ℹ️ How the app calculates

One number from 0 to 100. The higher it is, the cheaper Bitcoin is against its own history. The score stands on two pillars.

• Valuation: how far price is from its long term trend. The scale adapts as the market matures and its tops get weaker.
• On-chain data: how holders behave, meaning who buys, who sells and at what profit.

Sentiment and ETF flows are shown for information only. They don't count toward the score because they didn't improve results in testing.

Zones are the same everywhere in the app: 0 to 20 Strong Sell, 20 to 40 Sell, 40 to 60 Hodl, 60 to 80 Buy, 80 to 100 Strong Buy.

Several cycle top indicators watch for an overheated market. When more of them fire at once, the score caps itself. The honest record since 2017: they fired together five times. Twice at a top (Nov 2017 and Jul 2025), twice in the middle of a rally and once after the top. A benefit for DCA results hasn't been proven, so we show the warning mainly as information.

The probability that price will be higher than today in 3 months and in a year, and the risk of a 30% drop within six months. The model only learned from years that have already ended.

• Designs were tuned only on 2016 to 2021. The years 2022 to 2026 were locked and evaluated once, after the designs were frozen.
• On the locked years: Brier skill score 0.134 for 3 months, 0.415 for a year and 0.364 for drop risk (0 = no better than the base rate).
• One year hit rate (starts 2022 to Sep 2025): after a bullish label price was higher in 96% of cases, after a neutral one only in 23%. For 3 months 70%, 40% and after a bearish label 37%.
• Drop risk: in 2022 to 2026 a 30% drop followed elevated risk in 45% of cases and almost never followed low risk. In 2016 to 2021, though, it followed low risk 24% of the time, so low risk doesn't mean safe.

Today's cycle against past ones, aligned on the capitulation point. It shows what followed on the same day in past cycles. There are only four cycles, so this is a comparison, not a tested probability.

The score for every historical day is computed from data known up to that day. Time travel in the header therefore shows the app as it would have looked in the past.

An honest caveat:we chose the shape of the model on the same history we measure results on. Part of the hit rate may therefore look better than it will in the future. That's why we also track live operation since July 2026.

Every change has a date and an impact on results. After a change the whole history is recomputed, so the backtest and live operation calculate the same thing.

The buy amount grows with how rarely cheap today is against history. You choose the cap for a single buy yourself (1.5 to 5× the average) and over the long run the strategy targets your average amount. On cheap days it buys even more than has come in so far and takes the difference from a reserve.

Over the window Jul 17, 2017 to Sep 28, 2026 (a buy every Monday, default cap 3×) it bought 30% more bitcoin per dollar spent than plain DCA and ended 30% higher. It spent 100% of deposits and in the best periods went up to 55 weekly deposits ahead of them.

Recommended horizon: at least 1 year, ideally 2 years or more. 74% of one year and 89% of two year periods of the strategy ended in profit, against 67% and 75% for plain DCA. It beat plain DCA in 74% of one year, 94% of two year and 100% of three year periods. The periods overlap, so this covers roughly two cycles.

Same deposits, the reserve counts. Both strategies get the same deposits. Any reserve used counts as money deposited, with no interest, so the strategy gets nothing for free. With the score randomly shifted in time, the same rule comes out around zero. The edge comes from timing, not from putting money in earlier.

• Valuation: calculations in USD, transactions in EUR and CZK at the official rate on the day of purchase. The currency switch only changes the display, percentages stay the same.
• Profit and loss: average buy price, realized profit by the average price of the amount sold.
• Taxes (Czech rules): FIFO, 3 year holding test, CZK 100,000 annual value test. Indicative only, not tax advice.
• Comparison: your actual deposits mirrored into 100% BTC and into the S&P 500.
• Forecast: a range from historical returns. A model, not a prophecy.

The app reads aggregated on-chain metrics, the daily Bitcoin price, a market mood index, spot ETF flows, a stock index and FX rates. Your portfolio stays in your browser only. Your account and strategy settings are stored on the server.

• Backtests are a historical simulation. Past performance doesn't guarantee future results.
• Bitcoin has about four cycles behind it, and rare events like tops are few.
• Popular indicators that didn't improve results are shown for information only.
• The Live operation card on the Strategy page keeps showing how the model performs live.
• Bull or Bear largely measures the position in the cycle. A plain calendar from the halving was more accurate for 3 months in 2022 to 2026. The app deliberately doesn't use it.
• The Bull or Bear gauge didn't catch the tops.