Green Crypto Weeks Mostly Fail to Repeat: 35.4% Follow-Through Rate

By Olivia Hart

After a green week, crypto's 50 most-traded perpetual contracts on Bybit turned green again just 35.4% of the time. After a red week, the next week was green 45.7% of the time. SignalPilotLabs, the trading analysis account that ran the study, measured 2,648 week-to-week pairs and found that momentum, in this sample, ran backwards.

What exactly was measured?

The study covers daily closes for the 50 most-traded perpetual contracts on Bybit over the 54 weeks ending 11 October 2026. A week runs Monday to Sunday. It counts as green when it closes above the prior week's close. That produced 2,648 pairs of one week and the following week.

Across all weeks in the sample, 42.1% were green — a year in which most of the coins fell. Of the 50 coins, 42 were green less often after a green week than after a red one. In plain terms, four out of five coins showed the same reversal lean.

Is the pattern statistically meaningful?

SignalPilotLabs flags its own key caveat: coins move together, so most of a week's color belongs to the whole market, and 42 coins do not amount to 42 separate tests. To account for this, the analysts shuffled the order of the weeks 10,000 times, applying the same shuffled order to every coin.

A gap as large as the observed 10-percentage-point spread appeared in 8.9% of the shuffles. The study's own verdict: "That is a lean, not proof." For context, a result would typically need to fall below a 5% frequency to clear conventional significance thresholds. This one does not.

Did the lean hold across both halves of the year?

The analysts split the 2,648 pairs at the start of April 2026. They concede the split came after the headline results were known, making it "a check rather than a test."

The lean appeared in both halves:

• First half: 28.6% green-after-green versus 36.3% green-after-red
• Second half: 39.9% green-after-green versus 57.6% green-after-red

The gap widened in the second half, from 7.7 points to 17.7 points. A widening gap cuts both ways: it strengthens the visual pattern while raising the question of whether one regime shift, not a stable tendency, drove it.

What did the researchers change before running the numbers?

The original plan called for 59 weeks starting in late August 2025. One coin, AKE, only began trading at the end of September 2025. The analysts shortened the window to the 54 weeks that every coin shared — a decision made, they note, before computing anything. That pre-registration-style discipline is worth crediting; it removes one common avenue for cherry-picking.

What does the chart show?

The accompanying chart tracks Bitcoin weekly from 7 July 2025 to 11 October 2026 on a logarithmic scale, where equal vertical steps represent equal percentage moves. Weeks before 29 September 2025 were not part of the measurement.

Each green Bitcoin week connects to the next via a shaded span: teal when the following week also closed green — which happened 10 times — and red when it did not, which happened 15 times. So Bitcoin itself followed a green week with another green week only 40% of the time in this window, slightly better than the 50-coin average but still below the red-week baseline. A side panel breaks down how the 50 coins' subsequent weeks split after green weeks versus red ones.

What the study does not tell you

The analysis is explicit about its limits: it says nothing about what next week will do. It covers one year of weekly data, on one exchange, during a year when most of the sampled coins declined. SignalPilotLabs labels the work "Observations, not recommendations."

That disclaimer deserves to be taken at face value. A 10-point spread that appears in 8.9% of random shuffles falls short of statistical significance, the sample spans a single bearish regime, and the analysts themselves downgraded their split-sample check to a post-hoc exercise. Traders looking for an edge here have roughly one year of falling prices on one venue — enough to justify watching the next cycle, not enough to trade on.