We claimed 'the first 5 days of the month are gold' — then debunked our own claim

During testing we noticed something striking: trades were clearly better in the first 5 days of the month (Sharpe 1.75 — above the overall average, with roughly half the drawdown). It was tempting to say "we found an edge!"

But robustness first

Before getting excited, the question we always ask: is this consistent year over year?

• In 2023 the start-of-month effect was negative (Sharpe −1.01)
• And whichever 5-day window we looked at, one of them came out "amazing" — Sharpe +2.49 for days 16–20, −3.47 for days 11–15

This is a classic data-snooping signature: try enough slices and one shines by chance. Because crypto moves in trends, a fake pattern can stick to any calendar label.

Our decision: rejected. Most sellers would have marketed that out-of-sample 1.75 as an "edge." We tested it before trusting our own finding, it turned out fake, and we did not add it.

This is what we mean by "honest": we test ourselves before we sell to you.