THE THEORY LAB

Every Bitcoin price theory, side by side

Turn models on and off, change their settings, and see how each one has actually performed. Model output, not a forecast.

    Modelsloading

    Switch a model on and its lines appear on both views of the graph below.

    Next: macro models, the street-targets ledger and the Consensus Curve.

    Model medianloading

    Median switches

    How it is computed

    One vote per family by default: each family's median first, then the median across families, so one idea gets one vote however many models express it. The shaded band runs from the 25th to the 75th percentile. Before the latest close it mixes history-only and forward models; after it, only forward models count. A model whose 365-day walk-forward error is over 100% leaves the median until a rebuild brings it under.

    Scorecardloading
    (loading)
    Market-impliedloading
    (loading)
    Options expiryMedian25th to 75th10th to 90th

    Futures basis

    Polymarket

    Market-implied, not a forecast.
    Model settingsloading
    (loading)
    Production costelectricity, fleet, all-in
    (loading)
    (loading)
    Fleet efficiency, Hashpoint assumption
    YearJ/TH

    Log-linear between years. No line before the first year. A miner stress line, not a floor.

    Halving-cycle overlaydampening
    Stock-to-flowrefit

    Gold parity / wealth shareshare, year, growth

    Logarithmic growth curvesten points
    Growth curve points, daily closes
    CurveDateClose, $Source

    Each curve is a straight line through its five points in log price against log time.

    Capital S-curve scenarioceiling, pace, FOMO, inflation

    The dashboard's S-curve section explains each choice. The scenario draws on both views of the graph.

    Model output, not a forecast.
    Signalsloading

    Puell multiple

    (loading)
    Daily issuance value over its 365-day mean. Under 0.5 miners earned little against the past year; over 4 they earned a lot. A signal, not in the median. Origin: David Puell, 2019 · Model output, not a forecast.

    Hash ribbons

    (loading)
    30-day against 60-day mean hashrate from Hashpoint's node. Red marks a capitulation (the 30-day falls below), green a recovery. Markers only, not in the median. Origin: Charles Edwards, 2019 · Model output, not a forecast.
    The LedgerEvery component in one table
    Switch
    Today is the latest daily close. The error is the model's 365-day walk-forward median absolute error, one figure per model. Model output, not a forecast.
    Get the HashpointFree by email, double opt-in
    How to useSwitches, the median, the scorecard, zoom and sharing

    What it shows

    Each model is one theory of what Bitcoin should be worth. Switch one on in Models and its lines appear on the graph at the bottom of the page, in both views: price and bands, and the long view to 2140. The Models menu above the graph does the same job.

    Model median

    The white line is the median of every included model's central value on each date, with the shaded band running from the 25th to the 75th percentile. Before the latest close it mixes models that only describe the past (moving averages, production cost) with models that project; after it, only projecting models count. The gold or wealth scenario is left out unless you include it, because its sliders set the answer.

    Families

    Models that rest on the same idea are grouped: trend fits, cycle, scarcity, adoption, price structure, holder cost, network, mining cost, the scenario and the market. By default each family takes the median of its members first, then the Model median is the median across families, so one idea gets one vote however many models express it.

    Market-implied

    What real money prices in today: the Deribit options smile turned into a price distribution at each expiry (the fan runs from the 10th to the 90th percentile), Deribit futures prices by expiry, and the median of Polymarket's BTC price ladders. These are risk-neutral prices, so they carry the cost of carry and any premium traders pay for protection. The family joins the median only forward of today and only out to its last expiry.

    The 100% rule

    A model whose 365-day walk-forward error is over 100% stays on the graph and in the scorecard but leaves the Model median and its band. It comes back on its own if a later scorecard rebuild brings the error under 100%.

    Zoom

    Drag to pan, hold Shift and drag to zoom a box, use the wheel or pinch to zoom, or pick a preset. Log time spaces the x axis by the log of days since the genesis block, so a power law draws as a straight line.

    Production cost

    The teal line is what the mining fleet pays for electricity to win one new coin: the 7-day median hashrate, times the joules the average machine burns per terahash and the power price, over the coins mined that day (144 blocks times the subsidy). Fees are left out. Fleet efficiency and the electricity price are not published anywhere, so move both and watch how far the line moves. Read it as a miner stress line, not a hard floor: when price falls, the weakest miners switch off and cost falls with it.

    Sharing and saving

    Every switch and setting is written into this page's address. Share link copies it, and whoever opens it sees the same view. Your settings are also kept in this browser, so the dashboard's charts and this page stay in step.

    Keep in mind

    Every line here is model output or a market price, not a forecast. The walk-forward scorecard is the honest test: each model rebuilt from data knowable at the time and scored on what came next.

    The graph

    loading
    Model output and market prices, not forecasts.