Finding, not used

The compression of the premia

Four independent fee streams, measured separately, falling by the same order of magnitude on the same dates.

The question: If this market's premia are being arbitraged away, can it be seen? And when did it happen?

What was found

It's the project's most general finding and the one that best explains all the others. The basis carry, the cross-exchange spread and the variance premium are three independent measurements, in different units, made with different data. All three compressed by between 3.4 and 5.6 times between 2020-2021 and today, and all three did it at the same moment: 2021-2022. Three independent measures falling in the same proportion aren't three coincidences: it's capital arriving and competing the payments down. And since then they've been flat at a low level rather than still falling — which is exactly the shape of a market finishing its maturation. The quarterly basis of expiring futures, measured later and with a different instrument, is the fourth premium with the same profile.

0.028.056.084.0112.0100.0100.0100.02020-202125.123.447.62022-202329.517.828.12024-2026Basis carryCross-exchange spreadVariance premium
Each premium indexed to its own 2020-2021 level = 100. Three independent measures, in two different units, falling together and levelling off together.analysis/scripts/phase16b_premium_decay_2026-09-07.py

Try it yourself

Set a 2021 premium and today's, and look at the compression factor — the quantity that turned out to be the same for three independent measures.

Compression factor = 2020-2021 premium ÷ Today's premium
It has compressed by a factor of

Percentage of what it used to be:

Illustrative example numbers for practice — not real data.

How it was tested, step by step

  1. Each premium is measured separately and in its own units: the basis carry as an annual percentage, the cross-exchange spread as an annual percentage, and the variance premium in volatility points.
  2. They're grouped into three eras (2020-2021, 2022-2023, 2024-2026) and each is indexed to its own starting level. Comparing quantities in different units is only legitimate this way: what's compared is the RATE of decline, not the level.
  3. Two of the four turn out to be unreachable today not because they're small but because they're frictional. The cross-exchange spread changes sign between 130 and 153 times a year, not 52 as had been assumed without checking: at 2 basis points a leg the fees are 10-12% a year against a 3.60% gross.
PeriodBasis carryCross-exchange spreadVariance premium
2020-202123.91%19.93%+21.78 pts
2022-20236.00%4.67%+10.36 pts
2024-20267.05%3.54%+6.13 pts

Compression of 3.4x / 5.6x / 3.6x. The variance premium is the only one still clearly above zero.

What this does NOT say. One intuition that turned out false and is worth stating: alts do NOT pay more carry than Bitcoin. Measured across 53 symbols with history and liquidity, the per-symbol median is 6.35% against BTC's 11.59% — they pay roughly half, and the basket was negative in 3 of the last 5 years. Concentration is fine (the top 3 contribute 17% of 53), so the problem isn't diversification: the premium simply isn't there. And about the conclusion itself: three eras are three points, and nothing is extrapolated from a three-point pattern. What the data does support is the descriptive claim — the premia compressed and are flat — not a prediction about whether they'll come back.

Tracking

This finding has no live number: it was measured on archived data, and putting a chart here would imply continuous tracking that doesn't exist. The figures are above, with the script that produced them.

← All research