Price Prediction Methodology
How our price forecasts work
CoinDataFlow’s main price forecast uses historical price patterns to construct a possible future path. It is based on the idea that some features of past market cycles may recur.
We examine the sequence of past rises and falls and their size in percentage terms. Selected patterns are compared on a common scale and combined into one pattern. Starting from a recent observed price and the estimated current stage of the cycle, the model applies this pattern to build a sequence of projected prices.
This page explains the historical-pattern forecast. Technical indicators and other comparison tools shown alongside it are separate analytical views. Historical patterns do not guarantee that a cycle will repeat.
How a projection takes shape
The animation follows the journey from sections of price history to a forecast built from their recurring movements.
From historical movements to a forecast
- 01SelectPatterns within price history
- 02ExtractSelected historical movements
- 03CompareChanges on a common scale
- 04CombineOne combined pattern
- 05ProjectStart from an observed price
The estimated current stage of the cycle selects the relevant part of the combined pattern. The forecast then starts from the latest price used in the calculation.
How to read the forecast
The projected line shows a possible sequence of future prices. Period summaries describe values along that path; a range around the line shows a different kind of variation.
| Displayed value | How to interpret it |
|---|---|
| Price on a date | The projected value at that point on the path, built from the starting price and the movement pattern. |
| Period Min / Max | The lowest and highest projected values reported for the named period. These labels alone do not describe a statistical confidence interval. |
| Average price | The midpoint between the period’s projected minimum and maximum. It summarises that range; it is not the average of all daily projected prices or a year-end target. |
| Scenario range | Where shown, variation based on historical observations and model assumptions. It has no specified probability of containing future prices unless explicitly stated. |
A range is not a price floor or ceiling
- Scenario range
- Projected path
- Possible outcome
Why a forecast changes
Each forecast begins from a price observed at a particular time. New observations can change that starting point; updates to historical data or model assumptions can also change the projected path. Today’s calculation is not a record of what the model predicted in the past.
A forecast is a snapshot
- Illustrative observations
- Earlier projection
- Updated projection
Historical market data comes from third-party sources. Coverage, delivery and processing times can differ across assets. Forecast values are calculated estimates, not real-time quotations. The date at the top of this page describes this document, not the timestamp of an individual forecast.
What the model assumes - and what can change
Historical resemblance provides a reference, not a fixed schedule for the market. The projection depends on assumptions about the timing, scale and persistence of historical movements.
- Cycles can change
Similar phases may arrive earlier or later, become weaker or fail to recur.
- Markets respond to new events
Economic conditions, regulation, company developments and market shocks can move prices away from the projected path.
- History is a limited sample
Available observations cannot represent every condition an asset may face.
- Distant values depend more on assumptions
Extending a trajectory further into the future does not add new evidence about what will happen.
How forecast quality should be assessed
A good fit to known history does not establish forecasting accuracy. A meaningful evaluation separates model development from the outcomes used to judge it, uses only information available at each forecast start, and compares errors at clearly stated horizons.
What to look for in a published evaluation
- Whether the forecast was reconstructed from past data or recorded before the outcome occurred.
- The dates, assets, forecast horizons and number of observations tested.
- Which data and model settings were available when each forecast began.
- The error measure and a benchmark, such as an unchanged-price forecast.
- For a range, both its width and how often observed prices fell inside it.
This page describes evaluation principles; it does not report a validated accuracy score.


