Who is the real god on Polymarket? After scrutinizing 90,000 addresses, we discovered these harsh truths.

Author: @clairegu1, Hubble AI

Polymarket boasts numerous "god-like" addresses that have achieved single-transaction profits of $100,000. However, with hundreds of thousands of accounts, a core question consistently plagues participants: Is this replicable alpha, or unsustainable luck?

Existing leaderboards have a serious blind spot: they only show short-term results and fail to reveal the stability of a strategy. To eliminate the element of luck, we bypassed simplistic leaderboards and directly analyzed 90,000 active addresses and 2 million settled transactions on the blockchain.

After eliminating the interference of unrealized profits, we discovered four counterintuitive patterns in the market prediction process, which are both brutal and real, and redefined the screening criteria for copy trading.

TL;DR

• Mid-frequency efficiency trap: The most active retail investor group (mid-frequency) has the highest win rate across the entire network, but due to limited capital efficiency and lack of systemic advantages, the median actual return is close to zero.

• Certainty Trap: Betting on high-probability events (>0.8) faces an extremely asymmetric risk-reward ratio (a small profit if you win, zero if you lose), with a negative long-term expected value.

• Gold Odds Range: True Alpha is highly concentrated in the price range of 0.2 - 0.4. This is the area with the greatest market divergence and the optimal risk-reward ratio (Odds).

• Focus Premium: Data shows that "all-rounder" traders struggle to survive. Experts specializing in a few niche sectors typically earn four times the returns of diversified traders.

Finding 1: Mid-frequency trading carries the highest risk.

We categorize addresses into three tiers based on the number of transactions:

• Low-frequency trading: ~0.35 trades per day | Win rate ~40%

• Mid-frequency trading: Average of ~3.67 trades per day | Win rate ~43%

• High/Ultra Frequency: >14 trades per day | Win rate ~21-26%

Based on the surface data, mid-frequency traders seem to be the best performers in the market: they have the highest win rate, reaching ~43%; and the lowest percentage of losing accounts, with a loss ratio of only 50.3%, far lower than the 77.1% of the High group.

This gives the illusion that as long as you maintain a moderate number of 3-4 trades per day, you can make steady profits.

However, when we introduce PnL (profit and loss) data, the truth comes to light:

• Median PnL (Median Profit/Loss): The value for the median group is 0.001, which is almost zero.

• What does this mean? It means that for the vast majority of mid-frequency traders, even if you are researching and betting every day and seem to win more than you lose, your account equity is still stagnating.

In contrast, while High Frequency (HF) and Ultra Frequency (UHF) had median losses (-0.30 and -1.76 respectively), their Mean Profit/Loss (MPL) was driven up to +922 or even +2717 by a very small number of top-performing addresses. This illustrates that the HF field is a "battleground for machines"—surviving by relying on low win rates, high profit/loss ratios, and systematic strategies (such as market making and arbitrage), a model that ordinary people cannot replicate.

In-depth attribution: Why do mid-frequency frequencies fall into the "mediocrity trap"?

• Lacking systematic alpha, they become mere "coin-flipping" players: mid-frequency traders are mostly active retail investors. A win rate of approximately 43% and a median return close to zero indicate that this group's overall performance resembles a random walk. They participate in the market based on intuition or fragmented information. While they avoid significant drawdowns due to strategy failure like high-frequency bots, they also fail to build a true competitive advantage. They are repeatedly participating in the market, rather than profiting from it.

• Survivorship bias masks tail risks: there is a huge gap between the average PnL (+915) and the median PnL (-0.001) of mid-frequency traders. This indicates extreme polarization within the mid-frequency trading group. A very small number of "experts" with core insider knowledge or exceptional judgment have inflated the average, while the remaining 50% or more are doing futile work.

• High frequencies are difficult to replicate, and low frequencies are insufficient: Ordinary users cannot imitate the systematic high-frequency strategies of the High/Ultra bands (high technical barriers, low win rates, and high psychological pressure), yet they are unwilling to accept the extremely low success rate of the Low bands. As a result, a large amount of funds and energy are poured into the Mid band, making it the most crowded, most competitive, and most mediocre "red ocean."

Practical Insight: Data reveals a harsh truth: if you simply strive to become a "diligent mid-frequency trader," you'll likely end up wasting your time. The real value lies not in mimicking the average behavior of "mid-frequency" traders, but in identifying the differences.

• Avoid pitfalls: The vast majority of mid-frequency addresses are just performing Brownian motion and have no value in following their trades.

• Mining: The real Alpha is hidden in the right tail of the mid-frequency group—those very few who outperform the "zero gravity" at the same frequency.

This is precisely the core value of our copy trading tool: helping you skip the trial-and-error phase of "long-term mid-frequency trading with no advantage" and using algorithms to directly lock in the 1% of Alpha addresses that truly generate excess returns from a massive amount of mediocre mid-frequency trading denominators.

Discovery 2: "Picking up coins" and "buying lottery tickets" will eventually both end up with nothing.

We stratified traders' risk preferences by position price and discovered a harsh reality: whether you only buy "lottery tickets" (<0.2) or only buy "certainty" (>0.9), you will be a loser in the long run.

We have defined three typical strategies:

• High certainty strategy (Consensus Betting): Concentrate positions on prices >0.9, specifically targeting events that are "almost certain".

• Long-shot Betting Strategy: Positions are concentrated at prices below 0.2, focusing on low-probability upsets.

• Dynamic Strategy: Balanced position allocation, not fixated on extreme odds.

The data reveals a huge revenue gap:

Data Interpretation: The average return of hybrid strategies is 13 times that of high-certainty strategies. It's worth noting that the median return for all groups is ≤0. This means that even in the best-performing hybrid group, profits are highly concentrated in the hands of top players, and the vast majority of participants do not even beat transaction fees.

1. Why does betting on "certainty" fail?

Intuitively, buying at 0.95 seems like a low-risk, "sure-win" strategy. However, from a financial mathematics perspective, it's an extremely poor trade:

• Extremely asymmetric downside risk: Entering at 0.95 means you're risking 1.0 of your principal to gain 0.05. If a black swan event occurs (such as Biden suddenly withdrawing from the race, or a game being reversed in the last minute), the loss from a single event would require you to make 19 consecutive correct trades to break even. Over the long term, the probability of a black swan event is often higher than 5%.

• Alpha depletion (Priced In): When the price reaches >0.9, market consensus has already formed. Entering the market at this point is essentially taking over for those who were already aware of the market, and there is no longer any informational advantage to speak of.

2. The "Lottery Trap" of High-Odds Strategies

Betting on low-probability events with a stake of <0.2 also performed poorly, for the following reasons:

• Overestimation Bias: Retail investors often overestimate their ability to predict "unpopular" trends. In efficient predictive markets, prices usually already contain most of the implicit information. Long-term purchases of "lottery tickets" that are correctly priced by the market will inevitably result in the continuous erosion of one's principal.

• Low capital efficiency: Although the profit multiple per trade is high, the extremely low win rate will cause the capital to be in a state of drawdown for a long time, making it difficult to form a compound interest effect.

Practical Insight: Avoid being a "one-track mind" trader. When selecting copy traders, avoid addresses with extreme price distributions (all red or all green). True Alpha players are characterized by strategic flexibility—they bet on divergence at 0.3 and take profits at 0.8, rather than mechanically sticking to a particular odds range.

Finding 3: The optimal risk range is 0.2-0.4.

We stratified addresses by average purchase cost (Implied Probability) in an attempt to find the "sweet spot" with the highest risk-adjusted returns.

The data reveals a clear non-linear distribution of returns: the true alpha does not exist at the extremes, but is concentrated in the price range of 0.2 - 0.4.

Performance comparison across different price ranges:

In-depth insight: Why is 0.2-0.4 the most profitable?

1. Capturing “Pricing Divergence”

A purchase price between 0.2 and 0.4 indicates that the market consensus suggests the probability of this event occurring is only 20%-40%.

Traders who consistently profit within this range are essentially engaging in "cognitive arbitrage." They are able to identify events that are underestimated by public sentiment (e.g., the market is overly pessimistic and misjudging the probability of a candidate's comeback). Compared to simply following the consensus (buying > 0.8), betting in the divergence zone, once validated, can yield explosive returns of 2.5 to 5 times.

2. The perfect "asymmetric risk/reward structure"

• In the >0.8 range (certainty trap): investors face poor odds of "a small profit if they win, and zero if they lose." As the data shows, the average return in this range is negative, and the win rate is only 19.5% (meaning that most people who buy in >0.8 eventually perish due to black swan events).

• In the 0.2-0.4 range (Alpha comfort zone): This is a range with "convexity." Downside risk is locked in (principal), while upside potential is flexible. Skilled traders maximize profits in this range through the dual advantages of a high win rate (49.7%) and high payouts.

• Avoid the "lottery trap" (<0.2): Although the theoretical odds are highest in the extremely low price range, data shows that its performance is far inferior to the 0.2-0.4 range. This indicates that events with odds <0.2 are often true "garbage time" or pure noise, and excessively betting on extremely low probability events lacks a statistically positive expected value.

Practical Insight: Focus on "Divergence Hunters." When selecting traders to copy, prioritize those whose average buy price consistently remains between 0.2 and 0.4. This data characteristic indicates that the account neither blindly chases high-risk lotteries nor tries to "pick up pennies" in low-odds consensus zones, but rather focuses on finding undervalued opportunities where market pricing has become ineffective. This is the core competency most worth replicating.

Finding 4: A centralized strategy is superior to a decentralized strategy.

We calculated the Focus Ratio (total number of transactions / number of market participants) for each address and divided it into two categories:

• Diversified strategy: Participate in a large number of markets, with fewer transactions in each market.

• Concentrated strategy: Focus on a few markets, with a high number of trades in each market.

The results show:

The returns of a centralized strategy are four times that of a decentralized strategy ($1,225 vs $306).

It is worth noting that the win rate of the concentrated strategy was actually lower (33.8% vs 41.3%).

Concentrated strategies have yielded significant returns on a few high-odds opportunities.

explain:

• In-depth research creates an advantage. By focusing on a few markets, it is easier to discover market pricing discrepancies, thereby obtaining excess returns in a few transactions.

• Win rate is not the key metric. What matters is not the number of wins, but the ratio of winning profits to losing losses. Concentrated strategies accept a lower win rate in exchange for higher returns per trade.

• The limitations of a diversified strategy: Participating in too many markets leads to insufficient research depth in each market, making it more susceptible to market consensus and making it difficult to discover true alpha.

analogy:

As Buffett said, "Diversification is the self-protection of the ignorant." If you have an informational or judgmental advantage, you should focus on the few opportunities you are most confident in.

• Copy trading insights: Prioritize traders who focus on specific market types (such as particular sports leagues or political events in specific countries). Their level of specialization often implies a deeper understanding and stronger predictive abilities.

II. Implications for copy traders: How to identify real "smart money"?

To quantify the level of professionalism among traders, we constructed the Focus Ratio metric (Focus Ratio = Total number of trades / Number of markets participated in), and divided addresses into two distinct groups:

• Generalists: They participate extensively in a wide range of markets, with low frequency of transactions in a single market, and attempt to reduce risk through diversification.

• Specialists: They focus on a few markets, repeatedly trading and adding to their positions in a single market, exhibiting a strong "sniping" characteristic.

The data reveals a striking "focus premium": Strategy Type Average Return (Avg PnL) Win Rate Number of Addresses Generalists $306 41.3% 68,016 Specialists $1,225 33.8% 22,458

Data Interpretation: The average return of a concentrated strategy is four times that of a diversified strategy. However, a highly misleading phenomenon has emerged: the win rate of a concentrated strategy (33.8%) is significantly lower than that of a diversified strategy (41.3%). This reveals the true profit logic of advanced players in the prediction market.

In-depth attribution: Why is "less is more"?

1. Building a competitive advantage through information asymmetry (Information Edge)

Prediction markets are essentially information games.

Decentralized traders attempt to span multiple fields such as politics, sports, and crypto, which results in them only having a "shallow understanding" of any single market, making them vulnerable to being exploited.

Concentrated traders, on the other hand, establish an informational advantage in a vertical market by focusing on a single sector (such as studying only NBA player data or tracking only polls in swing states). This depth allows them to detect subtle discrepancies in market pricing.

2. Debunking the "Win-Rate Fallacy"

Data shows that high returns are often accompanied by a relatively low win rate.

This is because concentrated experts tend to act when there are high odds/high divergence (e.g., buying when the odds are 0.3), rather than picking up "certainty coins" with odds >0.9.

• Diversified approach: Wins many small amounts of money (high win rate), loses a large amount of money once (black swan event), and ultimately achieves mediocre returns.

• Concentrated approach: This approach allows investors to tolerate numerous small-scale trials (low win rate) in exchange for a few highly profitable, targeted investments (high profit/loss ratio). This is typical venture capital (VC) logic, not the logic of working for someone else.

3. Validation of Buffett's logic in market prediction

As Buffett said, "Diversification is the self-protection of the ignorant."

In the stock market, diversification is used to mitigate unsystematic risk; however, in the zero-sum game of the prediction market, diversification often means a dilution of focus. If you are confident that you have some edge, the best strategy is not to cast a wide net, but to concentrate your firepower on the few opportunities you are most confident in.

Practical Insight: Finding "Experts in a Vertical Track". In copy trading, a high focus ratio is a more important indicator than a high win rate.

• Bad sign: Avoid those "generalists" who buy everything.

• Good news: Prioritize accounts active only under specific tags. For example, an address that only trades "US Election" and has a stable profit curve is far more valuable than an address that trades both "NBA" and "Bitcoin." The level of specialization directly determines the purity of the alpha.

III. From Data Insights to Practical Tools

This report is not only a data review, but also the underlying logic for building our Smart Copy-Trading system.

To achieve long-term profitability on Polymarket, manually sifting through 90,000 addresses is unrealistic. We are packaging these exclusive data insights into an automated screening and risk control tool to solve the three most challenging problems in copy trading:

1. Intelligent elimination of market maker noise

The current public trading lists are mixed with a large number of market makers (MMs) and arbitrage bots that inflate trading volumes. Following their trades will not only fail to generate profits, but may also result in losses due to slippage.

Solution: Utilizing proprietary order book analysis and trading feature recognition algorithms, we automatically filter out systematic market makers, focusing solely on active traders who genuinely profit based on their insights.

2. Vertical matching based on "focus"

A general "profit ranking" has limited significance; you need experts in specific fields.

Solution: Based on Focus Ratio and historical behavior, we assign highly precise "capability tags" to addresses (such as US elections, NBA sports events, CryptoWhale). The system will then accurately match you with vertical experts who possess informational advantages in the relevant fields based on your areas of interest.

3. Dynamic Style Drift Detection

The most hidden risk of copy trading lies in the sudden failure of a trader's strategy or a sudden change in behavior.

Solution: We have established a real-time risk control model. When a long-term stable address suddenly deviates from its historical behavior characteristics (e.g., from low-frequency focus to high-frequency broad casting, or an abnormally large single risk exposure), the system will identify it as an abnormal signal and issue a timely warning to help users avoid drawdown risks.

Conclusion & Product Beta Testing

Prediction markets are a brutal zero-sum game. Data from 90,000 addresses proves that long-term winners win because they are extremely restrained: they focus on specific areas and look for pricing discrepancies.

All the core metrics mentioned in this report (Focus Ratio, pricing range analysis, market maker exclusion) are integrated into Hubble's data backend. Our initial motivation for building this tool was simple: to replace blind retail investor intuition with an institutional-grade data perspective.

Beta Testing Application: Hubble's Polymarket intelligent copy trading tool is currently undergoing limited, phased testing. If you agree with the above data analysis logic and wish to experience this product:

• Like/share to support this content;

• Leave a comment saying "Waitlist";

We will send you beta testing invitations via private message. We hope this data-driven selection system will help you truly outperform the market.

(Data Note: This study is based on settled transaction data from the Polymarket platform since its launch. All conclusions are derived from Hubble's proprietary on-chain PnL algorithm analysis. Author: Hubble @clairegu1)