Behind 27,000 Transactions: The Survival Algorithm and Illusion of High Win Rates of Polymarket Whales

Original Author: Frank, PANews

Recently, the heat around prediction markets has continued to rise, especially with the arbitrage strategies of "smart money" being held up as the gold standard. Many have begun to imitate and experiment, seemingly heralding the start of a new gold rush.

But behind this hype, how effective are these seemingly clever and logical strategies in reality? And how are they actually executed? PANews conducted an in-depth analysis of 27,000 transactions from the top ten most profitable whales on Polymarket in December to uncover the truth behind their profits.

After analysis, PANews found that while many of these "smart money" operations did execute hedging arbitrage strategies, this hedging differs significantly from the simple hedging interpreted on social media. The actual strategies are far more complex, not merely simple combinations of "yes" or "no." Instead, they leverage rules within sports events like "over/under" and "win/loss" to achieve combined hedges. Another crucial discovery is that behind the historically displayed, significantly high win rates lies a result embellished by a large number of these whales' unclosed "zombie orders." The real win rate is far lower than the historical one.

Next, PANews reveals the real operations of this "smart money" through concrete cases.

1. SeriouslySirius: A 73% Win Rate Embellished by "Zombie Orders" and a Complex Quantitative Hedging Net

SeriouslySirius was the top-ranked address in December, with profits of approximately $3.29 million for the month and total historical profits of $2.94 million. Looking solely at his completed order history, his win rate is as high as 73.7%. However, the reality is that this address still has 2,369 open orders, with 4,690 orders settled. Among these, 1,791 currently open orders have actually already completely failed, but the user has not closed them one by one. On one hand, this saves significant effort and fees. On the other hand, since he typically only closes profitable orders, the historical settled order data displays an extremely high win rate. If these unclosed "zombie orders" are considered, this address's real win rate drops to 53.3%, only slightly better than flipping a coin.

In his actual trades, about 40% of orders are hedge orders betting on multiple outcomes for the same event. However, this hedging is not a simple "YES" + "NO." For example, in an NBA game market for 76ers vs. Mavericks, he simultaneously bought positions in Under, Over, 76ers (home team), Mavericks (away team), and 11 other directions, ultimately profiting $1,611. In this process, he indeed employed an arbitrage strategy based on insufficient implied probability, such as buying 76ers to win at an implied probability of 56.8% and Mavericks at 39.37%, with a total cost of approximately 0.962, achieving a state where profit is guaranteed regardless of the outcome. He ultimately made $17,000 on this game.

However, this strategy does not always profit. In a market for a Celtics vs. Kings game, he participated in 9 directions and ultimately lost $2,900.

Furthermore, many orders have severely imbalanced capital allocation. For instance, while orders are placed on both sides, the capital invested differs by more than 10 times. This outcome is likely due to insufficient market liquidity, showing that while arbitrage strategies may seem appealing, liquidity can become the biggest issue in practice. Even if an opportunity arises, you may not be able to achieve a perfectly balanced hedge on both sides.

Moreover, due to automated execution, buy and sell orders in such situations can ultimately translate into significant losses.

Nevertheless, the core reason SeriouslySirius achieved substantial profits with this strategy lies in his competent position management, with a profit/loss ratio of about 2.52. This is the main reason he can be profitable despite a not-so-high real win rate.

Additionally, this strategy is not always profitable. Before December, this address's profit and loss situation was not optimistic, remaining below the break-even line for an extended period, with maximum drawdowns even reaching $1.8 million. Whether the current mature strategy can maintain such profit expectations remains to be seen.

2. DrPufferfish: Turning Low Probability into High Probability, The Art of Extreme "Profit/Loss Ratio" Management

DrPufferfish was the second-highest earning address in December, with monthly profits of approximately $2.06 million. His historical win rate is even more exaggerated at 83.5%. But factoring in a large number of "zombie orders," his win rate also returns to 50.9%. However, this address's strategy differs significantly from SeriouslySirius's. Although he also has close to 25% of orders as hedge orders, this hedging is not opposite-direction hedging but rather dispersed betting. For example, for the final champion of Major League Baseball, he simultaneously bought positions on 27 lower-probability teams, with their combined implied probability exceeding 54%. Through this strategy, he turned a low-probability event into a high-probability one.

Furthermore, the main reason he can achieve huge returns is his ability to control the profit/loss ratio. Taking Liverpool as an example, this Premier League team is his favorite. He has predicted outcomes for this team multiple times (123 times), ultimately earning about $1.6 million. Among the profitable predictions, the average profit was about $37,200, while the average loss among failed predictions was about $11,000. Moreover, he would sell most of these losing orders early to control position losses.

This operational approach gives him an overall profit/loss ratio of 8.62, indicating a high profit expectation. Overall, his strategy is not simple arbitrage hedging but rather achieves massive returns through professional predictive analysis and strict position management. Another point is that most of his hedge trades are in a loss-making state, with total profit/loss for this portion of orders at -$2.09 million. Thus, it appears this whale uses hedging trades mostly as a form of insurance.

3. gmanas: High-Frequency Automated Assembly Line Operations

The third-ranked address, gmanas, has a style similar to DrPufferfish, achieving overall profits of $1.97 million in December. His real win rate is 51.8%, close to DrPufferfish's. However, his trading frequency is higher, having completed over 2,400 predictions, clearly indicating his strategy is executed via automated programs. His betting style is similar to the previous address and won't be elaborated on further here.

4. Hunter simonbanza: The Swing Trader Who Trades Prediction Probabilities Like "Candlesticks"

The fourth-ranked address, simonbanza, is a professional prediction hunter. Unlike previous addresses, his strategy contains no hedge orders. His realized profit is about $1.04 million, with "zombie order" losses of only $130,000. Compared to previous addresses, while his capital scale and trading volume aren't high, his real win rate is the highest at about 57.6%. Additionally, among settled orders, his average profit is about $32,000, and average loss is $36,500. The profit/loss ratio data isn't high, but relying on a higher win rate, he ultimately achieved good profit results.

Furthermore, this address has very few "zombie orders," only 6. This is because he typically doesn't wait until the event concludes to settle; instead, he profits from fluctuations in probability. Simply put, he takes profits when available and doesn't hold out for the final outcome.

This is also a unique investment approach in prediction markets. In his logic, these probability changes are more like the rises and falls in financial investments. Of course, the specific logic behind his high win rate remains unknown—it's his exclusive survival secret.

5. Whale gmpm: Asymmetric Hedging Strategy, Using "Large Positions" to Seek Certainty

The fifth-ranked address, gmpm, while only ranking fifth in December's profit/loss ranking, has a total historical profit higher than the previous addresses, reaching $2.93 million. Additionally, his real win rate is about 56.16%, also at a relatively high level. His operational approach shares similarities with the fourth-ranked address but has its own core strategic secret.

For example, you'll see this address also frequently places orders on both sides of the same game, but his strategy doesn't seem to be about capturing the arbitrage space between the two directions. Instead, he allocates higher capital to the side with higher probability and less capital to the side with lower probability. This achieves a hedging effect where the position is larger when the high-probability event occurs, but the loss isn't too high if the low-probability event happens.

From a practical effect standpoint, this is a more advanced hedging strategy. It doesn't rely solely on the mathematical arbitrage of "yes" + "no" < 1 but combines comprehensive judgment of events with hedging for loss reduction.

6. Model Worker swisstony: High-Frequency Arbitrage in an "Ant Moving" Style

The sixth address, swisstony, is an ultra-high-frequency arbitrage address. He has the highest trading frequency among these addresses, with a total of 5,527 transactions. While accumulating over $860,000 in profit, the average profit per transaction is only $156. Strategically, this address follows an "ant moving" style. Similar to other arbitrage addresses, this address typically buys into all market directions for a game. Taking a Jazz vs. Clippers game as an example, the address bought into 23 market directions. Since the investment amounts are not large, the capital allocation is relatively balanced, achieving a certain degree of hedging effect.

However, this strategy seems to heavily depend on the details of the buys. For instance, "yes" + "no" must be less than 1. For some reason, the total cost of his hedge orders often exceeds 1, meaning these orders are guaranteed to lose regardless of outcome. Nevertheless, with a reasonable profit/loss ratio and win rate, his profit situation still shows positive expectation.

7. The Outlier 0xafEe: The Unconventional "Pop Culture Prophet"

The seventh address, 0xafEe, is a genuine low-frequency, high-win-rate player. His trading frequency is very low, averaging only 0.4 trades per day, with a real win rate of 69.5%.

Among his completed orders, he has earned approximately $929,000 in profit with his超高 win rate, and has very few "zombie orders," with only about $8,800 in unrealized losses. Furthermore, he never engages in hedge orders, focusing solely on predictions. His predictions mainly target Google search trends and pop culture-related content, such as "Will Pope Leo XIV be the most Googled person this year?" or "Will Gemini 3.0 be released before October 31st?" In these areas, he seems to possess unique analytical methods, resulting in奇高 win rates. Among the top-ranked whales, he is considered "unconventional," the only address not focused on sports.

8. Manual Hedging Player 0x006cc: Strategy Upgrade from Simple to Complex Hedging

The eighth-ranked address, 0x006cc, is similar to the aforementioned complex hedging addresses, with an overall net profit of about $1.27 million and a real win rate of about 54%. However, compared to other addresses using automated programs, his operational frequency is low, averaging only 0.7 trades per day. Judging from his early operations, this address might have been a manual operator initially using a "simple hedging strategy."

After entering December, this simple hedging strategy also upgraded to a complex one. Looking at his operational history, as hedging strategies become more widely understood in this market, they are gradually evolving.

9. Cautionary Tale RN1: When "Hedging" Becomes a "Loss Formula"

The ninth-ranked address, RN1, is the only address among December's top ten profitable addresses that is currently in an overall loss position. His realized profit is about $1.76 million, but unrealized losses amount to $2.68 million, resulting in a total loss of $920,000. As a cautionary tale, there are many points for reflection regarding RN1.

First, his real win rate is only 42%, the lowest among these addresses, and his profit/loss ratio is only 1.62. Combined, these two metrics indicate a negative profit expectation; overall, this strategy is not profitable.

Looking closely at the details reveals that this address also clearly employs an arbitrage strategy. However, in many of his hedge trades, while the condition "yes" + "no" < 1 is met, he often allocates more capital to the lower-probability side and less to the higher-probability side. This leads to actual position imbalance, ultimately causing tangible losses when the higher-probability event occurs.

10. Gambler Cavs2: One-Sided Heavy Positions on the NHL Ice, Luck Outweighs Strategy

The tenth-ranked address, Cavs2, is also a prediction gambler who favors one-sided heavy positions, with his most擅长 area being NHL hockey games. However, looking at the overall data, his actual total profit is about $630,000, with a real win rate of about 50.43% and a low risk hedging ratio of 6.6%. The data is mediocre, with a significant component of luck, having hit a few high-paying single-game outcomes. The actual strategic reference value is not high.

Five Harsh Truths After Demystifying "Smart Money"

After in-depth analysis of these "smart money" trades, PANews summarizes the reality behind the "wealth stories" in prediction markets.

1. "Hedging arbitrage strategies" are far from just meeting probability conditions. Under fierce market competition and liquidity constraints, they can easily become a backfiring loss formula. Blind imitation is not advisable.

2. "Copy trading" seems equally ineffective in prediction markets, mainly for several reasons. First, the commonly seen rankings or win rates are based on "distorted" data from historically settled profits. Behind such data, a large portion of "smart money" isn't that "smart." Real win rates exceeding 70% are extremely rare; most win rates are similar to flipping a coin. Additionally, the trading depth in prediction markets is currently relatively poor. The same arbitrage opportunity may only accommodate very small capital entry, and copy traders are likely to be squeezed out in this process.

3. Managing the profit/loss ratio and position size is more important than pursuing a high win rate. Among the addresses with excellent strategy performance, their common characteristic is being very adept at managing the profit/loss ratio. Addresses like gmpm and DrPufferfish even specifically exit positions based on probability trend changes to reduce losses and improve the profit/loss ratio.

4. The real secret lies outside the "mathematical formula." Currently, social media has many interpretations of "arbitrage formulas." At first glance, these strategies seem very reasonable. But in actual operation, the real capability of this smart money appears to lie outside these "mathematical formulas"—either possessing极强的 judgment ability for certain events or having unique analytical models for pop culture. These invisible decision algorithms are the key to their success. For users without such "decision algorithms," the prediction market is equally a冷酷 "dark forest."

5. The profit scale of prediction markets is still very small. Looking at the收益情况 of these top smart money addresses in December, the address with the largest total profit currently has收益 of only around $3 million. Compared to the crypto derivatives market, this market seems to have a clear upper limit on profit potential. For those dreaming of overnight riches and entering this market, its scale is显然 not large enough. Such a market, filled with unique specialization and small scale, likely holds little attraction for institutions for the time being. It's unknown if this is also a major reason restricting the growth of prediction markets.

In the seemingly gold-filled prediction market of Polymarket, the so-called "god-tier whales" are mostly just surviving gamblers or diligent "搬砖工" (arbitrageurs). The real wealth code isn't hidden in those inflated win rate leaderboards but in the algorithms that a few top players bet on with real money, after filtering out the noise.