Can you spot trading bots on the Bitcoin chart? The chart shows the time, not the culprit

On 23 February 2026, the administrator of Terraform Labs sued Jane Street over trades around the 2022 Terra collapse. Two days later, Glassnode co-founders Jan Happel and Yann Allemann wrote: "Jane street Lawsuit gets made public, and miraculously the 10am $btc slam disappears." Zerohedge posted a chart with the caption "would you look at that".

The idea may look familiar from your own chart: a big red candle at the same time every day. That has to be a machine, and once you can see the machine, you can trade against it.

Two years of 5-minute candles for Bitcoin, Ether and Solana give a clear answer. The chart does show a pattern: the clock of the New York stock exchange. It does not show a bot.

The opening hour really was the weakest of the day

First, the part that supports the claim. Between 3 November 2025 and 20 February 2026 there were 75 US trading days. In the hour after the open, 9:30 to 10:30 New York time, Bitcoin lost 0.19 percent on average. No other hour of the day was weaker.

Add up the 75 opening hours and you get minus 14.4 percent. Over the same stretch, Bitcoin fell from $107,900 to $67,800. On weekends and US holidays, when the exchange is closed, the same time of day averaged minus 0.01 percent.

The red morning candle that winter was real. Whether a single seller was behind it, striking every day, is a different matter.

Two to three percent a day did not happen

The claim had three parts: consistently, two to three percent, within minutes. The same 75 days:

On 33 of 75 days, Bitcoin rose in exactly the hour the bot was supposed to be selling. In the first 15 minutes, the price did not fall two percent on a single day. Even the average is shaky: with 75 days you cannot say for sure that the hour was negative on average at all. The statistical interval runs from minus 0.46 to plus 0.09 percent.

The consistency was in the posts, not in the candles. That still leaves the question of why this hour was weak so often.

The pattern came before the claim and returned after the lawsuit

In the year before, November 2024 to October 2025, Bitcoin rose from $69,400 to $110,200, or 59 percent. The opening hour was still the weakest of the day that year, more clearly for Ether and Solana than for Bitcoin. For Solana, the whole statistical interval lies below zero:

The weakest month in the whole two years was March 2025. Bitcoin's opening hours added up to minus 11 percent then, nine months before the first Jane Street post.

After the lawsuit, the hour turned positive on average for Bitcoin. In May and June 2026 it came back: in May it closed lower on 15 of 20 trading days, and in June it added up to minus 5.6 percent. For Ether it stayed negative after the lawsuit.

A pattern that was there before the supposed culprit and returned after the supposed retreat needs a different explanation. It sits in the same chart, on a different axis.

What the chart shows is the Wall Street clock

What matters is not which way Bitcoin moves at the open, but how far. The measure is the range, the distance between the high and the low within the hour:

At the same time of day, Bitcoin moves three to four times as much on a trading day as on a weekend. The clearest evidence comes from daylight saving time. Bitcoin trades around the clock and has no summer time. Yet the most volatile hour moves with it: 14:00 UTC in summer, 15:00 UTC in winter. It follows New York.

US economic data comes out at 8:30 New York time, the stock market opens at 9:30, and only then do the Bitcoin ETFs trade. Almost twice the volume of a typical hour trades then, and part of it is executed by algorithms run by market makers and ETF traders. When selling is already under way in a falling market, a lot of it lands in this hour. In February, analyst Alex Kruger found that the IBIT Bitcoin ETF closely tracked the Nasdaq between 10:00 and 10:30.

Compare that with the funding times of perpetual futures, every eight hours at 0:00, 8:00 and 16:00 UTC. Automated settlements definitely run there. On the chart they barely show: the first 5-minute candle after funding moves 0.071 percent at the median, other full hours 0.067 percent.

The chart tells you when a lot happens. It does not tell you who is trading.

Bots are found in account data, not in candles

The best-known bot in Bitcoin's history was real. In 2013, a program on the Mt. Gox exchange, later called "Willy", bought 10 to 19 bitcoin every few minutes, 268,132 bitcoin in total for $112 million, without paying for them. Researchers from the University of Tulsa and Tel Aviv University analysed it in 2017. Mt. Gox CEO Mark Karpelès admitted in a Japanese court that he had operated the bot.

Willy was not found on the chart. The evidence came from Mt. Gox trading logs leaked in March 2014: accounts that only bought, never sold, and even traded while the exchange was offline. On the chart, the same period looked like a bubble: Bitcoin went from about $150 to over $1,000 in two months.

The one well-documented Bitcoin bot looked like enthusiasm on the chart. Proving a bot takes accounts and orders, not candles.

Trading against the 10 a.m. dump would have cost money

Belief in a daily bot invites trading against it: short Bitcoin at 9:30 every trading day and buy it back at 10:30. Calculated with 0.05 percent fees per side, typical for futures:

The first row is hindsight: it assumes you already knew when the pattern would end. Starting after the post and simply carrying on left you at minus 19 percent by October. Even in the year before, when the opening hour was also the weakest, fees ate the edge.

The clock is still useful, just somewhere else. A stop one percent below the price sits within the normal swing of a single hour at the US open. The same order usually gets through the same time on a weekend with ease. You can plan around the clock. Betting against it does not pay.

What price data cannot prove

The figures come from Binance spot prices. Other exchanges and ETF trading itself are not calculated separately, but arbitrage keeps them close. The analysis cannot rule out that someone sold at the open. It only shows that no daily seller appears on the chart and that the pattern is not tied to the Jane Street period. The lawsuit against Jane Street concerns the 2022 Terra collapse, not the Bitcoin price at the open. Jane Street calls the lawsuit's claims "baseless, opportunistic". Periods, windows and controls were fixed before the first run.

The next time the chart shows a time of day

When someone shows you the next machine that strikes at the same time every day, three questions test it. Does it really happen on most days, or only on the days that get posted? Was the pattern there before the supposed culprit showed up? And does the price move the same way at that time on weekends?

For the Bitcoin morning, all three answers point the same way. What is left is an hour in which more happens than usual. Your stop should know that.

FAQ

Why 9:30 when the posts said 10 a.m.? The exchange opens at 9:30 New York time. Whale Factor wrote about the "cash open", Negentropic about the "10am slam". Both windows were checked. 10:00 to 10:30 was also slightly negative that winter (−0.07 % on average, down on 52 % of days) and fell two percent or more on exactly one day.

Does this mean Jane Street did nothing? No. Price data can neither prove nor rule that out. What it does show: the chart has no daily sell-off, and the pattern existed before and after the Jane Street period.

Aren't many trades automated anyway? No reliable public count for crypto could be found. And an algorithm buys as readily as it sells: market makers quote both sides, arbitrage programs buy on one exchange what they sell on another. That is why "an algorithm did it" does not explain a price direction. The question is who trades in which direction, and that is not on the chart.

Not investment advice, not a recommendation, not a forecast — historical patterns are no guarantee.

Study the Past — Improve your Future 🥋