ClawFomo Bot Skill

Play ClawFomo (Fomo3D on Base) with algorithmic strategy.

Overview

Automated player for the ClawFomo game on Base chain. Evolved through 5 strategy iterations in 2 hours based on game theory research, on-chain data analysis, and live P&L feedback.

Strategy Evolution

Version	Strategy	Keys/Bid	Result
V1	Aggressive	25	Massive losses — bonding curve destroys you
V2	Capped (3 bids)	5	Lost — cap meant we could never defend
V3	Cumulative EV	5	Won 3/5 rounds but still net negative — 5 keys too expensive
V4	Vulture (1 key, capped)	1	Right idea, wrong cap — folded every contested round
V5	Smart Vulture	1	Dividend-aware EV, opponent profiling, whale dodging

V5 — Smart Vulture Strategy

Core principles learned from game theory + on-chain analysis:

1. 1 key per bid — same win probability as 25, fraction of the cost
2. No arbitrary bid caps — pure EV math controls all decisions
3. Opponent profiling — track active bidders, dodge known whales
4. Dividend-aware EV — factor earned dividends into round profitability
5. Activity detection — wait for quiet moments before entering (30s minimum)
6. Round selection — skip rounds with 4+ opponents or whale activity
7. Frontrun protection — reject if cost spikes >50% between calculation and execution

Why 1 Key?

The game rewards the last buyer, regardless of how many keys they bought. Buying 5 keys costs 5x more but gives the same win probability. The only benefit of more keys is dividends, but the math doesn't justify the cost increase.

With 1 key at ~5K CLAWD, you can defend 10+ times for less than one old 5-key bid cost (~50K).

EV Calculation

projectedPot = currentPot + (bidCost × 0.65)  // 65% of bid reaches pot
projectedWin = projectedPot × 0.50             // winner gets 50%
dividendEstimate = (ourKeys / totalKeys) × avgBidCost × 0.225 × expectedBids
netEV = projectedWin + dividendEstimate - totalRoundSpend - thisBidCost

// Only bid when netEV > 0

Entry Conditions (ALL required)

• In snipe window (timer ≤ 120s)
• Timer > 5s (TX needs time to land)
• Pot:cost ratio ≥ 4x
• No known whales in round
• ≤ 4 active opponents
• Round quiet for ≥ 30s (first entry only)
• Net EV > 0 after all round spending

Game Mechanics

• Contract: 0x859e5cb97e1cf357643a6633d5bec6d45e44cfd4 (Base)
• Token: CLAWD (0x9f86dB9fc6f7c9408e8Fda3Ff8ce4e78ac7a6b07)
• Timer: 300s max, resets on buy
• Anti-snipe: Buy within 120s extends timer TO 120s
• On buy: 10% burned, 25% of rest → dividends, 65% → pot
• On win: 50% pot → winner, 20% burned, 25% → key holders, 5% → team
• Key price: 1000 + totalKeys × 110 CLAWD (bonding curve)

Scripts

Script	Purpose
scripts/play-v5.mjs	Live bot — Smart Vulture strategy
scripts/status.mjs	Check current round state
scripts/check-pnl.mjs	P&L tracking for cron monitoring

Usage

source ~/.axiom/wallet.env
export NET_PRIVATE_KEY

# Live play
node scripts/play-v5.mjs

# Dry run
node scripts/play-v5.mjs --dry-run

# Custom params
node scripts/play-v5.mjs --ratio 6 --quiet 60 --poll 3000

# Check status
node scripts/status.mjs

# P&L check (for cron)
node scripts/check-pnl.mjs

Key Lessons

1. Bonding curves are exponential traps — buying more keys costs quadratically more
2. The 10% burn is the house edge — every bid loses 10% immediately. You MUST be selective.
3. Arbitrary caps lose money — if the math says bid, bid. If it says stop, stop. No in-between.
4. 1 key = optimal — same win probability, minimum cost, maximum flexibility
5. Opponent awareness matters — whales will outspend you. Don't fight them.
6. Dividends are real income — factor them into every decision
7. Patience is the edge — most players overbid. The patient vulture wins.

Dependencies

• viem (Ethereum client)
• NET_PRIVATE_KEY environment variable
• Base RPC (defaults to https://mainnet.base.org)

Open Source

Part of axiom-public — open-source agent tools for on-chain operations.