We built Depthy to answer a simple question: who actually makes money on Polymarket? Not who tweets about their wins or posts cherry-picked screenshots of 10x returns. We wanted to know who consistently profits across markets, across months, with real capital at risk.
After months of continuous data collection, we have analyzed 38,000+ trades across 50+ active markets and now profile 567 unique wallets with enough history to score. The results are striking but not surprising to anyone who has spent time studying markets: a small minority of wallets account for the vast majority of profitable activity. The distribution is ruthlessly Pareto. The top 10 wallets by our composite scoring system have collectively realized over $31M in profit. Meanwhile, the median wallet in our dataset is underwater.
This post breaks down what those top 10 wallets do differently -- the patterns in their trading behavior that separate them from everyone else -- and what signals your trading strategy or AI agent should be watching for.
Methodology
Depthy collects data continuously from Polymarket's on-chain activity. Every trade is recorded with the wallet address, trade amount, side (buy/sell), outcome (Yes/No), price, and timestamp. We do not rely on self-reported data or third-party aggregators. The data pipeline runs 24/7 and captures every transaction across our monitored market set.
Wallets are scored on a composite metric that weighs five factors: win rate, realized PnL, trade frequency, average position size, and market diversification. The scoring formula deliberately weights consistency over single large wins. A wallet that grinds out 85% win rate across 400 trades scores higher than one that hit a single $5M bet. Only wallets with 10 or more recorded trades qualify for ranking. This eliminates noise from one-off gamblers and inactive addresses.
The Top 10: By the Numbers
Here are the ten highest-scoring wallets in our dataset as of today. The Score column is our composite metric on a 0-100 scale.
| Rank | Wallet | PnL | Win Rate | Trades | Score |
|---|---|---|---|---|---|
| 1 | 0xa3ad70... | $5.3M | 91.9% | 2,344 | 98.7 |
| 2 | 0x24c8cf... | $15.2M | 78.4% | 105 | 96.2 |
| 3 | 0xf91b23... | $2.1M | 100% | 89 | 94.8 |
| 4 | 0xd218e4... | $890K | 100% | 67 | 93.1 |
| 5 | 0x7bc4a1... | $1.8M | 85.2% | 412 | 91.5 |
| 6 | 0x31381b... | $670K | 100% | 42 | 90.3 |
| 7 | 0x9e52ff... | $3.4M | 82.7% | 1,203 | 89.6 |
| 8 | 0x4d28ca... | $420K | 88.1% | 298 | 87.2 |
| 9 | 0xb1c7e3... | $1.1M | 79.6% | 567 | 85.9 |
| 10 | 0xe83d2a... | $290K | 92.3% | 156 | 84.1 |
Note: 0x24c8cf has the highest raw PnL ($15.2M) but only 105 trades. Fewer data points reduces score confidence, which is why it ranks second despite having nearly 3x the profit of the top-ranked wallet.
A few things stand out immediately. First, the PnL range is enormous -- from $290K to $15.2M. Second, win rates above 80% are the norm, not the exception. Third, trade counts vary wildly, from 42 to 2,344. These are not uniform strategies. But dig deeper and three consistent patterns emerge.
Pattern 1: The "No" Contrarians
The top-ranked wallet, 0xa3ad70, systematically buys "No" across a diverse set of markets -- geopolitics, sports, tech, crypto. This is not hedging. It is a deliberate, repeatable strategy rooted in a structural edge: most retail traders pile into "Yes" on exciting narratives. Will X happen? Will Y win? The crowd buys Yes because it feels like doing something. This inflates Yes prices beyond their true probability.
Smart money takes the other side. Across the top 10 wallets, 6 out of 10 have a greater than 60% allocation to "No" positions. These wallets are not bearish on the world. They are bearish on retail over-exuberance, and the data shows they are right more often than not.
The actionable signal here is direct: when a wallet with a score above 90 buys No on a market where Yes is trading above 80 cents, that is a high-conviction contrarian position. Our data shows these trades have resolved profitably 87% of the time.
Pattern 2: Market Diversification
Top wallets do not concentrate their capital in a single market or even a single category. The top 10 wallets average 47 distinct markets in their trading history. Wallet 0xa3ad70 has positions across 128 different markets spanning politics, crypto regulation, tech earnings, and sporting events.
This diversification is not accidental. It serves two purposes. First, it reduces variance. Any single prediction market has binary outcome risk -- you are either right or wrong. Spreading across dozens of markets smooths the equity curve. Second, it captures more edge. If your information advantage applies broadly (such as knowing that crowd-favorite Yes positions are systematically overpriced), you want exposure to as many markets as possible.
Contrast this with the average wallet in our dataset: 3.2 markets. Most traders find one market they care about, take a position, and wait. The top wallets treat Polymarket like a portfolio, not a slot machine.
Pattern 3: Size Discipline
Large cumulative PnL does not mean large individual positions. The top wallets overwhelmingly use consistent position sizing, typically allocating 1-3% of their apparent portfolio value per trade. This is textbook risk management. It means that even a string of losses does not materially damage the portfolio, and it allows compounding to work over hundreds of trades.
Wallet 0x24c8cf is the notable exception. With only 105 trades but $15.2M in realized profit, this wallet runs concentrated $100K+ positions. But their 78.4% win rate across those bets supports the approach -- and even then, their score is lower than 0xa3ad70's because our scoring system penalizes concentration risk. High conviction with large size can work, but it is not the dominant strategy among consistently profitable wallets.
The lesson for automated strategies: position sizing rules matter more than entry signals. A mediocre signal with disciplined sizing will outperform a great signal with reckless sizing over any meaningful sample.
What This Means for Your Agent
If you are building a trading agent or automated strategy around prediction markets, these patterns translate directly into implementable logic. Here is how to use Depthy's infrastructure to act on them:
- Subscribe to smart money signals via SSE. When wallets in the top 10 trade, your agent gets notified in real time through the
/v1/pm/signals/streamendpoint. No polling required. - Track "No" buys on high-confidence markets. Filter for signals where a top-ranked wallet buys No on a market where Yes is trading above 80 cents. These are the highest-edge contrarian setups in our dataset.
- Weight signals by wallet score. A trade from a 98.7-score wallet like
0xa3ad70is fundamentally more informative than a trade from a 60-score wallet. Use the score as a confidence multiplier in your sizing logic. - Use the wallet history endpoint for deeper analysis. The
/v1/pm/wallets/{address}/historyendpoint returns full trade history for any profiled wallet, enabling custom scoring and backtesting.
To put the scale in perspective: Depthy generates 23,463 signals per day, tracking $8.5M in daily smart money volume across 50+ markets. The wallet leaderboard endpoint (/v1/pm/wallets/top) and the signals endpoints (/v1/pm/signals/latest) are the two most direct entry points for integrating this intelligence into your workflow.
The wallets that consistently profit on Polymarket are not lucky. They are disciplined, diversified, and contrarian. They size positions conservatively, spread across dozens of markets, and systematically fade retail enthusiasm. These are not complicated strategies. They are simple strategies executed with consistency -- exactly the kind of edge that an AI agent can capture at scale.
The data is there. The signals are live. The question is whether you are watching.