A price prediction is not a forecast you should trade on directly. It is a structured opinion that depends on assumptions — usually undisclosed. Reading a prediction well means understanding what produced it, what range of outcomes it represents, and what would invalidate it.
The three categories of predictions
- Analyst forecasts — a human reads news, looks at charts, picks a number. Highly variable quality. Often anchored on the analyst’s prior views. Almost never tracked for accuracy.
- Model output — a deterministic algorithm processes inputs and produces a number. Quality depends on whether the inputs are causal and the algorithm is sound. Should be auditable.
- Market-implied — derived from options prices, futures basis, or other market data. The most honest read of “what the market thinks,” but reflects current positioning, not necessarily fair value.
TheWeal predictions are model output. Our model is published in full at /methodology/.
Bear / Base / Bull is more honest than a single number
A prediction that says “BTC will hit $150k in 12 months” is wrong before you finish reading it — even if BTC reaches $150k. The actual outcome will not be exactly $150k. The useful question is: what is the range of plausible outcomes?
Bear/base/bull formats answer that. The base is the central estimate; the bear and bull bands give you the model’s uncertainty. If the bear-bull range is narrow, the model is confident; if it is wide, the model is hedging.
Horizon scaling
A 24-hour prediction and a 5-year prediction are fundamentally different objects. The 24-hour prediction depends on near-term momentum and is testable next week. The 5-year prediction depends on macroeconomic, technological, and regulatory variables that no model can reliably forecast.
Wider bands at longer horizons reflect compounding uncertainty. A model that gives identical confidence at 24h and 5y is overstating its long-term certainty.
What invalidates a prediction
For each prediction, ask: what would have to happen for this to be wrong?
- Momentum-driven predictions assume momentum continues. They break when momentum reverses (regime change).
- Mean-reversion predictions assume the asset behaves like its history. They break when the asset is a fundamentally different thing (Ethereum post-Merge, BTC post-ETF).
- Sentiment overlays assume current sentiment is informative. They break when sentiment becomes uncorrelated with subsequent prices.
A good prediction tells you what would invalidate it. A bad prediction is vague about its own assumptions.
Three questions to ask any prediction
- How was it produced? Hand-wave, model, market-implied?
- What is the bear case? If the bear is not stated, the prediction is incomplete.
- What is the historical accuracy? Most prediction makers do not publish their track records. The ones that do are more credible by definition.
How to use a prediction
Use predictions as one input among many. Cross-reference with:
- Your own analysis of fundamentals
- The news flow around the asset
- Market structure (ETF flows, futures basis, options skew)
- Your own time horizon and risk tolerance
Never make a position-sizing decision based on a single prediction. Use them to inform your thinking, not to replace it.
What predictions are not
- Guarantees
- Trading signals
- Personal recommendations
- A substitute for understanding the asset
Where to go next
- TheWeal prediction methodology — full breakdown
- How to read a price chart
- See a prediction in action: Bitcoin