BTC Intelligence Brain

Timeframe	Trend	Momentum	Signal	Note

How to read these numbers (plain words). A probability of 0.55 means the model thinks "up" is a little more likely than a coin flip — it does NOT mean certainty. The baseline is what you'd score by always guessing the most common answer; skill only exists above that line, which is why we print it next to every hit rate. The Wilson lower bound answers "given this sample size, what's the worst this hit rate could plausibly really be?" — small samples get wide, humble bounds. And maker expectancy subtracts trading fees, because a 55% guesser can still lose money after costs. If a row doesn't clear all three bars, we say so.

Loading the measured record…

Horizon	n	Hit rate	Always-up baseline	vs majority	Maker expectancy	Wilson 95% LB
Loading accuracy artifact…

How to read this: a horizon only has real edge if its hit rate beats the majority baseline and the Wilson 95% lower bound clears 50% and expectancy survives fees. Maker expectancy assumes a 2 bps round trip. Green means the bar is cleared, amber means promising-but-unproven, red means below baseline.

Loading signal artifact…

Each resolved forecast is treated as one paper trade of a fixed 1-unit stake in the forecast direction, entered at the recorded entry price and exited at the horizon close, minus a 2 bps maker round trip. This is a derivation from the public ledger, not an account statement — no real money is traded by this page.

Read our own post-mortems. The strongest trust signal we can offer is publishing our falsification work: the edge autopsy (where the edge lives and where it dies), the upgrade master plan, and the latest promotion/kill decision. Raw artifacts for auditors: forecasts.jsonl · resolutions.jsonl · accuracy.json · edge_report.json · full research library →

⟳ Auto-Rotate: ON ⊙ Reset View ⚡ Pulse All

• Volume confirmation on closed candles (live, see Live Market panel above)
• Derivatives — funding rate, open interest (live or via data/public/derivatives.json)
• ETF / institutional flows (no live feed wired into this page)
• Macro context — DXY, M2, Fed policy (research chapters only)
• Sentiment — Fear & Greed, retail positioning (no live feed wired into this page)

Page generated by static build · last commit timestamp lives in git history. Coverage Score Δ: editorial — not a performance metric For measured accuracy & freshness see the Trust & Transparency section.

• Ledger-backed metrics only. Hit-rate / Brier / log-loss numbers are computed from ledger/public/accuracy.json, which is generated from append-only forecast and resolution streams. No author-set hit rates appear in the measured panels.
• Sample-size aware. Until at least n=20 forecasts are resolved, accuracy is reported as insufficient data. Rolling windows (7d / 30d / 90d) are shown only when each window meets its own minimum.
• Models run in shadow mode. All models in models/public/registry.json are gated by the Phase 4 promotion criteria (resolved-per-horizon ≥ 30, shadow days ≥ 14, ECE ≤ 0.08, must beat baseline Brier, manual approval). Until those gates pass, their forecasts are recorded but never shown as “live”.
• Source freshness, not just price. Every data feed exposes a freshness budget in data/public/source_health.json; panels stamp themselves stale when their source is past budget.
• Confidence intervals are not fabricated. Probability bands appear only when generated by a valid calibration artifact. Otherwise the field reads unavailable.
• Daily briefings. The archive renders only artifacts that exist on disk. No email or external delivery integration is claimed unless its artifact is present.
• Hub status: not yet integrated. Cross-site HUB readiness (Phase 5) is being built in a separate repository and is not yet merged here. Any HUB-dependent panel will show not connected until that work lands.
• Not financial advice. Do not size positions or allocate capital based on anything shown here.

One email a week: the ledger's live hit-rate vs. baseline, regime shifts, and what changed in the models. Join now — you'll get the first issue when delivery goes live.

The deal: a paid tier only opens if and when the public accuracy ledger above beats its baseline out of sample. Until then, everything here stays free. Not financial advice.

Join free

Reminder: Educational research only. Forecasts are a research artifact; until the public ledger has enough resolved samples and the Phase 4 promotion gates pass, no model on this page is rated “live”. Do not trade on this page.

What is Bitcoin?

Bitcoin is digital money that no company, bank, or government controls. Instead of a bank keeping the record of who owns what, thousands of computers around the world keep identical copies of one shared record (the blockchain), and they constantly check each other. To cheat, you would have to fool most of them at once — which is what makes the record trustworthy without needing a boss.

Why do people care?

Two reasons. First, scarcity: the rulebook says only 21 million bitcoins will ever exist, and nobody can change that without convincing the whole network. Regular money can be printed in unlimited amounts; Bitcoin cannot. Second, permission-free: anyone with a phone can receive or send it, anywhere on Earth, without asking a bank for an account. Whether those two properties make it valuable is exactly what the market argues about every day — that argument is the price chart.

Why does the price jump around so much?

Because nobody knows what it "should" be worth, the price is pure supply and demand — global, 24/7, amplified by borrowed money and emotion. Bitcoin has climbed thousands of percent and it has crashed by more than half many times (see Chapter 2 and Chapter 4). Both are normal for this asset.

The halving — Bitcoin's built-in clock, interactive

New bitcoins are created as a reward to the computers that secure the network (mining). Roughly every four years that reward is cut in half — so new supply keeps slowing until it stops entirely around the year 2140. Drag the slider to watch the schedule play out:

How to use this site

Two ways in, depending on what you want:

• Learn how Bitcoin works — read the chapters in order. The 10-minute path: Fundamentals → Halving Cycles → Bull & Bear Markets → Risk Management. Then test yourself. Dotted-underlined words show definitions when you hover or tap them.
• See the live data — the Live Market Command Center shows what the market is doing right now, and the Forecast Scoreboard shows how our prediction experiments are actually going, measured in public.

Every chapter starts with an "In plain words" box — if the full text feels heavy, the boxes alone tell the story.

Author-compiled estimates from public research and backtests — not measured by this site’s forecast ledger. Measured accuracy lives in the Forecast Scoreboard.

⏳ Archived: figures below reflect April 2026 conditions and will disagree with the live panels. That disagreement is the lesson.

Funding rates settle every 8 hours and represent the cost of holding a leveraged position. Extreme readings are among the most reliable cycle sentiment signals in BTC markets.

Rate (8h)	Annualized	Signal	Action
+0.1% or higher	109%+	Extreme greed — leverage overheated	Reduce longs, watch for cascade
+0.01% to +0.05%	11–55%	Elevated but manageable	Monitor, normal bull regime
-0.001% to +0.01%	-1% to 11%	Neutral — healthy market	No signal, trend-follow
-0.01% to -0.05%	-11% to -55%	Bearish leverage heavy — shorts piling in	Watch for short squeeze
Below -0.05%	-55%+	Extreme fear — capitulation signal	High-probability accumulation zone

Price	Open Interest	Signal	Interpretation
↑ Rising	↑ Rising	BULLISH	New long positions entering — confirmed uptrend
↓ Falling	↑ Rising	BEARISH	New shorts being added — bearish pressure
↑ Rising	↓ Falling	SHORTS COVER	Short squeeze — longs may not sustain
↓ Falling	↓ Falling	LONGS FLUSH	Deleveraging — possible bottom after flush completes

Event	Liquidated	Price Drop	Trigger
COVID Black Thursday (Mar 2020)	$1.2B	-53% in 24h	Global risk-off, crypto-native panic
May 2021 China ban	$8.6B	-35% in 72h	Hashrate crash + regulatory fear
LUNA collapse (May 2022)	$2.1B	-25% in 7 days	UST depeg, contagion across DeFi
FTX collapse (Nov 2022)	$5.2B	-28% in 1 week	Exchange fraud, trust collapse
Oct 10, 2025 (largest ever)	$19.35B	-14% in 24h	OI $56.6B → ETF outflows → cascade
Feb 5–6, 2026	$6.18B	-8% in 48h	7-catalyst convergence, ETF outflows

Period	Capacity (BTC)	Channels	Context
Mar 2018	<10	<1,000	Mainnet launch, experimental
Dec 2019	866	35,000	Early growth phase
Jan 2021	1,060	38,000	Pre-bull surge
Oct 2021	3,200+	80,000+	El Salvador legal tender law
Dec 2022	5,100	82,000	Cash App integration, ATH despite bear
Dec 2023	4,800	70,000	Binance/OKX integration
Dec 2025	5,637	~65,000	All-time capacity high
Q1 2026 (est.)	~5,200	~62,000	Slight decline with BTC price drop

During the 2022 bear market, LN capacity hit an all-time high (5,100 BTC) while BTC price was down 77%. This was a historically bullish signal — the network was growing independent of price speculation. The Federal Reserve found that LN could have reduced 2017's on-chain congestion by 93%. Long-term, LN capacity growth relative to price represents genuine adoption vs speculation, making it one of the few metrics that improves during bear markets.

Tier	Input	Type	Signal Quality	Notes
TIER 1	Halving Cycle Position	Leading	VERY HIGH	4/4 cycles followed halving math within 15%
TIER 1	Funding Rate Extreme	Contrarian	HIGH	Extremes (>±0.05%) signal reversals 73% of cases
TIER 1	MVRV Zone	Lagging	HIGH	Cycle bottoms always in MVRV <1.0 zone
TIER 1	Volume Confirmation	Coincident	HIGH	Breakouts with <1.5× avg vol fail 67% of time
TIER 2	Fear & Greed Extreme	Contrarian	MEDIUM	F<20 → 90d median return +42% (2019-2026)
TIER 2	ETF Net Flows	Leading	MEDIUM	3-5 day lag from flows to price response
TIER 2	OI Direction	Coincident	MEDIUM	OI rising with price = confirmation; against = warning
TIER 3	Social Sentiment	Lagging	LOW	Noisy. Only useful at extreme readings
TIER 3	News Catalysts	Coincident	LOW	Reaction often fades. Trend remains dominant