What happened to Bitcoin after similar charts?

After charts that looked like today's, Bitcoin went up about as often as it went down. In 300 tests of Binance BTCUSDT hourly charts from January 1, 2020, to September 29, 2026, the ten closest past charts were closely split after 24 hours in 65.00% of tests: four to six of them ended higher. Their middle move pointed the right way in just 50.67% of tests, close to a coin flip.

For the same test across 12 markets, see Similar crypto charts, different outcomes.

What counts as a similar Bitcoin chart?

A bar records prices during a set time. An hourly closing price is the price at the end of that hour. We built each test chart from 48 completed hourly closing prices.

A chart analog is a past chart with a similar shape. We put price changes on a common scale, so charts could match even when their dollar prices or move sizes differed.

For each test chart, we found the ten closest past windows that did not overlap each other. Bitcoin on Binance matched only its own Binance history. Bitcoin on Coinbase matched only its own Coinbase history. Each past case and its later outcome were already known at the time of the test chart.

The markets differ too. Binance BTCUSDT here is a perpetual futures contract: a contract tied to Bitcoin's price that never expires. Its price is in USDT, a token designed to track the US dollar. Coinbase BTC-USD is spot Bitcoin, meaning Bitcoin traded directly for dollars. See why Bitcoin prices differ between exchanges for more on these differences.

How often did the closest past charts disagree?

Each row below covers 300 test charts on one exchange. A “close split” means four, five or six of the ten past cases rose over the stated time. To find the middle move, called the median, we sorted the moves and averaged the middle pair. “Direction matched” means that middle move pointed the same way as Bitcoin's actual later move. “Test charts up” is the share of test charts followed by a rise.

How often did the closest past charts disagree?

Bitcoin market	Time after chart	Close split among past cases	Direction matched	Test charts up
Binance BTCUSDT	12 hours	65.00%	51.33%	52.67%
Binance BTCUSDT	24 hours	65.00%	50.67%	49.33%
Binance BTCUSDT	48 hours	67.00%	52.33%	52.33%
Coinbase BTC-USD	12 hours	64.33%	47.67%	53.00%
Coinbase BTC-USD	24 hours	64.33%	47.33%	49.00%
Coinbase BTC-USD	48 hours	60.67%	52.67%	52.00%

Close splits were more common than other results at every time shown. Looking further ahead did not make the Binance matches agree more. Coinbase had fewer close splits at the longest time, but that alone does not show that the matches could predict the next move.

A split also says nothing about move size. Similar counts of rises and falls can hide large moves in both directions. Comparand’s shape_analogs search lets you read each matched path separately. Our guide to why past crypto cases disagree explains how to read mixed results.

Did similar charts beat a simple direction check?

We also checked what happened on those same test dates without using chart matches. Using the same dates keeps the comparison fair.

In the Binance day-later row, Bitcoin fell slightly more often than it rose. Calling every move down would have matched direction just as often as using the middle past move. We can see that after looking at the results; it was not known before the test.

For Coinbase's day-later row, using the middle move did worse than calling every move up or every move down. At the longest time, it did only slightly better than calling every move up. The results show no clear, steady advantage from using the middle past move. Our test across 12 crypto markets found the same thing.

Results also changed by year. In the Binance Bitcoin test using hourly bars, the middle move matched the actual 48-hour direction in 72.73% of 44 cases in 2023. It matched in 36.36% of 44 cases in 2024. Reporting only the better year would hide that change.

The share of past matches that rose is not today's chance of a rise. We chose those cases because their shapes looked similar. The matches also change a lot when you change the chart length; see why chart length matters.

How wide were Bitcoin's usual price moves?

We also measured Binance BTCUSDT hourly closing prices from September 8, 2019, through the October 20, 2026, cutoff. Here we used every hour with a complete price record through the later endpoint. This is Bitcoin's base rate: what happened without first choosing similar charts.

“Share up” counts only rises; an unchanged price does not count as up.

The low and high cutoffs are the 10th and 90th percentiles. Roughly a tenth of moves fell below the low cutoff, and a tenth rose above the high cutoff. These are not the worst and best moves.

How wide were Bitcoin's usual price moves?

Time after hourly close	Cases	Share up	Middle move	Low cutoff	High cutoff
12 hours	62,347	51.30%	+0.03%	-2.09%	+2.28%
24 hours	62,335	51.88%	+0.07%	-3.07%	+3.49%
48 hours	62,311	52.23%	+0.15%	-4.33%	+5.06%

The middle move was near zero, but the cutoffs show much larger gains and losses. A small middle move did not mean prices usually stayed near their starting point. The longer windows had wider results on both sides.

These are ordinary hours, not the selected chart matches, so this table does not show whether matches had a narrower range of moves.

A takeaway for your own research: before you read a set of past cases, write down this normal range for the same wait. For one day on Binance, a tenth of ordinary moves fell below -3.07% and a tenth rose above +3.49%. If most of your past cases sit inside that range, they look like ordinary days, not a special setup.

What happened after fees?

For the same 300 Bitcoin tests per exchange from January 1, 2020, through September 29, 2026, we tested imaginary trades based on the middle move. A positive middle move meant a trade that gained if Bitcoin rose. A negative middle move meant a short: a trade that gained if Bitcoin fell.

We assumed a fee of 0.05% to enter and 0.05% to exit each trade. Binance's average results were -0.15%, -0.30% and -0.07% over 12, 24 and 48 hours. Coinbase's were -0.05%, -0.25% and +0.03%, in the same order. The small positive result belongs alongside the losses.

These figures describe separate imaginary trades, not growth in an account. They leave out borrowing costs (Coinbase shorts would need borrowed Bitcoin), payments between perp traders, delays, the gap between buy and sell prices, and getting a worse price than expected.

How we measured this

We used public hourly Bitcoin prices from Binance and Coinbase and kept each market separate. At dates spread across the test period, we compared charts built from 48 completed closing prices. We kept ten earlier matches whose later moves were already known, then checked their middle direction against Bitcoin's next move. The imaginary trades include 0.05% fees per side but leave out the other costs described above. The base-rate table uses a longer history through October 20, 2026, and both tests exclude windows with missing prices. Some later paths share hours, and we did not show that small differences were more than chance; see how the searches work.

Try this in Comparand

With Comparand connected to your AI app, ask: “Find the ten closest past charts for Binance Bitcoin, BTCUSDT, using the latest 48 completed hourly bars. Show each case's price path and what happened 12, 24 and 48 hours later.” This uses the chart search called shape_analogs.

The search returns:

• Past Bitcoin cases from Binance only, using the same hourly bar size.
• One match per past episode, with no match overlapping the chart you searched.
• Each case's own price path and later price changes at 12, 24 and 48 hours for this chart length.

The price changes come in basis points: 100 basis points equals 1%. Compare the same later time across cases and include the falling cases too. The search does not return an average, middle move or success rate; the tables here come from our separate test. These are past cases, not a forecast.

A shape_analogs search costs $0.01, paid from credits. For a free first look, open the page to try 3 coins free. For the AI connection steps, read the connection and search docs.

The short version

• Similar Bitcoin charts often had mixed later outcomes on both exchanges.
• Their middle move showed no clear, steady advantage in this test.
• Bitcoin's usual moves included gains and losses far larger than the middle move.
• Each past path adds detail that a single summary can hide.