SatoshiMacro Model (SMM)

All six tiers are live: 48 signals feed the composite, weighted Cycle Timing 30%, Valuation 25%, Sentiment 20%, Rotation 10%, Miner 10% and Macro 5%.

For decades, the analytical frameworks institutions used to read crypto cycles sat behind paywalls, Bloomberg terminals, and prop-desk-only access. Retail got the leftovers: surface-level metrics, equal-weighted composite scores, and a steady stream of paid courses pitching what professional desks already had for free.

I traded allocated institutional capital at a Sydney proprietary trading firm. I saw the asymmetry from the inside. Professional desks were not running 9-indicator equal-weight composites. They were running weighted confluence models calibrated to historical signal quality, layering on-chain valuation with derivatives positioning, sentiment extremes, rotation flows, miner stress, and macro context. The retail tools never caught up because the gap was the business model.

The SatoshiMacro Model is what that gap actually looks like, published free. 48 signals across 6 weighted tiers - cycle-position valuation, mass psychology, miner economics, sentiment & positioning, institutional rotation flow, and macro context - calibrated against three completed Bitcoin cycles, with every input value, normalized score, tier weight, and historical reading disclosed transparently on this page. No paywall. No signup. No funnel into a course at the end. This is the tool. It costs nothing because the institutional playbook should not cost anything to use.

For years, institutional traders had access to information retail never saw. That asymmetry is finally breaking. SatoshiMacro exists to hand you the analytical playbook, and the SatoshiMacro Model is the first piece of it. Govind Satoshi · Former Institutional Trader · Founder, SatoshiMacro

More on the background, methodology, and why this is published free on the About page.

• Dataset: daily Bitcoin and Ethereum SMM scores since 2013 with all six tier scores, on Hugging Face, Kaggle and GitHub, archived on Zenodo with a citable DOI (10.5281/zenodo.23131351).
• Methodology paper: An Expanding-Window Percentile Confluence Model of Bitcoin Market Cycles (working paper, 16 pages) documents the construction, data and in-sample diagnostics, including where the model falls short.
• Interactive explorer: the Bitcoin Cycle Explorer charts the full history by zone, and a Kaggle notebook walks through tier behaviour and time spent in each zone.
• AI assistants: connect Claude, ChatGPT or any MCP client to https://satoshimacro.com/mcp to ask for the current reading (setup).

Live badge for your site or README

Updates automatically twice a day. Paste one of these snippets:

Other badges: /assets/badges/smm-eth.svg (Ethereum) and /assets/badges/altseason.svg (Altcoin Season Index). For the full interactive gauge, use the embed code below.

The SatoshiMacro Model is a composite cycle confluence model built from up to 48 on-chain, derivatives, sentiment, mass-psychology, and macro signals. Each signal is computed at every historical date back to January 2013, normalized to its own historical distribution (0-100 percentile rank or calibrated mapping), and aggregated into six tier scores. The six tier scores are then weight-summed to produce the final SMM score. Every constituent signal also lives as a standalone tool in the Bitcoin & Crypto Charts Dashboard if you want to drill into any individual signal.

Why a confluence model?

No single indicator calls cycle tops reliably. On AUD data, Pi Cycle Top crossed only at the December 2017 top, and the Mayer Multiple read 5.8 at the 2013 top but 1.4 at the November 2021 top. A model that averages many signals is less exposed to any one of them failing. CBBI popularised this approach with 9 indicators; SMM extends it to 48 signals in six weighted tiers.

Tier structure

Indicators are grouped into six tiers by what they measure, then combined with fixed weights. The weights are a judgement about how directly each family measures cycle position, not estimated from data; the historical tops and bottoms still land in their target zones with equal tier weights.

• Cycle Timing & Mass Psychology (30%). The 4-year halving clock (a fixed schedule of days since the last halving, rising from 35 at the halving to 90-95 around days 500-650 and falling to 10 near day 1,000, shaped on the completed cycles), drawdown from the all-time high, days since that high, a rolling profitable-days share and the 90-day return.
• Valuation & Cycle Position (25%). MVRV Z-Score (currently a 4-year moving-average proxy, see caveats), Pi Cycle Top, Pi Cycle Bottom, Mayer Multiple, 2-Year MA Multiplier, 200-Week MA distance, Power Law deviation, Golden Ratio Multiplier, Rainbow Chart position and the Bitcoin Risk Metric. Power Law deviation, Rainbow position and Risk Metric come from the same power-law fit and rank identically, so in effect they act as one signal carrying about 7.5% of the composite.
• Sentiment & Positioning (20%). Fear & Greed, Google Trends (Bitcoin / buy Bitcoin / Bitcoin tax), funding rate, open interest, Coinbase Premium, Deribit put/call ratio (inverse), Deribit DVOL implied volatility, the Bybit-Binance price spread and the 3-month futures basis.
• Rotation & Institutional Flow (10%). BTC Dominance (inverse), Altcoin Season Index, Strategy (MSTR) and Coinbase (COIN) share prices, IBIT assets, ASX Bitcoin ETF assets, cumulative US spot ETF holdings, 30-day ETF net flows and Strategy's quarterly BTC purchases. Most of these only exist from 2020-2024, so this tier is thin in earlier cycles.
• Miner & Production Stress (10%). Puell Multiple, Hash Ribbons state, miner fee-share anomaly, Stock-to-Flow deviation, hashrate versus its 365-day peak and the latest difficulty adjustment.
• Macro Headwinds/Tailwinds (5%). USD Index (inverse), M2 growth YoY, 2s10s yield curve, VIX (inverse), S&P 500, Nasdaq 100 and gold (inverse).

Normalization

44 of the 48 signals are normalized to their own expanding-window percentile rank (0-100): each day's value is ranked only against that signal's history up to the same day. A Mayer Multiple of 1.5 doesn't feed in as "1.5"; it feeds in as, say, "the 78th percentile of Mayer readings so far," which puts it on the same footing as a funding rate at its own 78th percentile. The other four (the halving clock, drawdown from the all-time high, days since that high and the Hash Ribbons state) use fixed mappings instead.

Calibration curve

The raw weighted composite read 82.6, 86.8, 80.1 and 73.0 at the four historical tops, so three of the four fall short of the 85+ Cycle Top zone before calibration, partly because diversifying signals (sentiment, macro, rotation) don't all peak at the same moment as cycle-position valuation - the weighted average gets diluted by signals that haven't fired yet. CBBI doesn't have this dilution because it uses only 9 indicators, all highly cycle-correlated. SMM keeps the 48-signal diversity (more robust against false signals and single-source outages) and applies a calibration curve on the final composite to stretch the upper half of the 0-100 scale.

The calibration is piecewise linear:

• Raw 0-40 (identity): bottoms pass through unchanged. 2018-12 (22.3) and 2022-11 (22.8) read the same raw and calibrated.
• Raw 40-55 (slope 1.5): mid-cycle stretch. Raw 55 maps to 62.5.
• Raw 55-64 (slope 2.5): steep late-cycle stretch. Raw 64 maps to 85, the start of the Cycle Top zone.
• Raw 64-78 (slope about 0.71): the body of the top zone. Raw 78 maps to 95.
• Raw 78-83 (slope 1.0): the tip. Raw 83 maps to 100.
• Raw 83+ (clamp): stays at 100.

Calibrated readings at historical cycle tops (48-signal version): 2013-12 (99.6), 2017-12 (100), 2021-04 (97.1), 2021-11 (91.4). All four register in the Cycle Top zone (85-100). The knots were chosen so that the seven reference dates land in their target zones, which makes that result in-sample. The zone is also not rare: the calibrated score has sat at 85 or above on roughly one day in five since 2013, often for months before a top. The raw pre-calibration score is also emitted as raw_smm in the output JSON so consumers can compare or recalibrate if needed.

Zone interpretation

• 0-15 Deep Value. Deepest-value zone. No documented cycle bottom since 2013 has closed this low on the calibrated score; the three bottoms read 20, 22.3 and 22.8, in Accumulation.
• 15-30 Accumulation. Favourable risk/reward. Below-trend valuation.
• 30-50 Neutral. Mid-cycle. No strong directional edge.
• 50-70 Caution. Late-mid cycle. Begin de-risking on strength.
• 70-85 Distribution. Historical late-cycle zone. Trim aggressively. About 46% of past Distribution days fell within 12 months of a major cycle top; the rest came during mid-cycle rallies.
• 85-100 Cycle Top. Historical top zone. Maximum caution. Calibrated readings (48-signal version): 2013-12 (SMM 99.6), 2017-12 (SMM 100), 2021-04 (SMM 97.1), 2021-11 (SMM 91.4). All four major BTC cycle tops register in this zone with the calibration curve applied.

Caveats and limitations

• SMM starts at January 2013, not Bitcoin's genesis. Pre-2013 Bitcoin was in raw price discovery: thin liquidity, single-venue trading (Mt. Gox), and a market cap measured in tens of millions of USD. Cycle-position metrics computed against that period produce unreliable readings because the "historical distribution" itself is too small and too dominated by a single venue's order book to be meaningful. Same reason the Mayer Multiple and most other cycle indicators are conventionally backtested from 2013 onward, not from the 2009 genesis block. Treat any chart visualisations that extend earlier than January 2013 as price-history context only; the SMM line begins at the first reading where the underlying distribution is broad enough to support meaningful percentile ranks.
• Percentile ranks use expanding-window history. Each historical date's score is computed only against indicator values that existed up to that date - no lookahead bias. The rank step is lookahead-free, but the halving schedule, the power-law and stock-to-flow fits, the zone boundaries and the calibration knots were all chosen with knowledge of the full history.
• MVRV Z-Score currently uses a proxy. Standard MVRV compares market cap with realised cap. The free realised-cap feed (CoinMetrics community API) has not been reachable from the build servers, so the MVRV input is a 4-year moving-average proxy, labelled "(4Y MA proxy)" in the indicator list. If a working realised-cap feed is connected, the model switches to the standard formula automatically.
• Indicator coverage varies by date. Earlier periods (pre-2018) have less data from sentiment, derivatives, and macro tiers, so the composite there uses re-normalized weights across only the available tiers, and readings before 2018 rest on fewer independent inputs than later ones.
• This is not financial advice. The SMM is a research tool for cycle position context. It does not predict prices. Cycle indicators have failed before and will fail again. Use as one input among many.

How SMM differs from CBBI

CBBI uses 9 indicators with equal weight. SMM uses 48 signals across 6 tiers with fixed weights that put cycle timing and valuation first, plus a calibration curve on the composite. The larger panel makes SMM more robust to single-source data outages: any one feed going dark only marginally moves the composite. Neither approach has a true out-of-sample record yet. Both read the known tops correctly in hindsight, so treat SMM as a structured summary of many indicators rather than a proven forecaster.

How to use it

The SMM is a position-sizing input, not a buy/sell trigger. Sample uses:

• SMM in Deep Value zone (0-15): consider increasing crypto allocation toward your maximum.
• SMM in Distribution zone (70-85): trim positions, raise cash, take profits on strength.
• SMM in Cycle Top zone (85-100): maximum caution. Each completed cycle ended in a 74-83% drawdown after a Cycle Top run, but the zone has also held for months while prices kept rising.
• SMM in Neutral zone (30-50): no signal. Manage based on your base strategy.