AgentStack
SKILL verified MIT Self-run

Magpie Strategy

skill-senpi-ai-senpi-skills-magpie · by Senpi-ai

>-

No reviews yet
0 installs
10 views
0.0% view→install

Install

$ agentstack add skill-senpi-ai-senpi-skills-magpie

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Magpie Strategy? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

🐦 MAGPIE v1.0 — IPO / New-Listing Event Hedge Fund

Magpie trades the new-listing event that just became Hyperliquid's biggest story: SpaceX (SPCX) did $1.4B in day-1 perp volume — 30% of all HIP-3 volume — while real exchanges ran out of shares. Magpie productizes the full arc: accumulate the pre-IPO perpetual, then ride the explosive momentum when it converts to a full equity perp. One producer script (magpie-producer.py) serves both books; the MAGPIE_LEG env var selects which.

| Book | Style | Wallet env | Runtime | Scanner | |---|---|---|---|---| | pre_listing | Accumulate IPOPs into the IPO | MAGPIE_PRE_LISTING_WALLET | runtime-pre_listing.yaml | magpie_pre_listing_signals | | graduation | Ride post-conversion momentum | MAGPIE_GRADUATION_WALLET | runtime-graduation.yaml | magpie_graduation_signals |

What makes Magpie different

  • vs. Cougar (equity long/short): Cougar trades the ongoing equity market;

Magpie trades the event — the pre-IPO ramp and the conversion pop.

  • vs. Lemur / Falcon (single agents): Lemur holds IPOPs, Falcon trades the

conversion. Magpie packages both into a two-wallet fund with its own risk controls — and reuses their exact, proven detection logic.

How an IPOP is detected (both books)

trade.xyz pre-IPO perpetuals carry a structural funding signature: very low funding (|funding| ≤ ipopFundingMaxAbs ~1e-7, a ~1% multiplier vs the ~100× standard) and a low leverage cap (≤ ipopMaxLeverageCap 5). When the company IPOs and the product converts to a standard equity perp, funding jumps ~100× and the cap lifts — Magpie keys off exactly that transition.

PRE-LISTING book — accumulate the ramp (both directions)

Each tick: fetch_ipop_universe filters the xyz board to IPOPs (funding + leverage + min-volume signature). For each, score the pre-listing trend — 4h structure sets direction, 1h confirms, Smart-Money confirms (sparse pre-listing → trend-only fallback). minScore 5. Holds up to 3 IPOPs (today ~just SPCX; auto-expands as trade.xyz lists ANTHROPIC / OPENAI / STRIPE / …). margin_pct 12%, 3x (Discovery Bounds throttle IPOP velocity).

GRADUATION book — ride the conversion (both directions)

Each tick: classify every xyz instrument IPOP-vs-STANDARD, compare to the prior-tick class cache to detect IPOP→STANDARD flips, and stamp each flip into a conversionWindowHours (72h) eligibility window. Within the window, score post-conversion momentum (minMomentumPct 3% / strongMomentumPct 8%) + SM + rising-volume bonuses, and ride the move. minScore 5. margin_pct 15%, 5x (the venue cap lifts post-conversion). The window means momentum that develops over hours/days is still tradeable, not just the single flip tick.

Execution & exit

  • both books: FEE_OPTIMIZED_LIMIT entries + exits, taker-true
  • PRE-LISTING slots 3, 3x — DSL moderate-wide (multi-day discovery): phase1 max_loss 10%, time-cuts OFF, hard_timeout 7d; phase2 10%→0 / 22%→45 / 45%→65 / 80%→80 / 130%→90
  • GRADUATION slots 3, 5x — DSL wide let-winners-run (price discovery trends hard): phase1 max_loss 12%, time-cuts OFF, hard_timeout 5d; phase2 12%→0 / 28%→45 / 55%→65 / 100%→80 / 160%→90

State (graduation book)

The graduation book persists a class-state cache (class-state-graduation.json, last-tick IPOP/STANDARD per instrument) and a conversion-window cache (conversions-graduation.json, when each flip was detected). The first tick just seeds the class cache (no flips); flips are only detected against a known prior.

XYZ handling + auth

Everything is XYZ — candle/instrument fetches pass dex="xyz". Requires user-scope auth for leaderboard_get_markets (Smart-Money confirmation). The main/xyz clearinghouse sections are two VIEWS of ONE cross-margined wallet, so get_positions() takes accountValue ONCE via max() — never sums.

Risk gates (risk.guard_rails)

| Gate | prelisting | graduation | |---|---|---| | dailylosslimitpct | 12 | 12 | | maxentriesperday | 4 | 4 | | maxconsecutivelosses | 4 | 4 | | cooldownminutes | 90 | 90 | | drawdownhaltpct | 18 | 20 | | perassetcooldownminutes | 360 | 360 | | dataretentionhours | 168 | 168 | | drawdownresetonday_rollover | true | true |

Files

| File | Purpose | |---|---| | runtime-pre_listing.yaml / runtime-graduation.yaml | Per-book runtime spec | | scripts/magpie-producer.py | Book-aware producer (IPOP discovery + conversion detection + scoring) | | scripts/magpie_config.py | Leg resolution + SenpiClient wrapper + class-state/conversion caches | | config/magpie-pre_listing-config.json / config/magpie-graduation-config.json | Per-book tunables |

Operator install

See [README.md](README.md) — two daemons (MAGPIE_LEG=pre_listing and MAGPIE_LEG=graduation) on two wallets (default 50/50), each with its own runtime YAML.

Hard rule for user-conversation Claude sessions

User-conversation Claude sessions MUST NOT call any of: create_position, close_position, edit_position, ratchet_stop_add, ratchet_stop_edit, ratchet_stop_delete, cancel_order, strategy_close, strategy_close_positions.

These tools are reserved for the producer daemon (entry path) and the DSL ratchet engine (exit path). User-conversation sessions are read-only.

License

Apache-2.0 — Copyright 2026 Senpi (https://senpi.ai)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.