| name | roach-strategy |
| description | ROACH v1.0 — Striker-only scanner. Stalker mode disabled entirely. Tests whether Stalker adds any value or is pure drag. Only trades violent FIRST_JUMP explosions backed by 1.5x volume. Will have long stretches of silence — that patience is the edge. All hardened gates preserved. XYZ banned. Leverage 7-10x. DSL exit managed by plugin runtime via runtime.yaml. |
| license | MIT |
| metadata | {"author":"jason-goldberg","version":"2.0","platform":"senpi","exchange":"hyperliquid","config_source":"striker-only-experiment","requires":["senpi-trading-runtime"]} |
🪳 ROACH v1.0 — Striker Only. Stalker Disabled.
Cockroaches survive anything. ROACH survives by not trading when there's no explosion.
⛔ CRITICAL AGENT RULES — READ BEFORE ANYTHING ELSE
RULE 1: Install path is /data/workspace/skills/roach-strategy/
The skill MUST be installed to exactly this path.
RULE 2: MAX 3 POSITIONS — check before EVERY entry
Before opening ANY position, call strategy_get_clearinghouse_state and count open positions. If positions >= 3, SKIP.
RULE 3: Scanner output is AUTHORITATIVE — never override from memory
The scanner is the single source of truth for all trading parameters.
RULE 4: Verify runtime is installed on every session start
Run openclaw senpi runtime list. Runtime must be listed. The position tracker and DSL exit are handled by the plugin runtime.
RULE 5: Never retry timed-out position creation
If create_position times out, check clearinghouse state first.
RULE 6: Never modify your own configuration
No adjustments to leverage, margin, scoring, or any parameter.
RULE 7: Do NOT enable Stalker mode
Stalker is disabled by design. This is an experiment. Do not re-enable it, do not build your own accumulation detection, do not invent alternative entry logic. If the scanner outputs "stalkerDisabled": true, that means it's working correctly. Long periods of silence with zero trades are EXPECTED and CORRECT.
The ROACH Experiment
Hypothesis: Stalker mode is pure drag. Fox v1.0 produced 17 Stalker trades at score 6-7 with a 17.6% win rate and -$91.32 net P&L. The one Striker signal (ZEC LONG, score 11) was the only explosive entry worth taking. If ROACH outperforms or matches the other variants with dramatically fewer trades, Stalker should be permanently demoted or removed from the scanner.
What this means in practice: ROACH will be quiet. Striker signals require FIRST_JUMP or IMMEDIATE_MOVER (10+ rank jump from #25+), rank jump >= 15 OR velocity > 15, volume >= 1.5x, score >= 9 with 4+ reasons. In choppy markets, this might fire 0-1 times per day. In trending markets with SM explosions, maybe 2-3.
The question: Does fewer + higher quality beat more + mixed quality? Polar proved it at the macro level (29 trades, +28.1% vs Ghost Fox's 1,078 trades, -58.5%). ROACH tests it at the signal-type level within the same scanner framework.
Single Entry Mode
STRIKER (Explosion) — Score >= 9, min 4 reasons
- FIRST_JUMP or IMMEDIATE_MOVER (10+ rank jump from #25+)
- Rank jump >= 15 OR contribution velocity > 15
- Raw volume >= 1.5x of 6h average
- Score >= 9 with at least 4 distinct reasons
- 4H trend aligned, XYZ banned, asset not in cooldown
STALKER — DISABLED
- Scanner still builds scan history (needed for Striker's rank jump detection)
- Stalker signals are never generated or emitted
- Output always contains
"stalkerDisabled": true and "stalkerSignals": []
Exit Management
DSL exit is handled by the plugin runtime via runtime.yaml. The position_tracker scanner auto-detects position opens/closes on-chain. See runtime.yaml for configuration details.
Entry flow:
- Scanner outputs Striker signal
- Verify positions < 3
- Verify exchange max leverage >= 7
- Call
create_position
- Send ONE notification: position opened
position_tracker detects the new position automatically- Plugin DSL monitor applies trailing stop-loss protection
Monitor positions:
openclaw senpi dsl positions — list all DSL-tracked positionsopenclaw senpi dsl inspect <ASSET> — full position details
Runtime Setup
Step 1: Set your strategy wallet address in runtime.yaml:
sed -i 's/${WALLET_ADDRESS}/<STRATEGY_WALLET_ADDRESS>/' /data/workspace/skills/roach-strategy/runtime.yaml
Replace <STRATEGY_WALLET_ADDRESS> with the actual wallet address.
Step 2: Set telegram chat ID for notifications:
sed -i 's/${TELEGRAM_CHAT_ID}/<CHAT_ID>/' /data/workspace/skills/roach-strategy/runtime.yaml
Replace <CHAT_ID> with the actual Telegram chat ID.
Step 3: Install the runtime:
openclaw senpi runtime create --path /data/workspace/skills/roach-strategy/runtime.yaml
Step 4: Verify:
openclaw senpi runtime list
Bootstrap Gate
On EVERY session start, check config/bootstrap-complete.json. If missing:
- Read the senpi-trading-runtime skill:
cat /data/workspace/skills/senpi-trading-runtime/SKILL.md — this provides all CLI commands for runtime management and DSL position inspection.
- Verify Senpi MCP
- Set wallet in runtime.yaml:
sed -i 's/${WALLET_ADDRESS}/ACTUAL_ADDRESS/' /data/workspace/skills/roach-strategy/runtime.yaml
- Set Telegram in runtime.yaml:
sed -i 's/${TELEGRAM_CHAT_ID}/CHAT_ID/' /data/workspace/skills/roach-strategy/runtime.yaml
- Install runtime:
openclaw senpi runtime create --path /data/workspace/skills/roach-strategy/runtime.yaml
- Verify runtime installed:
openclaw senpi runtime list
- Remove old DSL cron (if upgrading): run
openclaw crons list, delete any cron containing dsl-v5.py via openclaw crons delete <id>
- Create scanner cron (90s, main)
- Write
config/bootstrap-complete.json
- Send: "🪳 ROACH v1.0 online. Striker only. Stalker disabled. Waiting for explosions. Silence = no explosion."
If bootstrap exists, still verify runtime and scanner cron on every session start.
Expected Behavior
| Metric | Expected |
|---|
| Trades/day (chop) | 0-1 |
| Trades/day (trending) | 1-3 |
| Days with zero trades | Common and correct |
| Win rate | Higher than Stalker variants (60-70% target) |
| Avg winner | Larger (explosive entries with 1.5x volume) |
| Avg loser | Similar to other variants (DSL-managed) |
| Total P&L | Hypothesis: matches or beats Stalker variants with far fewer trades |
SILENCE IS CORRECT. If ROACH goes 24-48 hours without a trade, that means there were no FIRST_JUMP explosions worth taking. That's the experiment working, not a bug.
Risk Management
| Rule | Value |
|---|
| Max positions | 3 |
| Max entries/day | 6 |
| Leverage | 7-10x |
| Daily loss limit | 10% |
| Per-asset cooldown | 2 hours |
| XYZ equities | Banned |
| Stalker mode | DISABLED |
Notification Policy
ONLY alert: Position OPENED (Striker signal with score + reasons), position CLOSED (P&L + reason), critical error.
NEVER alert: Scanner found no Striker signals (this is normal and expected), any reasoning about whether to enable Stalker, any analysis of market conditions.
Files
| File | Purpose |
|---|
scripts/roach-scanner.py | Striker-only scanner (Stalker disabled in run()) |
scripts/roach_config.py | Config helper |
config/roach-config.json | Config (Stalker settings present but unused) |
runtime.yaml | Runtime config for plugin (DSL exit + position tracker) |
License
MIT — Built by Senpi (https://senpi.ai).
Source: https://github.com/Senpi-ai/senpi-skills
