Start with the ugly year
In 2022, ETHUSDT returned 7.873% for the full year, amplified — 0.787% before principal scaling.
The same table shows 375.621% for the same pair in 2021. A factor of nearly 48 between two rows built from the same collector, the same pair and the same strategy.
That row is printed on the front page, at the same size as the good ones. Not as a gesture — because a table containing only good years does not get believed. A reader who cannot see a bad year has no way to tell whether bad years were hidden or simply did not happen, and the rational response is to discount everything on the page, including the parts that are true.
Why the gap is that wide
Funding is a transfer between traders. Its size tracks how far the contract price has drifted from spot, and that gap is a direct readout of positioning.
2021 was a leveraged bull market. Everyone wanted to be long, the contract traded at a premium for months at a time, and positive settlements were frequent and rich. The short side — the side this strategy sits on — collected without doing anything clever.
2022 was a bear market. The crowd wanting leveraged long exposure thinned out, the premium compressed and at times inverted, positive settlements became sparse, and negative funding appeared in stretches. During negative funding the short side is the one paying, and the correct action is to be out of the position entirely.
So returns from this strategy correlate with market enthusiasm, even though the strategy itself takes no directional view. Those are two different statements and they are compatible.
Does the average still mean anything
Yes, read correctly.
The figure on the performance page is 124.296% amplified, and it has to be read together with its basis: it covers only the five complete, period-verifiable years from 2021 to 2025, averaged across ten cells — five years times two pairs.
The two excluded years were not dropped for being ugly. Quite the opposite: 2020 is a carried-over total from the old dashboard with an unknown settlement count, so it cannot be reconciled period by period; 2026 has run only 711 settlements and is not finished. 2020 is a strong year (187.646% and 315.810% amplified) and it still does not enter the average — a good year excluded on principle says more about the table’s rules than a bad year included for show. Both remain printed; they just do not feed that number.
One more thing before the figure can be read at all: 124.296% is the amplified column. Funding itself averaged 12.43% over the same span. The tenfold difference is interest-free principal scaling, not yield. Treating the amplified figure as the rate itself overstates earning power by an order of magnitude — and the same multiple scales costs, volatility and forced-reduction risk, each of which is itemised in the risk breakdown.
More useful than the average is seeing how far the ten cells spread:
| Year | BTCUSDT (amplified) | ETHUSDT (amplified) |
|---|---|---|
| 2021 | 306.352% | 375.621% |
| 2022 | 41.649% | 7.873% |
| 2023 | 78.421% | 82.379% |
| 2024 | 119.805% | 130.185% |
| 2025 | 51.327% | 49.352% |
There is no clean “triple digits in a bull market, double digits in a bear market” ladder here: 2024 is triple digits in both cells, 2025 falls back to around fifty, and in 2022 the two pairs differ from each other by a factor of five. Funding tracks sentiment, and sentiment does not queue up by calendar year.
If you are planning around annual cash flow, budget against the bottom band and treat anything above it as a surprise. Projecting take-home from 124% is a separate exercise — the calculator piece walks through the five steps that sit between a gross annualised figure and what actually reaches you.
What did not change
Returns varied by a factor of nearly 48. The risk structure did not: across the whole span the position carried no directional exposure.
2022 earned little. It also did not lose principal while ETH fell from 4,800 to 900 — a year in which simply holding the asset cost about eighty percent.
That is the accurate content of the phrase “market-neutral”: it removes directional price risk, not variance in returns. The two get conflated constantly, and the cost of conflating them is either using a bear year’s returns to doubt a bull year’s data, or using a bull year’s returns to plan a bear year’s budget. What exactly delta neutral is neutral to is the mechanism behind that sentence.
Where the funding itself comes from in the first place is the basics piece. The full year-by-year table, including the sample count behind each figure, is on the performance page. What remains after price risk is removed is listed on risk and boundaries.
The above is historical back-test data. It does not indicate future returns and is not investment advice.