emerging-movers

📁 senpi-ai/senpi-skills 📅 1 day ago
9
总安装量
5
周安装量
#32887
全站排名
安装命令
npx skills add https://github.com/senpi-ai/senpi-skills --skill emerging-movers

Agent 安装分布

mcpjam 5
gemini-cli 5
claude-code 5
junie 5
windsurf 5
zencoder 5

Skill 文档

Emerging Movers Detector v3.1

Tracks Smart Money market concentration across all Hyperliquid assets and flags assets accelerating up the ranks before they become crowded top-3 plays. By the time an asset hits the top of the SM leaderboard, the easy money is gone. This catches the trajectory.

One API call per scan. Near-zero LLM tokens. Runs every 60 seconds.

How It Works

The SM Profit Concentration Leaderboard

Senpi’s leaderboard_get_markets returns all assets ranked by percentage of total Smart Money profit in the last 4-hour rolling window. This isn’t trader count — it’s where the money is actually flowing.

#1  ETH SHORT   31.4%  286 traders
#2  BTC SHORT   25.1%  436 traders
#3  HYPE SHORT  24.2%  330 traders
...
#36 ASTER SHORT  0.2%   18 traders  ← 60s later: #13, 0.82%, 65 traders

The script tracks this leaderboard over time and detects acceleration.

Detection Signals

Immediate Action Signals (v3+)

Signal Condition Priority
IMMEDIATE_MOVER 10+ rank jump from #25+ in ONE scan Highest — act now
NEW_ENTRY_DEEP Appears in top 20 from nowhere Very high
CONTRIB_EXPLOSION 3x+ contribution increase in one scan Very high
DEEP_CLIMBER 5+ rank jump from #25+ High

Trend Signals

Signal Condition
NEW_ENTRY First appearance in top 50
RANK_UP Jumped 2+ positions in one scan
CLIMBING 3+ positions up over several scans
ACCEL Contribution % increasing scan-over-scan
STREAK Consistently climbing every check
VELOCITY Sustained positive contribution growth

v3.1 Quality Filters

These prevent false IMMEDIATE signals that looked great on rank jump alone but failed on execution:

Filter Rule Rationale
Erratic rank >5 rank reversals in history → erratic: true, downgraded Bouncing ranks are noise
Velocity gate contribVelocity < 0.03 → lowVelocity: true, excluded from IMMEDIATE No momentum behind the move
Trader count floor <10 traders → SKIP IMMEDIATE Single whale risk
Max leverage check max leverage < 10x → SKIP Not worth the limited position sizing

See references/quality-filters.md for implementation details and real-world examples.

Architecture

┌────────────────────────────────────┐
│ Cron: every 60 seconds             │
├────────────────────────────────────┤
│ scripts/emerging-movers.py         │
│ • Loads scan history from JSON     │
│ • Fetches leaderboard (1 API call) │
│ • Parses top 50 markets            │
│ • Compares with previous scans     │
│ • Detects signals + v3.1 filters   │
│ • Saves updated history            │
│ • Outputs JSON with alerts         │
├────────────────────────────────────┤
│ Agent reads output:                │
│ • IMMEDIATE alerts → evaluate now  │
│ • Deep climbers → queue for review │
│ • No alerts → silent               │
└────────────────────────────────────┘

Files

File Purpose
scripts/emerging-movers.py Scanner script
emerging-movers-history.json Auto-managed scan history (last 60 scans)
max-leverage.json Optional: asset max leverage reference

Output

See references/output-schema.md for the complete JSON schema.

Key top-level fields: alerts[], topMovers[], immediateMovers[], deepClimbers[], scanCount, timestamp.

Per-alert fields: asset, direction, rank, prevRank, contribution, traderCount, reasons[], contribVelocity, isImmediate, isDeepClimber, erratic, lowVelocity.

Cron Setup

*/1 * * * * python3 scripts/emerging-movers.py

Agent Response Logic

  • isImmediate: true + erratic: false + lowVelocity: false → Evaluate immediately for entry via Scanner
  • isDeepClimber: true → Queue for next scanner run
  • erratic: true or lowVelocity: true → Log but do not act
  • No alerts → Silent

Companion Recipes

  • opportunity-scanner — use Scanner to deep-dive assets flagged by Emerging Movers
  • autonomous-trading — full loop integrating Emerging Movers as entry trigger
  • wolf-strategy — uses IMMEDIATE_MOVER as primary entry signal