Methodology
How our tools actually work.
TMU builds its own research tools. This page explains, in plain terms, what each one measures, what it puts on your screen, and what it deliberately does not do. Everything here is decision support for members. None of it is a buy or sell signal, and none of it is financial advice.
We build tools we can stand behind, so every one of them is held to the same rules. Each tool reads real data and judges it against history rather than against a headline. Each one is honest about its limits, and where a tool has a weakness we tell you here instead of hiding it. No tool on this page issues buy or sell instructions, and no tool replaces your own judgement. A tool you can trust is one that tells you what it cannot do. Every framework below was built by one of the two analysts behind this platform, and is named accordingly.
Where Bitcoin sits in its cycle on a cheap-to-expensive scale, read from on-chain valuation and holder behaviour against Bitcoin's own history. It is scored from 15 on-chain metrics across five families: valuation, holder conviction, absorption and spending, miner economics, and cycle timing. Each metric is percentile-ranked against its own history from 2009 to 2021, then applied out of sample to 2022 and onward. The score updates daily and the written analysis publishes every Monday.
A single score from 0 to 100 with a zone label, where 0 is the cheapest on-chain reading in Bitcoin's history and 100 the most extended. It tells you whether Bitcoin is historically cheap, mid-cycle, or extended on-chain, and flags separately when capitulation conditions are present.
It identifies zones, not turning points, and it is not a timing tool. It reads bottoms and tops early, often by months, and it is deliberately never framed as a sell signal. Bitcoin only, on-chain only. No price structure, no macro, no sentiment.
Where Ethereum sits in its own on-chain cycle, from the low end of its range to overheated. It is a separate engine from the Bitcoin Barometer, built on Ethereum's own record rather than copied across from Bitcoin's. On-chain only, with every measure percentile-ranked against Ethereum's own history and the training window frozen so recent years stay out of sample.
A score from 0 to 100 expressed as one of six named zones, running from deep accumulation at the low end to overheated at the high end. Beneath the headline, group-level readings show what is driving it. Where a group of measures cannot yet be scored, the page shows the hole rather than filling it with a zero: staking and new-supply flows are currently not scored, and it says so on screen.
It reads the level, not the turn, and can sit cheap or heated for long stretches. The top end rests on less history than the rest of the range, so it is directional there rather than precise. There is no accuracy figure and no claimed track record. Ethereum only, on-chain only. It says nothing about ETH against BTC.
A cycle-timing framework. It identifies windows in time where a market's cycle low has historically tended to form, and once a window opens it tracks how far a recovery has progressed. It brackets when, not how low. KAIROS reads price structure and cycle timing only and consumes no on-chain data of any kind. It runs across Bitcoin, Ethereum, the S&P 500 and Gold, on a defined set of timeframes per market.
Per market and timeframe, an active cycle-low window when one exists: the market context, how far the recovery has progressed in staged steps, a countdown to the window closing, and how comparable past windows resolved, including the size of the sample behind it. Where the historical base is thin, the tool labels it as limited rather than quietly averaging it in.
It is not a buy signal, not a confirmed bottom call, not a price target, and not a strategy with a win rate. Its statistics describe how past windows resolved, not what any trade would have returned, because no entries, stops or sizing are scored anywhere in it. It cannot predict individual outcomes: windows that failed looked the same at the open as windows that worked. Cycle-top windows showed no demonstrated edge and are turned off.
Whether today's conditions for Ether relative to Bitcoin resemble the setup that real historical ETH/BTC rotations shared. It is about ETH versus BTC only, not ETH versus alts, and it says nothing about any specific altcoin. It pairs a read of whether the broader setup looks favourable with a read of whether the ETH/BTC ratio has actually turned up.
A state, not a number. One of Quiet, Watch, Caution or Aligned, with a plain headline and the conditions sitting behind it. Aligned means both of the measured conditions that past rotations shared are present now. Alongside it, the tool shows the small set of historical rotations it rests on and says on screen that the base is thin.
It does not score, predict, or name a date, and it says nothing about any altcoin. Aligned is explicitly not a buy. It confirms a turn rather than forecasting one, and two known false dawns still read as strong setups under its current design. Those are documented rather than tuned away.
A weekly read on whether Bitcoin's conditions resemble a market bottom. It combines several weekly market conditions, including how the week closed, whether real buying or selling dominated traded order flow, how heavy the volume was, where price sits against its recent range, and how bullish or bearish investor sentiment is, into a single weekly state. It runs every Monday at 7am UTC against the just-closed weekly candle. Bitcoin only.
One weekly state. The only affirmative reading it publishes is Strong Bottom, and the default, by far the most common, is No Signal. A separate Top Potential watch flag is surfaced and clearly marked as unvalidated. Every week carries a confidence marker showing whether all its inputs were available, and the Monday reading is flagged as preliminary until the sentiment input lands mid-week.
It answers one question over a horizon of weeks. It is not a price forecast, gives no targets or levels, is not a trade trigger, is not a position-sizing tool, and covers no asset other than Bitcoin. Weak signals are still computed but no longer published, because they showed no forward edge. Historical labels are recomputed as the rules improve, so no fixed track record is claimed from them. It does not tell you what happens next week.
The macro regime of the current year, read through seven macro assets covering growth demand, credit appetite, real borrowing costs, the dollar, credit conditions, energy and the shape of the rate curve. Each asset sets a price range in January. Macro Concordance reads how each one breaks that range through the year, turns the result into the year's macro fingerprint, then compares that fingerprint against every year on record to find the closest historical analogues.
Two panels. This year's macro picture across the seven assets, each marked bullish, bearish or neutral. Then past years grouped by how closely they match, with the matching years listed by name and how each group tended to resolve for the asset you select. The match is a plain visible count out of seven, not a hidden score, and there are no proprietary weights inside it.
It is regime context, not a trading signal, a price target, or a timing mechanism. It reads monthly closes only, so it moves slowly and does not react to anything until a month ends. Usable history for most of the assets begins in the early 2010s, and the closest-matching groups can contain very few years, so their percentages should be read with that in mind. Where data is missing it says so rather than filling in.
The macro backdrop for Bitcoin. It tracks whether the broad conditions that have historically supported or undermined Bitcoin, such as liquidity, credit health, business activity and market stress, are improving or deteriorating. It is built from public market and economic data, with fallback sources on every input so it keeps working when a single feed goes down.
A label for the current macro regime, a composite reading with its standing against its own history, separate leading and coincident readings with the gap between them, and the component values behind each. Context labels sit on top to help you tell a macro-driven move from a crypto-native one, alongside a staged bottom sequence and a separate top-warning classification.
It is a regime and risk filter, not a buy or sell signal. It cannot flag tops driven by crypto-specific factors while the wider economy still looks healthy, and that blind spot is by design rather than an oversight. Its highest-confidence top warning needs weeks of sustained confirmation, so it is deliberately late rather than early, and its bottom signal will not fire in shallow drawdowns however good the other conditions look.
The crypto market read from the inside, and the counterpart to the BCI's macro read. It tracks whether capital is rotating into risk, where the cycle sits relative to its own history, and whether the signals that have tended to lead price are improving or deteriorating. Inputs include cycle position, market structure and breadth, stablecoin deployment, and derivatives positioning. It runs weekly.
A composite reading with a separate leading line plotted alongside it, plus rows for cycle position, market structure, stablecoin deployment and derivatives positioning, a context label, and a staged bottom-recovery sequence showing which phase is active. On-chain profit-taking is shown as its own standalone context row.
It is a regime and risk filter, not a buy or sell signal. It runs on weekly bars only, so intra-week moves and sharp reversals are not reflected until the week closes. It can miss tops driven by factors it does not measure. The bottom sequence must progress in order and will not skip, so a recovery that does not follow the usual repair pattern may be caught late or not at all. The on-chain profit-taking row is deliberately kept out of the composite because testing showed it did not improve it.
Which alt names inside each sector deserve attention, deeper review, or caution right now, judged against their sector peers rather than against the whole market. It is a structured research and monitoring layer, led by analyst judgement, organised into focused sectors, and dated on every refresh. Reviews run on a defined cadence of run types, from a light weekly pass through to a deeper monthly rebuild, and each board discloses the coverage behind it.
Names grouped by sector, each placed in one of three phases: Watch-Only, Priority Review, or TMU Focus, with the date the board was last reviewed. Where a name has not been re-examined it is carried forward and shown as carried forward, rather than presented as reconfirmed.
It is not a score, a leaderboard, a heatmap, or a buy list, and it does not rank the whole market. There is no automated scoring engine behind it, and we do not claim one. It issues no buy signals and no price calls, and it does not replace your own judgement. Coverage has run partial on some sectors in the past, which is exactly why every board carries its review date and its coverage note.
A planning framework rather than a signal. It gives each tracked coin a defined set of accumulation entry zones, staged profit-taking targets, and a level at which the thesis is considered invalid, all shown against the live price so you can see how far away each one is. It covers 42 coins plus two market indices held for context, and where relevant it carries both a longer major-cycle set of levels and a shorter near-term set.
Per coin, the live price against its levels, a table of entry zones and staged targets with the distance to each, guidance scaled to a risk profile you choose, a clearly marked invalidation level, and optional per-level alerts. Reached targets are struck through and the next unreached target is highlighted.
It is a framework for planning entries and exits in advance, not a signal service, not auto-trading, not personalised advice, and not a prediction of price. The levels come from TMU's own research judgement; the tool displays and contextualises them rather than deriving them. Near-term targets are deliberately withheld during corrective phases, so a coin can show entry zones and no targets. The on-screen hit-rate readout is a rough progress indicator calculated in your own browser, not an audited track record, and it is not published as one.
Judge the tools, then decide.
Every framework above is built, maintained and used by Doc and Oz in the weekly research. If the method holds up, the membership is where you use it.