Deep Dive into 290,000 Market Data Points: Revealing 6 Truths About Polymarket Liquidity

Author: Frank, PANews

Previously, PANews had conducted in-depth research on market prediction strategies , and one important finding was that the biggest obstacle to the success of many arbitrage strategies may not be the mathematical formula of the strategy, but rather the liquidity depth of the market itself.

This phenomenon seems even more pronounced after Polymarket announced the launch of its US real estate prediction market. Since its launch, the market has only seen daily trading volume of a few hundred dollars, far from the expected buzz. Its actual market activity is far less than the discussion on social media. This seems both comical and abnormal, thus necessitating a comprehensive investigation into the liquidity of prediction markets to reveal several truths regarding liquidity within these markets.

PANews retrieved historical data from 295,000 markets on Polymarkert to date and obtained the following results.

1. Short-term market: A PVP battlefield comparable to MEME coins

Of the 295,000 markets, 67,700 have a cycle of less than 1 day, accounting for 22.9%, and 198,000 have a cycle of less than 7 days, accounting for 67.7%.

Of these ultra-short-term prediction events, 21,848 are currently active markets, of which 13,800 have zero trading volume in the past 24 hours, accounting for approximately 63.16%. In other words, a large number of short-term markets on Polymarkert are currently illiquid.

Does this situation seem familiar?

At the height of MEME's popularity, tens of thousands of MEME tokens were issued on the Solana chain, but the vast majority of these tokens either went unnoticed or died out quickly.

Currently, this situation is also being replicated in the prediction market, except that the lifecycle of events in the prediction market is certain, while the lifecycle of the MEME coin is unknown.

In terms of liquidity, more than half of these short-term events have less than $100 in liquidity.

In terms of categories, these short-term markets are almost entirely dominated by sports and crypto market predictions. The main reason is that the judgment mechanisms for these events are relatively simple and mature, typically involving questions like the price movement of a token within 15 minutes or a team's victory. However, perhaps due to its significantly lower liquidity compared to crypto derivatives, the crypto category isn't the most popular "king of short-term trading."

Sports events dominate the market, with analysis showing that sports events with prediction periods of less than one day on Polymarkert have an average trading volume of $1.32 million, while crypto events only have $44,000. This means that if you're hoping to profit by predicting short-term cryptocurrency movements in prediction markets, there may not be enough liquidity to support it.

2. Long-term market: a pool for large funds to accumulate.

Compared to the numerous event contracts in the short-term market, the number of market contracts with longer time horizons is much smaller.

On Polymarkert, there are 141,000 markets with a timeframe of 1-7 days, while there are only 28,700 markets with a timeframe of more than 30 days. However, these longer-term markets have accumulated the most capital. The average liquidity of markets with a timeframe of more than 30 days reaches $450,000, while the liquidity of markets with a timeframe of less than 1 day is only around $10,000. This also indicates that large funds prefer to position themselves in long-term predictions rather than participate in short-term speculation.

In long-term markets (greater than 30 days), all categories except sports showed higher average trading volume and liquidity. The most popular market category was US politics, with an average trading volume of $28.17 million and average liquidity of $811,000. The "Other" category also performed well in attracting capital, with an average liquidity of $420,000 (this "Other" category includes areas such as popular culture and social media topics).

In the crypto market, predictions tend to favor a long-term perspective, such as whether BTC will break $150,000 by the end of the year or whether a certain token's price will fall below a certain level within a few months. In the prediction market, crypto predictions are more like a simple options hedging tool than a short-term speculative tool.

3. Polarization of the sports market

Sports predictions are one of Polymarkert's primary sources of daily active users, currently boasting 8,698 active users, accounting for approximately 40%. However, the distribution of trading volume reveals significant differences across different timeframes in the sports market. On one hand, ultra-short-term predictions (less than one day) average $1.32 million in trading volume; on the other hand, medium-term (7-30 days) predictions average only $400,000, while ultra-long-term predictions (greater than 30 days) average a staggering $16.59 million.

Based on this data, users participating in sports predictions on Polymarkert are either pursuing "instant results" or making "high-stakes gambles" on the season, while mid-season event contracts are not very popular.

4. Real estate forecasts face challenges in adapting to local conditions.

After extensive data analysis, a superficial conclusion is that longer-term predictions seem to have better liquidity. However, this logic sometimes fails when applied to specific or more granular categories. For example, real estate predictions, mentioned earlier, are typically a market with relatively high certainty and a time horizon exceeding 30 days. Yet, predictions of the 2028 US presidential election outperform the entire market in both liquidity and transaction volume.

This perhaps reflects the "cold start dilemma" that new asset classes (especially niche and highly specialized ones) may face. Unlike simple, intuitive event predictions, real estate market participants require a higher level of expertise and knowledge. Currently, the market seems to be in a "strategy adjustment period," with retail investors' enthusiasm limited to observation. Of course, the inherently low volatility of the real estate market exacerbates this cold start; the lack of frequent event-driven fluctuations also reduces the enthusiasm of speculative funds. Under these combined factors, these relatively niche markets face the awkward situation where professional players lack counterparties, while amateur players dare not enter.

5. "Short-term" or "steady accumulation"?

Based on the above analysis, we can make new classifications of different categories of prediction markets. Markets like cryptocurrencies and sports, which are ultra-short-term, can be called short-term markets, while categories such as politics, geopolitics, and technology are more inclined towards long-term, stable markets.

These two types of markets correspond to different investor groups. Short-term markets are clearly more suitable for those with smaller amounts of capital or who need higher capital turnover. On the other hand, "steady" markets are more suitable for those with larger amounts of capital and relatively higher certainty.

However, when the market is segmented by trading volume, it becomes clear that markets with substantial capital accumulation (over $10 million) account for 47% of total trading volume, although they have the fewest contracts (only 505). Markets with trading volumes between $10 million and $100,000 constitute the vast majority in number, totaling 156,000 contracts, but their trading volume is only 7.54%. For the vast majority of prediction contracts lacking strong narrative capabilities, "going to zero upon launch" is the norm. Liquidity is not evenly distributed sunlight, but rather a spotlight concentrated around a very few mega-events.

6. The "Geopolitical" sector is emerging.

The growth momentum of a category can be seen from the "current active number / historical number". The most efficient growth segment is undoubtedly "geopolitics". The total number of historical event contracts in geopolitics is only 2,873, but there are 854 active ones at present, accounting for 29.7% of the total, which is the highest among all segments.

This data indicates that the number of new contracts related to "geopolitics" is rapidly increasing, making it one of the most pressing topics for prediction market users. This is also evident from the recent frequent leaks of insider addresses related to several "geopolitics" contracts.

Overall, behind the liquidity analysis of prediction markets, whether it's the sports sector as a "high-frequency casino" or the political sector as a "macro hedge," their ability to capture liquidity hinges on either providing immediate dopamine feedback or offering deep macroeconomic game-playing opportunities. Those "chicken-rib" markets lacking narrative density, with excessively long feedback cycles and little volatility are destined to struggle to survive in decentralized order books.

For participants, Polymarket is evolving from a utopian "predicting everything" model into a highly specialized financial instrument. Recognizing this is far more important than blindly searching for the next "100x prediction." In this arena, value is only discovered where liquidity is abundant; where liquidity is scarce, there are only traps.

This may be the biggest truth that data tells us about predictive markets.