Recall

We tracked 1,356 active polymarket wallets in April. 725 traded only BTC 5 minute markets. Most run the same play: Arbitrage settlement timing. They race the oracle after the price window closes. With 700+ wallets competing on the same strategy, the edge is pure infra and speed. We expect returns to compress fast.

Do prediction markets have an optimism problem? Takers overwhelmingly buy YES at longshot prices, while makers sit on the other side and collect. Is buying NO an underutilized strategy?

4/ Follow the money. We discovered this once we stopped stopped looking at markets and started looking at trades. No existing API gave us the full picture, so we built a data pipeline from scratch that classifies 500M+ fills and reverse-engineers strategies and PnL for every wallet. Turns out the biggest edge in prediction markets isn't a better prediction, but better data infrastructure.

3/ Compare this to the "insider" narrative. Almost none of the concentrated bettors who went big during the 2024 election cycle and cleared millions, have been consistently profitable since. Meanwhile, merge arb wallets have made money every week for 18 months.

1/ Most think prediction markets are about insider knowledge, e.g. someone knows the outcome so they buy big and cash in. But analyzing 400K wallets and 500M+ fills tells a different story.

We found a wallet that does this manually on geopolitical markets. - Buys tail risk at 1-5 cents - Wins 24% of the time - Profits $290K with 50x payoff on the wins Now imagine that wallet isn't a person checking Twitter. It's an agent ingesting a structured event stream in real time and placing bets within seconds.

Can an AI agent map event feeds like MTS to mispriced Polymarket outcomes?

In theory, this trading strategy works. In practice, we saw this only work for markets with 3 outcomes. Markets with 10+ outcomes were a disaster. Convert mints YES tokens on every remaining outcome, and when most of those outcomes have no real bids, you're holding inventory the market won't buy back, blowing up your agent's theoretical arb model.

Multi-outcome markets on @Polymarket have a structural inefficiency. When YES prices across all outcomes don't sum to $1, you can buy cheap NO tokens, call CONVERT on the smart contract, get guaranteed USDC back, and sell the leftover YES tokens. Free money, right?

April conviction rewards are now live. Visit app.recall.network to see what you've earned.

You don't need to predict anything to print on @Polymarket. This wallet has a 100% win rate over six months, exploiting the 30-minute window between when the outcome is obvious and the contract settles. The most boring edge in prediction markets can also be the most profitable.

One wallet made 1.93 million trades in 76 days on @Polymarket. That's 25,000 each day, every day, for 2.5 months. Prediction markets are not "retail vs retail." This is what you're trading against. Position your agent accordingly.

Total sponsored rewards on @Polymarket hit $1M+. Track every active sponsored market on our dashboard. Point your agents at them. Easy alpha, easy rewards. recall.network/pm-reward-trac…

@andrewxhill This model looks legit!

Will Smith eating spaghetti. March 23, 2023

Intelligence is the product. Software is the distribution.

The Recall Foundation is committed to the trust and security of public blockchain networks. Last year, we contributed $100K to challenge the community to audit the @ipcdevs stack.

Accelerating financial superintelligence

Opportunity is everywhere in markets. Your job is to tell your agent where to look.