SIAT turns forty, and our research is among the three finalists

The Italian Society of Technical Analysis was founded in Milan in 1986 as a non-profit association, to promote knowledge of the technical analysis of markets by developing research and training. In 2026 it turns forty — forty years in which the school of technical analysis has had, in Italy, a place where it could be taught and argued about.

At the SIAT Award 2026 our work — “A Falsificationist Cycle Analysis of Cryptocurrencies” — is one of the three finalists.

What is in the running

Not a forecast, and not a new method: a falsification protocol.

For more than half a century the cycle analysis of markets has produced rich conceptual frameworks, and has handed them down almost always as alternative doctrines: the classical American school of J. M. Hurst, the Italian school of the battleplan and the inverse cycle, John Ehlers' quantitative analysis in the frequency domain. They get adopted; they rarely get tested together on the same sample.

The work takes each founding claim of the three schools and holds it to the same yardstick, on a multi-asset sample: Bitcoin and Ethereum since 2017 and Solana since 2020, extended to thirty-seven instruments in six classes and to six stock indices with long daily history.

The part that outlives the results

These are the four requirements the protocol imposes, and they are worth more than any single outcome:

• the null value of a measure is measured, not assumed. Surrogate series are built that preserve the power spectrum and destroy the temporal structure, and one looks at where the observed value falls inside that cloud;
• a conditional rate must be accompanied by its own marginal, that is, by the same condition counted on the cases the event does not select. Without it, a high number only says that the condition is frequent;
• a negative outcome does not count until it has been shown that the test would recognize the phenomenon if it were there. A blind test falsifies nothing: it is silent;
• every level is measured on the timeframe that resolves it. The daily timeframe does not separate the short levels of a hierarchy, and measuring them there changes the very numbers under discussion.

The four are not a list of criteria to choose from: they are a chain. An assertion has to clear all of them, and the first one it fails stops it — which is why some widely quoted claims fall without any new data being needed.

What falls, and what holds

Some of the most quoted claims fall: fixed harmonic nesting and the principle of harmonicity, the sequential mean-reversion of durations, the canonical threshold for connecting cycles, the peak hold. Two claims of the Italian school fall by tautology — the inverse cycle, whose operational definition is not falsifiable before the fact, and the unconditional swing, whose success condition turns out to be true in every measured parent cycle, with and without a swing.

What holds, with convergence across asset classes: the cyclic commonality of durations, cyclical translation as an indicator of polarity, the violation of the previous peak, and, in a rewritten form, two signals of the Italian corpus.

And the most uncomfortable part

The same yardstick applies to the regularities the work could claim as its own, and there it is harsher than with other people's schools.

The short-cycle tail does not leave the surrogate cloud in any of the fifteen combinations of asset and level, and at the intermediate levels the observed value lies below the median: it is what the detector produces, not what the market does. The same holds for the deviation from the nominal period and for the position of the peak inside the cycle.

The chart above is that deviation, measured: thirteen levels across three instruments, each set against the duration the catalogue assigns to it. The regularity is plain to see — and that is exactly why it is worth saying what it is not: a property of the market.

The work's own starting axiom also remains unproven — the one by which a quaternary structure would be a binary of binaries: the trough that would separate the two binaries is the lowest of the three internal ones in 1.3 per cent of the 557 measured cycles, against 33.3 per cent for pure indifference — and inside the band that the same detector produces on noise. Tested on a purpose-built binary of binaries, the detector does not let it through to the count: the test cannot decide.

Why we publish it this way

Because it is the only form in which a nomination is worth telling. The recognition, if it comes, is for a protocol that refuted its own starting hypotheses before refuting anyone else's — and that is the part which stays useful to a reader, regardless of how it ends.

The paper is published on this site, on the preprint page: the PDF can be downloaded without registering, and the evidence files for verification are available with an account. The data, the period and the method are in the card alongside, as on every note of ours.