There is no scientific model of Bitcoin's price. There are only indicators with better or worse epistemics. This page explains the four-tier ranking used across Chainmeter.
Tier 1 — On-chain cost basis (measured)
Bitcoin's ledger records the price at which every coin last moved, so the aggregate cost basis of all holders is a measured quantity, not an estimate. Realized Price, MVRV Z-Score and NUPL are arithmetic on that measurement. They have clear economic interpretations (unrealized profit and loss) and no fitted parameters. This is as rigorous as Bitcoin analytics gets — which is still not prediction.
Exchange Netflow and Exchange Reserve are measured the same way but answer a different question: not cost basis, but which direction supply is structurally moving relative to exchanges, and what share currently sits there — a rate and a level of the same underlying phenomenon. Both sit alongside the rest of Tier 1 on rigor, each with its own caveat printed on its page.
Tier 2 — Production cost
Miners must sell coins to pay fiat-denominated costs, making them the market's only structural sellers. The Puell Multiple measures when their revenue is historically extreme; Hash Ribbons measures the same phenomenon through actual mining hardware participation instead of dollar revenue; Security Budget measures what share of that revenue already comes from fees rather than the shrinking block subsidy. The logic is causal, but each halving shrinks miner flow relative to total volume, so the signal decays by design.
Tier 3 — Statistical, fixed parameters
The Mayer Multiple and 200-Week MA Heatmap describe price relative to its own moving averages using conventions fixed decades ago. Nothing can be quietly re-tuned, which makes them honest — and purely descriptive. They are thermometers.
Tier 4 — Statistical, fitted parameters
The Power Law Corridor and Pi Cycle Top have parameters chosen by fitting history. So do the Rainbow Chart and Stock-to-Flow pages, both built as honest reconstructions of famous public models rather than references to them, and Metcalfe's Law, which fits market cap against active addresses instead of price or scarcity. The failure mode is universal across all five: when reality breaks the fit, the fit gets redrawn, so the model is unfalsifiable. We include them because they are widely watched, and we print their parameters, refit dates and out-of-sample failures directly on each page.
What "not scientific" means here
A scientific model makes falsifiable out-of-sample predictions from a causal mechanism. No Bitcoin price model clears that bar: the sample is ~4 cycles, the market's structure is non-stationary (ETFs, institutions, derivatives), and every published model that made hard predictions — most famously Stock-to-Flow — has failed and been revised after the fact. The honest use of everything on this site is describing present conditions against a short history, nothing more.
Where the Tools fit
Cycle Comparison, the Cycle Top & Bottom Timing page, the Halving Countdown, the Supply Schedule, Drawdown, Monthly Returns, Volatility, the Sharpe Ratio, the DCA Calculator and the Lump Sum Calculator aren't ranked in the four tiers above — they're not indicators making a claim about value, just ways of viewing the same underlying price and network history from a different angle. The Sharpe Ratio specifically combines the Volatility page's own output with return data rather than introducing a new data source, and the Supply Schedule is the one page whose future line is a protocol rule rather than an estimate about behavior. Cycle Top & Bottom Timing is the most explicitly caveated tool on the site: it detects past tops and bottoms with one fixed, disclosed rule (a 70% reversal), then mechanically projects the same historical offsets onto the next halving — labeled throughout as a naive extrapolation from an n=3 sample, not a forecast. The Market Dashboard is also unranked in this sense: it re-expresses five Tier 1–3 indicators as percentile ranks against their own history rather than fixed thresholds, and averages them equally rather than by tier weight — a deliberately different, complementary lens to the Composite score below. The Composite score is the exception to all of this: it's an aggregate of Tier 1–4 indicators, weighted by the same tier structure documented here.
Data & computation
All series come from the Coin Metrics community API (PriceUSD, CapMrktCurUSD, CapMVRVCur, SplyCur, IssTotUSD, FeeTotNtv, HashRate and related fields), fetched by your browser at load time. Realized cap and miner revenue aren't published on the free tier, so this site derives them from those primitives — market cap divided by the published MVRV ratio for realized cap, issuance plus fees for revenue. Formulas are stated on each page and implemented in ~100 lines of readable JavaScript per page — view source to audit any of them. Nothing is uploaded, cached server-side, or tracked beyond standard analytics.