Crypto Investor
Current Crypto Cycle: Late Markup — Agent Prepared the Exit Plan
Key Takeaway
On-chain data classified the current crypto cycle as late markup — the phase right before distribution — and the agent built a phased exit plan before the crowd even started talking about tops.
The Problem
Every crypto cycle has the same four phases. Accumulation. Early markup. Late markup. Distribution. And every cycle, people know this intellectually but fail to act on it emotionally.
Late markup is the hardest phase. Prices are high. Your portfolio looks incredible. Twitter is euphoric. Everything is going up. And somewhere in the back of your mind, you know it can't last — but selling feels stupid when things are still ripping.
I've been through 2017 and 2021. Both times, I knew we were in late markup. Both times, I held too long. The problem isn't analysis — it's execution. You need a plan before the emotions hit, and you need something external to hold you accountable to it.
That's what the agent does. It doesn't just classify the cycle. It builds the exit plan and tells me when to execute each phase. No emotions. No "this time is different." Just data and a plan.
The Solution
The Crypto Macro Regime Classifier skill ingests on-chain metrics, exchange data, and cycle-specific indicators to classify the current market phase. It runs daily and compares current metrics to historical patterns from 2013, 2017, and 2021 cycles.
The Process
The classifier tracks 12 core on-chain signals:
yamlShow code
# crypto-regime-config.yaml
on_chain_signals:
valuation:
- MVRV_ratio # Market value vs realized value
- NUPL # Net unrealized profit/loss
- puell_multiple # Miner revenue vs 365d avg
behavior:
- exchange_netflow # Coins entering/leaving exchanges
- LTH_supply_change # Long-term holder distribution
- STH_SOPR # Short-term holder profit ratio
network:
- active_addresses # Network activity
- transaction_volume # On-chain transaction value
- new_addresses # Growth rate
market_structure:
- funding_rates # Perpetual futures funding
- open_interest # Derivatives positioning
- stablecoin_supply # Dry powder on sidelines
cycle_phases:
accumulation: { action: "build_position", size: "full" }
early_markup: { action: "hold", size: "full" }
late_markup: { action: "begin_exit", size: "phased_reduction" }
distribution: { action: "accelerate_exit", size: "minimal" }
Current cycle classification output:
View details
Cycle Phase: LATE MARKUP (confidence: 78%)
Evidence:
├── MVRV: 2.8 (elevated — historical tops: 3.5-4.0)
├── NUPL: 0.62 (greed zone — not extreme yet)
├── Exchange netflow: -12K BTC/week → -3K BTC/week (outflows declining)
├── LTH supply change: -0.4%/month (distribution begun)
├── STH SOPR: 1.04 (still profitable but compressing)
├── Funding rates: 0.03% (elevated but not extreme)
├── New addresses: +2.1%/week (retail arriving)
└── Stablecoin supply: $165B (high — but growth slowing)
Cycle Comparison:
┌──────────────┬──────────┬──────────┬──────────┐
│ Metric │ 2017 Top │ 2021 Top │ Current │
├──────────────┼──────────┼──────────┼──────────┤
│ MVRV │ 4.2 │ 3.8 │ 2.8 │
│ NUPL │ 0.78 │ 0.72 │ 0.62 │
│ LTH selling │ heavy │ heavy │ starting │
│ Exchange flow│ large in │ large in │ neutral │
│ Retail surge │ extreme │ extreme │ moderate │
└──────────────┴──────────┴──────────┴──────────┘
Assessment: Late markup phase. Not at cycle top yet.
Historical pattern: late markup → distribution in 2-4 months.
Conditions for distribution trigger: MVRV >3.5, NUPL >0.70,
exchange inflows sustained >20K BTC/week.
The agent generated this exit plan:
jsonShow code
{
"phase": "late_markup",
"exit_plan": {
"phase_1": {
"trigger": "NOW (late markup confirmed)",
"action": "Reduce 20% of crypto allocation",
"target": "Stablecoins or fiat",
"rationale": "Lock partial profits while still in markup"
},
"phase_2": {
"trigger": "MVRV crosses 3.2 OR exchange inflows >15K BTC/week",
"action": "Reduce additional 20%",
"target": "Stablecoins",
"rationale": "Approaching historical distribution zone"
},
"phase_3": {
"trigger": "Distribution phase confirmed OR MVRV >3.5",
"action": "Evaluate remaining 60%",
"target": "Keep 30% for blow-off potential, exit 30%",
"rationale": "Asymmetric: blow-off top could 2x, but distribution could -60%"
}
},
"requires_approval": true
}
The Results
84% on 2017 + 2021 data
Cycle classification accuracy (backtested)
Late markup (78% confidence)
Current phase
~6 weeks
Time in late markup
2-4 months
Historical late markup → distribution
Yes — 20% to stablecoins
Phase 1 exit executed
Reduced from 18% to 14.4%
Portfolio crypto exposure
+34% on exited portion
Unrealized gains locked
The power of this system isn't in predicting the exact top. Nobody does that. The power is in having a plan that executes mechanically. When MVRV crosses 3.2, I don't need to think about it. I don't need to argue with Twitter. Phase 2 fires. That's it.
Try It Yourself
Install the Crypto Macro Regime Classifier skill. You'll need an on-chain data provider — Glassnode or CryptoQuant APIs work. The classifier needs at least 2 full cycles of historical data to calibrate (built-in for BTC and ETH). Configure your exit plan thresholds based on your risk tolerance.
The key: build the plan during accumulation or early markup, when you're calm and rational. Not during late markup when greed is screaming.
The best exit plan is the one you wrote when you didn't need it.
Related case studies
Crypto Investor
On-Chain Data Says Distribution Phase — My Agent Sold Before the Drop
How an AI agent classified crypto's distribution phase using MVRV, SOPR, and whale data — and reduced exposure before a 35% drop. On-chain analytics meets autonomous portfolio management.
Crypto Investor
DCA Automation on Base — $500 of ETH Every Monday, Zero Effort
AI agent automates weekly DCA purchases on Base L2 via Bankr — executing swaps, tracking cost basis, and pausing on abnormal gas. Zero manual effort crypto accumulation.
Crypto Investor
Fear-Greed Extremes as a Buy Signal — How My Agent Caught the Bottom
Using AI agents to monitor the crypto fear-greed index and trigger buy signals at extremes. Historical backtest shows 85% average returns buying at fear levels below 20.
Want results like these?
Start free with your own AI team. No credit card required.