Insights · updated November 21, 2026
Historical analogs in crypto: how to use the past without pretending it predicts the future
"History doesn't repeat, but it rhymes." Analog analysis takes that idea seriously: find the moments in the past that looked most like today, then study the full range of what happened next.
What is a historical analog?
An analog is a past date whose market state resembles the present: a similar price shape over the last few weeks, similar momentum, similar volatility and a similar distance from recent highs. Instead of drawing one projected line, you look at several analogs at once and see how wide the outcomes were.
Why use several analogs instead of one fractal?
A single "this looks like 2019" chart is easy to cherry-pick. Using the six to ten closest matches, scored objectively, shows you the spread: how often the price was higher, the median move, and the worst case. That spread is the honest answer.
How Atlas scores similarity
- Price shape: the last 30 days of log prices, normalised to today.
- RSI(14): how stretched the recent move is.
- Volatility: 30-day standard deviation of daily returns.
- Momentum: 90-day change.
- Drawdown: distance from the 180-day high.
Matches must be at least 20 days apart so one episode is not counted twice.
Common mistakes
- Treating the median as a target. It is a description of the past, not a forecast.
- Ignoring the worst case. Position sizing should survive the bad paths, not just the average one.
- Using too little history. More data gives more independent matches.
Try it
The Atlas demo runs this analysis live for BTC, ETH and SOL on three years of daily data.
Run it yourself
Atlas finds the past moments that look most like today and shows what followed. Free, in your browser, no sign-up.
Open the live demoQuestions
Can historical analogs predict crypto prices?
No. They describe how prices behaved after similar conditions in the past. They are useful for understanding the range of outcomes and risk, not for predicting a specific price.
Historical statistics only. Past behaviour does not predict future results. Not investment advice.