RWA Alpha v1.1 — Real World Asset Intelligence Trading Engine

Risk Warning: This strategy trades real tokens on-chain. Capital loss may occur due to RWA liquidity risk, macro prediction errors, smart contract bugs, or slippage. Start in paper mode. Deploy live only with capital you can afford to lose.

File Structure

RWAAlpha/
├── skill.md              ← This file (AI agent instructions)
├── config.py             ← All tunable parameters (edit this, not rwa_alpha.py)
├── rwa_alpha.py          ← Strategy engine (DO NOT EDIT unless fixing bugs)
├── dashboard.html        ← Web dashboard UI (http://localhost:3249)
├── .gitignore            ← Excludes state/ and runtime files
└── state/                ← [auto-generated at runtime]
    ├── positions.json    ← Open positions
    ├── trades.json       ← Completed trade history
    ├── signals.json      ← Signal log (last 200)
    ├── macro_events.json ← Detected macro events (last 100)
    └── yield_snapshots.json ← Yield ranking snapshots

onchainos

Startup Protocol

Step 1: Pre-flight Check

# Verify onchainos CLI
~/.local/bin/onchainos --version

# Verify wallet login (live mode only)
~/.local/bin/onchainos wallet status
~/.local/bin/onchainos wallet addresses --chain 1

Step 2: Configure via config.py

config.py

• MODE = "paper"

  or

  "live"

• PAUSED = False

  to enable trading

• STRATEGY_MODE = "macro_trader"

  (or

  yield_optimizer

  /

  full_alpha

  )

• TOTAL_BUDGET_USD = 1000

  (total USDC allocation)

• BUY_AMOUNT_USD = 100

  (per-trade size)

• ENABLED_CHAINS = ["ethereum"]

  (add

  "solana"

  if desired)

RWA_MODE

RWA_STRATEGY_MODE

RWA_BUDGET

RWA_BUY_AMOUNT

RWA_CHAINS

• Set

  ANTHROPIC_API_KEY

  env var to enable

• LLM_ENABLED = True

  in config.py (default)
• Uses Haiku (~$0.005/call) only for ambiguous headlines
• Set

  LLM_ENABLED = False

  to run purely on keyword matching

Step 3: Launch

cd /path/to/RWAAlpha && python3 rwa_alpha.py

http://localhost:3249

Architecture

┌────────────────────────────────────────────────────────────┐
│                  RWA ALPHA v1.1 ENGINE                      │
├────────────────────────────────────────────────────────────┤
│                                                            │
│  PERCEPTION LAYER (runs every CHAIN_POLL_SEC = 60s)       │
│  ├─ Price Cache: onchainos token price-info / advanced-info│
│  ├─ NewsNow API: financial headlines from 3 sources        │
│  │   └─ wallstreetcn, cls, jin10                          │
│  ├─ Polymarket API: prediction market probabilities        │
│  ├─ Gold price tracking: PAXG/XAUT price changes          │
│  └─ Volume spike detection: vol/MC ratio on gov tokens     │
│                                                            │
│  COGNITION LAYER                                           │
│  ├─ Macro Event Detection (3-layer)                        │
│  │   ├─ L1: keyword match (fast, free)                    │
│  │   ├─ L2: LLM confirm/override ambiguous (Haiku ~$0.005)│
│  │   ├─ L3: LLM classify unmatched RWA headlines          │
│  │   └─ 15 event types in MACRO_PLAYBOOK                  │
│  ├─ Sentiment Scoring (keyword-based, news + on-chain)     │
│  │   └─ 60% news weight + 40% on-chain weight            │
│  ├─ Yield Ranking (alpha_score for asset-backed tokens)    │
│  │   └─ NAV discount 30% + sentiment 25% + liquidity 25% │
│  └─ Signal Composition → risk gate → execute               │
│                                                            │
│  EXECUTION LAYER                                           │
│  ├─ onchainos swap quote → onchainos swap swap               │
│  ├─ onchainos wallet contract-call (TEE, requires user confirmation) │
│  ├─ Risk checks: daily limit, session stop, cooldown,      │
│  │   position concentration, category limit, liquidity     │
│  └─ Dual exit system: asset-backed vs governance tokens    │
│                                                            │
└────────────────────────────────────────────────────────────┘

RWA Token Universe (config.py → RWA_UNIVERSE)

Three Strategy Modes

1. Yield Optimizer (

yield_optimizer

)

• Only trades asset-backed tokens (USDY, OUSG, sDAI, bIB01, PAXG, XAUT, USDe)
• Focus: NAV discount entry + yield rotation between best alpha_score
• Ignores governance tokens entirely
• Lowest risk, fewest trades

2. Macro Trader (

macro_trader

) — Recommended

• Trades both asset-backed AND governance tokens
• Responds to macro events: Fed decisions, CPI, gold breakouts, SEC rulings
• Moderate conviction threshold (0.55)

3. Full Alpha (

full_alpha

)

• All strategies active: macro + yield rotation + governance momentum
• Volume spikes on ONDO/CFG/MPL/PENDLE/PLUME/OM/GFI/TRU trigger entries
• Highest trade frequency, highest risk

Macro Event Playbook (15 Events)

fed_cut_expected

fed_cut_surprise

fed_hold_hawkish

fed_hike

cpi_hot

cpi_cool

gold_breakout

gold_selloff

geopolitical_escalation

ondo_yield_increase

maker_dsr_up

sec_rwa_positive

sec_rwa_negative

credit_expansion

credit_tightening

1. Keywords (free, instant) — regex match on headlines from NewsNow (wallstreetcn, cls, jin10)
2. LLM classification (Haiku, ~$0.005/call) — confirms ambiguous keyword matches + catches headlines keywords miss. Only fires when: keyword conviction is in the 0.55-0.80 band, OR no keyword matched but headline contains RWA-relevant terms
3. Polymarket API — prediction market probabilities (e.g. rate cut > 65% → trigger)
4. On-chain price action — gold +/-2% triggers breakout/selloff, vol/MC > 10% triggers momentum

Exit System

Asset-Backed Tokens (USDY, OUSG, sDAI, bIB01, PAXG, XAUT, USDe)

• TP: NAV premium > 40 bps → sell
• SL: NAV discount > 100 bps (or PnL < -1%) → sell
• Yield Rotation: if another asset-backed token's alpha_score is 0.15+ better, sell current and buy replacement

Governance Tokens (ONDO, CFG, MPL, PENDLE, PLUME, OM, GFI, TRU)

• TP: +20% → sell
• SL: -10% → sell
• Trailing Stop: activates at +10% profit, triggers on 8% drop from peak

Portfolio-Level

• Max Drawdown: if total portfolio PnL < -8% of invested → close ALL positions

Risk Controls (config.py)

MAX_POSITIONS

MAX_SINGLE_PCT

MAX_CATEGORY_PCT

MAX_DAILY_TRADES

SESSION_STOP_USD

COOLDOWN_LOSS_SEC

MIN_LIQUIDITY_USD

MAX_NAV_PREMIUM_BPS

MIN_CONVICTION

SLIPPAGE_BUY

SLIPPAGE_SELL

onchainos CLI Commands Used

# Price data
onchainos token price-info --chain ethereum --address <token_addr>
onchainos token advanced-info --chain ethereum --address <token_addr>

# Wallet
onchainos wallet status
onchainos wallet balance --chain <chain_idx>
onchainos wallet addresses --chain <chain_idx>

# DEX trading
onchainos swap quote --chain <chain> --from <stable> --to <token> --amount <raw>
onchainos swap swap --chain <chain> --from <stable> --to <token> --amount <raw> \
  --slippage <pct> --wallet-address <addr>

# Transaction signing + broadcast
onchainos wallet contract-call --chain <chain_idx> --to <contract> --unsigned-tx <tx_data>  # requires user confirmation

# Transaction confirmation
onchainos wallet history --tx-hash <hash> --chain <chain_idx>

1

501

Dashboard

http://localhost:3249

• Portfolio allocation bars by category
• Macro pulse feed (detected events)
• Yield landscape table (ranked opportunities)
• Open positions with PnL
• Trade history
• Signal log
• Activity feed

GET /api/state

Slash Commands (for AI agent)

/rwa-alpha start

python3 rwa_alpha.py

/rwa-alpha status

/rwa-alpha stop

/rwa-alpha config

/rwa-alpha positions

/rwa-alpha trades

/rwa-alpha signals

/rwa-alpha events

Iron Rules

1. NEVER modify

   rwa_alpha.py

   to change strategy logic — edit

   config.py

   only
2. NEVER set

   MODE = "live"

   without user's explicit confirmation
3. NEVER commit state/ files to git
4. ALWAYS start in paper mode first
5. ALWAYS verify wallet login before live trading
6. ALWAYS check that

   PAUSED = False

   before expecting trades
7. If a sell fails, do NOT retry immediately — wait for cooldown
8. If portfolio drawdown triggers, ALL positions are closed — this is by design

Future: RWA Perps Split

• RWA Spot (this skill): asset-backed tokens, yield rotation, NAV arbitrage
• RWA Perps (new skill): leveraged macro bets on ONDO, CFG, MPL with funding rate arbitrage
• Shared: macro event detection, sentiment scoring, risk controls

Changelog

v1.1 (2026-04-02)

• Added: LLM headline classification (Haiku) — 3-layer detection: keyword → LLM confirm → LLM discover
• Fixed: onchainos CLI commands (

  dex quote/swap

  ,

  --wallet-address

  ,

  --chain

  )
• Fixed: sell amount decimal conversion for token amounts
• Fixed: yield rotation now buys replacement after selling
• Fixed: yield rotation timing uses interval tracking (not modulo)
• Added:

  gold_selloff

  event in MACRO_PLAYBOOK
• Added: NewsNow API integration (wallstreetcn, cls, jin10 headlines)
• Added: Polymarket API for macro event probability confirmation
• Added: Keyword-based sentiment scoring from news headlines
• Added: Composite sentiment (60% news + 40% on-chain)
• Fixed:

  interactive_setup()

  uses env vars instead of blocking

  input()

• Fixed:

  main()

  no longer requires interactive input

v1.0 (2026-04-01)

• Initial release with 3 strategy modes
• 9 RWA tokens across Ethereum + Solana (expanded to 15 in v1.1)
• 12 macro event types (expanded to 15 in v1.1)
• Dashboard on port 3249

Security & Data Trust

M07 — External Data Trust

M08 — Safe Fields for Display

Live Trading Confirmation Protocol

1. Credential Gate: Wallet must be logged in via

   onchainos wallet status

   before any trade
2. Per-Session Authorization: Live mode (

   MODE = "live"

   ) must be explicitly set by the user in config.py. Default is

   paper

   mode.

   PAUSED = True

   by default. The user must deliberately change both settings to enable live trading.
3. Autonomous Execution Warning: Once

   MODE = "live"

   and

   PAUSED = False

   , the bot executes trades autonomously based on macro signals without per-transaction user confirmation. All trades go through

   onchainos swap swap

   →

   onchainos wallet contract-call

   (TEE-signed).
4. Budget Limits: Per-trade (

   BUY_AMOUNT_USD

   ) and portfolio-level (

   MAX_PORTFOLIO_USD

   ,

   SESSION_STOP_USD

   ,

   MAX_DAILY_TRADES

   ) limits enforced in config.py
5. Risk Gates: Daily loss limit, consecutive loss cooldown, position concentration cap, category limits, and minimum liquidity checks — all prevent runaway losses