← Research · Replication · Market microstructure

Does one crypto lead another by a few minutes?

Not in our data. A 2026 paper reported Solana leading Cardano by four minutes with 92% directional agreement; on the paper’s own window we found that agreement in the same minute (0.93) and almost none at a four-minute lag (0.006), and shifting the Cardano series by four minutes by hand reproduces the paper’s figure. Across hundreds of coins and every year from 2021 to 2026, no lag of 1–10 minutes carried a tradeable lead.

Did not replicateCrypto perpetual futures, one-minute, 13–16 Jan 2026 and 2021 – 2026Last reviewed

The claim

Source we re-testTasci, M. (2026). TAM: ternary agreement model — a novel approach to lead-lag analysis in cryptocurrency markets. PeerJ Computer Science, 12, e3810. doi:10.7717/peerj-cs.3810

Tasci turns one-minute returns into up, flat and down states and measures how often a leader’s state agrees with a follower’s at different lags. Tested on series with lags inserted by hand, the method recovers them almost perfectly; on real data the paper names Solana a reference leader for the market and reports SOL leading ADA by four minutes with 91.9% agreement.

What we tested

We rebuilt the measure from the paper and the author’s code and ran it on one-minute perpetual futures over the paper’s window (13–16 January 2026), then on SOL against hundreds of other coins, and on SOL–ADA year by year from 2021 to 2026. For a tradeable test we built hourly portfolios of followers from the leaders’ last minutes, entered one minute after the signal, with costs.

Result

On the paper’s window, SOL–ADA agreement is 0.93 at lag zero and +0.006 at four minutes, and every lag from one to ten minutes stays below 0.07. Shifting the ADA series by four minutes by hand moves the 0.93 to lag four: the paper’s figure is same-minute agreement relabelled as a lead. Across all coins the strongest agreement at lags of 1–10 minutes was 0.23, in illiquid names at one minute, and SOL–ADA at four minutes ranged from −0.013 to +0.006 by year. Leader-follower portfolios broke even only at costs below 0.001% per trade.

Directional agreement between SOL and ADA (1 = always agree)

MeasureValue
Same minute (lag 0)0.93
Four-minute lag (paper: 0.92)+0.006
Best lag of 1–10 minutes, any coin0.23 (illiquid, 1 min)
Break-even cost of a leader-follower portfolio< 0.001% per trade

What it means

Crypto coins move together within the same minute, and small, illiquid coins catch up within seconds, not minutes. A lead measured on shifted data, or scored at the wrong lag, looks like a perfect signal.

What would change the verdict

A sub-minute lead might exist on second-level data with own execution. A lag of two minutes or more that survives the hand-shift test would change the verdict.

This note re-tests a claim made by the cited authors on our own crypto market data; the summary of their method is ours. Tested · published · last reviewed . Research, not investment advice.