Original Title:I Analyzed 112,000 Polymarket Wallets. Here's What Separates the Top 1% from Everyone Else, Author: darkzodchi(@zodchiii)
Compiled by | Odaily Planet Daily (@OdailyChina); Translator | Asher (@Asher_ 0210)
After systematically sorting and analyzing on-chain data from over 112,000 Polymarket wallets spanning 6 months, a rather intuitive yet surprising result emerged. Approximately 87.3% of users ultimately lost money trading on the platform.
This analysis covered multiple key dimensions, including every on-chain transaction record, trading volume, win rate, profit and loss, types of markets participated in, entry timing, and position size. The entire data processing took 3 weeks, and the final conclusion differs from many people's intuition.
Many believe that top players in prediction markets often possess some distinct advantage, such as access to insider information or the use of obscure, sophisticated computational models. However, the data shows this is not the case. The top 1% of elite players consistently and persistently do a few things right and repeat them. The other 99% of users often do the exact opposite and then wonder why their capital continues to dwindle.
Polymarket's Leaderboard is Actually Highly Misleading
If you open Polymarket's leaderboard now and sort by profit (PnL), you'll notice some anomalies. For example, the top-ranked wallet has only 22 total positions; the fourth-ranked wallet has only 8 trades; and the eighth-ranked wallet has even made just 1 bet, yet still ranks in the historical top ten.
These addresses can hardly be called genuine traders. In many cases, they are simply whales making a one-time bet of over $5 million on a single event and happening to be right; or they could be individuals with informational advantages, or a combination of both. But in any case, data from just a few trades offers almost no learnable trading rules. This outcome is more like flipping a coin with massive capital size rather than a replicable strategy.
Therefore, the first step of the analysis was to filter out this noisy data, retaining only samples with true statistical significance. The screening criteria included the following aspects:
- At least 100 settled positions, ensuring the sample size is statistically significant;
- Active trading for no less than 4 months, excluding accounts that won on luck just once;
- Participation in at least 2 different markets, avoiding reliance on a single event;
- Total trading volume exceeding $10,000, ensuring participants have genuinely invested capital.
Under these conditions, the initial 112,000 wallets were filtered down to approximately 8,400 wallet addresses with sufficient data value. These 8,400 addresses constitute the truly meaningful dataset for study, not the "hero accounts" on the leaderboard that made a few trades and earned millions. The common feature of these addresses is sustained trading and stable data, making it easier to observe real behavioral patterns.
Interestingly, after the screening was completed, the image of the truly consistently performing traders was completely different from those on the leaderboard. They are not prominent, and most people have never even heard their names. Their profit scale is typically between $50,000 and $500,000, not tens of millions of dollars.
But what is truly worth paying attention to is not how much money they made, but the trading process and methods behind it. Because what can truly be replicated is never the result, but the process.
Three Common Misconceptions to Debunk
Misconception 1: Top traders have win rates between 80% and 90%
This is not true. According to the filtered data sample, rather than the whale accounts on the leaderboard that made a fortune on one bet, the win rates of truly long-term profitable wallets are mostly only between 55% and 67%. That is, even top traders are wrong in a significant portion of their trades. For example, one address has completed over 900 settled positions, accumulating $2.6 million in profit, but its win rate is only 63%. In other words, he was wrong in over a third of his bets, yet still earned huge profits in the prediction market.
An obsession with win rate is often the easiest trap for novice accounts to fall into. Many newcomers like to buy contracts priced at $0.90 because it seems "very safe." The probability of YES is already 90%, seeming almost certain, so they buy at $0.90, and if the event happens, they only profit $0.10. But if they are wrong just once, they lose $0.90 directly, a risk-reward ratio of 9 to 1. This pattern, repeated enough times, quickly depletes account funds. In the dataset, this scenario has recurred across hundreds of addresses.
Misconception 2: The strongest traders trade every market
The reality is恰恰相反 (qià qià xiāng fǎn - exactly the opposite). The best-performing wallets typically participate in a maximum of three categories of markets, with most specializing in just one or two areas. Some addresses only make predictions related to cryptocurrency events; some only participate in weather-related markets; there's even one address that almost exclusively trades questions like "Will Bitcoin reach a certain price by Friday?"
In prediction markets, over-diversification often means a decline in judgment quality. General participants tend to perform averagely, while highly focused participants are more likely to profit consistently.
Misconception 3: Speed is everything
This is only true in very rare cases. For example, certain 15-minute settlement crypto markets do require quick reactions. But in the vast majority of markets, top traders do not win by relying on speed. Their more common practice is to gradually build positions over days or even weeks. They are not in a hurry to compete on click speed but patiently wait for significant price deviations. When the price deviates sufficiently, even if the market takes two weeks to correct, the overall mathematical expectation is still in their favor.
Five Trading Patterns Worth Learning
Pattern 1: Trading against extreme sentiment
In the entire dataset, this is the most obvious and stable profit signal. Among the filtered 8400 wallets, this behavior is almost the primary indicator of whether an account is profitable long-term.
When a contract is pushed up to 88% by market sentiment, many top wallets start selling YES; conversely, when the price drops to around 12%, they begin buying gradually. Of course, this is not blind contrarian operation, nor are they opposing the market just for the sake of it. They only enter the market on a large scale when they judge that market sentiment is clearly overreacting.
The effectiveness of this strategy is related to a classic phenomenon known as the "favorite-longshot bias." This phenomenon was first discovered in horse race betting research in the 1940s and appears in almost all markets where betting is involved. Simply put, people tend to overestimate outcomes that "seem almost certain to happen" and underestimate low-probability events.
Further statistics also found that the average entry price for the top 50 most profitable wallets typically deviates by 6% to 11% from the market consensus probability. They do not participate in bets at 50/50 odds but patiently wait for the odds to be明显 (míngxiǎn -明显) in their favor before entering. This trading style might seem boring, but it is stable and highly profitable in long-term data.
Pattern 2: Position sizing very close to the Kelly Criterion
Comparing the position sizes of the top 200 wallets by profit with the "implied edge" they faced at the time reveals a very clear correlation. In other words, they do not bet arbitrarily; their bet size almost changes in proportion to the size of the advantage they believe they have. That is, when they believe the edge is large, the position size increases significantly; with a smaller edge, they only bet a smaller amount; if there is no clear edge, they simply do not trade.
Whether these traders have actually read about the Kelly Criterion or simply developed this intuition through long-term losses and实战 (shízhàn - practical experience) is hard to determine. But mathematically, their behavior is very close to the Kelly Criterion.
The Kelly Criterion is usually written as: f* = (p × b − q) / b, where: p represents the probability the trader believes the event will actually occur; q = 1 − p; b represents the odds payout ratio (potential profit ÷ risk cost).
Take a simple example: suppose a trader judges an event has a 60% chance of occurring, and the market price is $0.45. The payout ratio is: b = (1 / 0.45) − 1 ≈ 1.22. Substituting into the formula gives: f* = (0.60 × 1.22 − 0.40) / 1.22 ≈ 0.272. This means the full Kelly strategy suggests betting 27% of the capital on this trade.
However, this approach carries extremely high risk in actual trading, with very high volatility, potentially dragging the account into huge drawdowns in a short time. From the data, truly profitable wallets usually use a more conservative version, approximately close to a quarter Kelly. In other words, if the full Kelly suggests betting 27%, they typically only bet around 7%.
In the most confident trade opportunities, the position might be increased to 12% to 15%; medium-confidence opportunities usually only allocate 2% to 5% of the capital; and they often choose not to participate in markets without a clear advantage. In contrast, losing accounts usually fall into two extremes. Either they bet 80% of their capital on one trade, relying entirely on luck; or they spread $10 across forty or fifty markets, thinking they are "diversifying risk." But in reality, this is more like constantly paying fees, making the account look busy.
Pattern 3: Extreme focus and specialized trading
Dividing the 112,000 wallets according to the categories of markets they participated in (including crypto markets, political events, sports events, weather, geopolitics, entertainment, and science) reveals very clear differences. The analysis shows:
- Wallets participating in only 1 to 2 categories had an average PnL of approximately +$4200;
- Wallets participating in 3 to 4 categories had an average PnL of approximately -$380;
- Wallets participating in 5 or more categories had an average PnL of approximately -$2100.
This relationship shows an almost clear linear trend. The more market categories participated in, the higher the probability of loss.
Different categories of prediction markets rely on completely different information systems. Crypto markets are often influenced by exchange fund flows, whale addresses, funding rates, etc.; political markets rely on polling data, grassroots information, congressional schedules, etc.; while weather markets rely more on NOAA weather models, atmospheric data, and satellite observations.
Two cases are particularly representative. Case One: Wallet A only trades Bitcoin 15-minute settlement prediction markets, never participating in other types, e.g., "Will BTC be above a certain price in the next 15 minutes." This address completed 502 predictions with a win rate of 98%, accumulating about $54,000 in profit. Its advantage is actually very simple: continuously monitoring Binance order book depth and trading quickly when Polymarket's price lags by 10 to 30 seconds. That is, an information gap of just十几秒 (shí jǐ miǎo - over ten seconds) was exploited hundreds of times.
Case Two: Wallet B only participates in weather-related markets. The strategy is also straightforward: reading NOAA's daily publicly released temperature forecast data and comparing it with Polymarket's market pricing. If the market price shows a significant deviation from these supercomputer predictions optimized over decades, they enter the trade directly. In the New York temperature prediction market alone, this address achieved an accuracy rate of 94%.
It's important to emphasize that these people are not geniuses. The key is that they found a niche they understand better than the average Polymarket participant and then repeatedly leveraged this advantage. They didn't frequently change strategies, nor did they suffer from FOMO due to market hype. They just executed the same logic over and over again around the same advantage.
Pattern 4: Trading price movements, not event outcomes
Most Polymarket users trade in a very simple way: buy a contract and hold it until event settlement, either profiting or losing—a typical binary outcome. But the approach of top wallets is completely different. Often, they buy at $0.40 and exit directly by selling when news or market sentiment pushes the price to $0.65. They don't care whether the event actually happens in the end; as long as the price reflects the new information, they complete the trade and exit.
In the dataset, some of the best-performing addresses don't even have any settled positions. They never hold contracts until final settlement but continuously engage in波段交易 (bōduàn jiāoyì - swing trading) based on price mismatches. Statistics show that the average holding period for top wallets is typically only 18 to 72 hours, while wallets in the bottom 50% by profitability often hold until settlement, sometimes for weeks or months.
This doesn't mean holding until settlement is always wrong. Sometimes, when conviction is very high, holding long-term is indeed the better strategy. But overall, the data shows that top wallets use their capital more actively and flexibly than most people imagine. They are not passive bettors but genuine traders.
Pattern 5: Always avoiding breaking news
We intuitively think that the sharpest money should enter the market immediately when sudden events occur, such as military conflicts, election results, corporate executive resignations, and other major news. But the data shows that top wallets often actively avoid the period immediately after news breaks. They usually wait for emotional funds to flood the market first, causing significant short-term price volatility, and only start trading after market sentiment stabilizes.
From the entire dataset, a very clear pattern is: the best trading opportunities often occur before the market notices an event or after the market has overreacted emotionally. And when everyone is discussing the same thing, it is often the worst time to enter. At this point, market prices are usually highly efficient, and the advantage that can be gained is minimal.
Five Operational Suggestions
Choose a niche and focus long-term
Whether it's crypto, politics, weather, or sports, anything is fine, but you must choose the field you are most familiar with. For the next至少三个月 (zhìshǎo sān gè yuè - at least three months), trade only this category of markets. No exceptions, and don't participate in other hot events on a whim. Even "just placing a casual bet on the election" can easily break your original judgment system.
Record every prediction
Before each trade, write down several key data points, including your judged true probability, current market price, expected edge, and planned position size. Review after accumulating over 50 trades. For example, if predictions marked as 70% probability actually hit close to 70%. If there's a significant deviation, it means there is bias in probability judgment, and calibration is necessary before increasing position sizes.
Manage positions to approximate quarter-Kelly
First calculate the theoretical position size given by the Kelly Criterion, then divide by 4 for the actual position. This number will often seem very small, but this is key to controlling risk. The result of over-leveraging is almost always one—account blow-up.
Only trade when the edge is sufficiently clear
If the expected edge is below 8% to 10%, just skip it. Even if the opportunity seems tempting, learn to wait. The best-performing wallets in the data typically only execute 2 to 3 trades per market category per week. Trade quality is far more important than trade quantity.
Persist in recording and reviewing
Create a comprehensive trading spreadsheet to record every trade, its outcome, and any issues that arose. Wallets that show continuous long-term improvement almost always systematically review their mistakes;而那些 (ér nàxiē - and those) stagnant or continuously losing accounts often just repeat the same mistakes, attributing the results to bad luck.
