net, daily rebalance-19.1%~8.7%/yr cost drag — over-trading turns the edge negative
Net, weekly rebalance (headline)+2.8%Sharpe 0.5 · max drawdown -5.3% · ~2.1%/yr drag
Net, biweekly rebalance+4.6%Sharpe 0.83 · ~1.1%/yr drag — less trading, more kept

Read plainly

The gross funding number was the same in every row. Cadence and costs, not the headline rate, decide whether funding farming is profitable. And funding can turn negative when the market flips net-short, so even a patient cadence isn’t risk-free.

Why most bots hide this

+19.3% / ~4%/yr annualised into “passive crypto yield” sells far better than an honest +2.8% net. Most bots quote the gross accrual, skip the two-leg costs, and never show the daily-rebalance case that goes negative. Fygga shows the edge and the costs — including the strategies we tested and rejected. New to the mechanics? Start with what funding-rate capture is, or watch the live funding monitor.

Frequently asked questions

Is funding-rate farming profitable?
Sometimes, modestly, and only when costs are respected. In Fygga's committed backtest (2022–2026, five liquid pairs on real Binance history), the gross funding accrual was +19.3% total — about 4% a year — which proves a real market-neutral edge exists. But after realistic two-leg trading costs, the honest net at a weekly rebalance was +2.8% total with a Sharpe of 0.5 and a -5.3% max drawdown. It is a low-drama, single-digit edge, not passive high yield. Backtest only — past performance is not indicative of future results.
Why does rebalance cadence matter so much?
Because the funding edge is small per period, every rebalance you pay for eats into it. In the same backtest, rebalancing daily turned the identical edge into -19.1% (a ~8.7%/yr cost drag), a weekly cadence netted +2.8% (~2.1%/yr drag), and a biweekly cadence netted +4.6% at a Sharpe of 0.83 (~1.1%/yr drag). The gross edge never changed — only how often it was charged trading costs. Cadence, not the headline funding number, decides whether anything survives.
Why do most funding-farming bots look more profitable than this?
Most quote the gross funding accrual — the pre-cost number — and annualise it to look like passive yield. They rarely subtract fees and slippage on both legs, and rarely show what over-trading does. A gross +19.3% headline sells better than an honest +2.8% net. Fygga publishes both, plus the daily-rebalance case that goes negative, so the costs are visible rather than hidden.
What are the risks even if it is profitable?
Funding can flip negative when the market turns net-short, so the hedged position pays instead of earns. Trading costs on both legs can erase the edge if you rebalance too often. There is execution and basis risk between the perp and spot legs, exchange and counterparty risk, and imperfect-hedge risk. Removing price direction removes one risk, not all of them. This is educational information, not financial advice.

Related reading

The rest of the funding-rate explainers, in plain English.

Follow the research