The SIAT Award 2026 gives first prize to the falsificationist protocol

Three days ago we wrote that “A Falsificationist Cycle Analysis of Cryptocurrencies” was one of the three finalist works for the SIAT Award 2026, and that the recognition, if it came, would be about a protocol and not a forecast. It came: in SIAT's fortieth year, the prize in the Open category went to our work.

It is worth saying straight away what this is not. A prize is not an independent verification and does not confirm a result: the published verdicts stand as they are, with the same asymmetries and the same stated limits. What was assessed is the way those verdicts were reached.

What was assessed

The official reasoning will be published by SIAT, and we are not anticipating it. What we can say from here is what the work demands of itself: four requirements of the protocol, which are worth more than any of its outcomes.

The first is the validation protocol itself. The work does not apply the historical models — Hurst, the Italian school, Ehlers — it puts them to the test: every founding claim goes through a strict null hypothesis, through surrogate series, and through a prefix-invariance test that prevents a past value from depending on future data.

The second is the marginal, kept next to every conditional rate. A conditional rate without its own marginal on the control group says almost nothing: it says the condition is frequent, not that it selects. Applied methodically, this requirement exposed the tautological nature of concepts accepted for decades without ever being counted — the unconditional swing is the clearest case.

The third is the question of the ruler: measuring a cycle on the wrong timeframe distorts its data. Assigning each analytical level to the timeframe that mathematically resolves it is not a technical detail; it is what dissolves some long-standing conflicts in the discipline without adding any theory.

The fourth is the severity applied to the author's own hypotheses, and it is the reason the candidacy was described as it was. Not all of them hold, and the two that do not hold fall in different ways: the bimodality of durations, announced in the first version, was withdrawn by those who had announced it; the tail of short cycles does not emerge from the surrogate cloud in any of the fifteen asset-and-level combinations. Neither of the two is the market, and saying so was worth more than keeping them.

What is left to do

The same things as three days ago, in the same order.

The extension to lower timeframes, where the sample of four-beat structures becomes large enough to rule on what the daily leaves without a verdict. The extension to equity and currency markets, where the scale of levels has different institutional origins and the protocol has to be redone, not carried over. And independent verification, which is the only thing that can do to a result what a prize does not: the dataset and the verdicts in JSON format are there for that.

The synthesis, the replication materials and the peer review are in the publications, for those with an account. The data, the period and the method are in the panel alongside, as on every note of ours.