Unprecedented! 2% of the "Invisible Whales" have actually swallowed up 90% of the trading volume. Is the retail investors' celebration just leftovers from an algorithmic feast?

Market observers point out that discussions about the prediction market platform Polymarket mostly focus on surface-level numbers like trading volume and user count. Few delve into who is actually trading behind these numbers. An analysis based on the complete wallet behavior over the past three months approaches from two dimensions: trading frequency and trading scale, sketching out the true structure of this market.

Data shows that the vast majority of trading volume is contributed by a small group of algorithms and high-frequency traders. A large low-frequency retail group lives almost in two parallel worlds with this group of professional players. Understanding this division directly relates to the platform’s fee design, product roadmap, and market strategy.

The distribution of trading frequency exhibits a typical logarithmic decay. The largest user group traded only 2 to 10 times throughout the entire study period, accounting for 32% of the total user base. Adding the group that traded 11 to 50 times brings the total to nearly two-thirds of all users. They typically participate only during elections, sports events, or major economic occurrences, with small amounts of capital at stake.

The distribution of trading volume, however, is completely different. Although high-frequency users are few, the trading volume presents a bell-shaped distribution on a logarithmic scale, peaking at $600 to $3,000 per user. This means that the typical active user’s trading amount is in the four-figure range, but the “long tail” users starting from $25,000, though very few in number, contribute the vast majority of the platform’s trading volume.

These two distribution charts together reveal a structural split: on one side is a massive number of low-frequency participants, while on the other are professional players with huge trading volumes. The latter are almost invisible on the user count chart but dominate the trading volume chart.

Simply categorizing users by frequency or volume will distort the picture. Trading 500 times with a total of $50 differs entirely from trading 500 times with a total of $5 million; these are two completely different types of participants. Analyzing both dimensions together categorizes wallets into seven tiers.

T1 to T7 represent trading frequency, from single transactions to over 10,000 transactions. V1 to V7 represent trading scale, from below $100 to over $2 million. This results in seven types of user profiles.

P1 is a one-time silent user, trading only once and under $100, indicative of a trial. P2 is a low-active retail trader, trading 2-10 times with a total below $1,000, purely event-driven. P3 is a moderate participant, trading 11-200 times, with amounts between $1,000 and $10,000, entering repeatedly but lacking systematic strategies.

P4 is a high-depth retail trader, trading 201-1,000 times, with amounts between $10,000 and $100,000, actively participating but not reaching institutional levels. P5 is a low-frequency high-net-worth individual, trading fewer than 50 times but often exceeding $100,000 per transaction, indicative of selective opportunity and heavy betting.

P6 is a high-frequency professional trader, trading over 200 times, with amounts exceeding $100,000, representing algorithmic strategies and institutional traders. P7 is a high-frequency low-value player, trading over 200 times but with a total under $10,000, highly active yet capital-constrained.

The number of P2 low-active retail traders reaches 849,000, accounting for 69% of total users. The P6 high-frequency high-capital users number only 27,000, about 2%. However, during the statistical period, the P6 group generated a total trading volume of $39 billion. This is an extreme manifestation of the Pareto principle: it is not 80/20, but rather that 2% of users support nearly 90% of the trading volume.

The user growth chart and the trading volume growth chart depict almost completely different groups. Platforms targeting user growth and those targeting trading volume growth will have vastly different product decisions.

Sports and $cryptocurrency are the two categories with the largest trading volume on the platform, accounting for 42% and 31% of total trading volume, respectively, but their user structure differs significantly. In the $cryptocurrency market, the proportion of high-frequency high-capital (P6) traders is much higher than the overall level of the platform, aligning closely with algorithmic trading. Their trading volume is high, and frequency is also high, indicating that execution is likely automated.

While sports betting is also dominated by high-frequency high-capital (P6) trading, the proportion of moderate (P3) and high (P4) participants is higher than in the $cryptocurrency category. This means the sports domain has both institutional algorithmic funds and a large number of experienced players relying on manual analysis.

Political users account for the highest proportion at 19%, and are relatively evenly distributed among user groups. Low-participation users (P2) have the highest proportion in the political category, typically being one-time retail traders driven by events, registering only to participate in election betting.

The economic and financial sectors attract a disproportionately high number of low-frequency high-capital (P5) participants. They trade infrequently but with massive amounts, betting large capital on macroeconomic outcomes.

The categories on the platform directly determine the user groups attracted and affect liquidity depth, user retention, and fee tolerance. A new $cryptocurrency market will attract algorithmic and high-frequency traders; a new political market will attract event-driven retail traders who may leave immediately after the event ends.

If the goal is trading volume, then products should be built for P6 users. If the goal is user growth and brand influence, then products should target P2 users. These two goals require entirely different category selections.

User segmentation profiles directly determine the fee design of the prediction market. A fixed fee per transaction would overly suppress the P6 and P7 groups, which are precisely the ones supporting the platform’s liquidity foundation. Therefore, differentiated fees by category hold value.

Polymarket’s current rate system reflects this logic: the effective fee for the $cryptocurrency sector is the highest at 1.80%; the sports sector is 0.75%; the political and financial sectors are at 1.00%; and the geopolitical sector has zero fees throughout.

This standard precisely matches the population structure and trading habits of each category. The $cryptocurrency sector is filled with P6 algorithmic funds, which can bear higher fees without damaging liquidity; the political sector primarily consists of low-barrier retail traders, necessitating reduced friction costs to maintain retention. Fee designs that disregard user profiles are essentially blind trial and error.

The core conclusions are as follows: the P6 high-frequency high-capital group constitutes only 2% of users but generates 88% of the platform’s trading volume; fee policies that harm P6 interests will severely damage the platform’s foundation; 69% of users are low-frequency small retail traders driven purely by hot event topics; $cryptocurrency trading is highly concentrated among algorithmic high-frequency funds, while the sports sector has a more diverse participant structure; ordinary users average only 12 transactions within 90 days, with a median total investment of $224; expanding into new categories requires anchoring to target user profiles rather than merely chasing hot topics.

Since trading volume is highly concentrated among a small core of high-frequency users, why does Polymarket still position itself as a retail product? Professional algorithmic funds underpin the vast majority of transactions, yet product experience, marketing strategy, and category layout seem to continuously accommodate ordinary retail traders.

Part of the answer may lie in structural evolution. The proliferation of agent frameworks, Telegram bots, and no-code tools has made it easy for retail traders to engage in automated trading. If retail traders have started to engage in algorithmic trading, the next natural evolution would be for AI agents to autonomously conduct large-scale high-frequency operations.

This is also why Polymarket may nurture the first killer application at the intersection of $cryptocurrency and artificial intelligence. In a market characterized by strong liquidity, event-driven dynamics, and binary outcomes, autonomous agents can operate with precision. They can absorb world events, social sentiment, and real-time reasoning information, identify mispricing trading opportunities, and execute trades without human intervention.

When this application achieves breakthrough progress, it will no longer just be a $cryptocurrency product. That will be the moment when agent trading enters the mainstream market.

Follow me: Get more real-time analysis and insights into the crypto market! $BTC $ETH $SOL

#GrowthValueLotteryWinGoldBar #CryptoMarketCorrection #BitcoinFluctuatesWeaken