Prediction Markets: How They Turn Disagreement Into Revenue

In the past, discussions about prediction markets focused more on their accuracy, trading volume, and whether they could become new information markets. But when viewed as a business, the core question shifts: What is the revenue model of prediction markets?

In the business world, high trading volume does not equate to a platform making money. A market can have massive volume and users trading frequently, but if most trades don’t fall within revenue-generating categories, or if activity is sustained purely by subsidies and points, then trading volume is merely impressive data—not healthy revenue.

For prediction markets, what truly tests a platform’s business acumen isn’t “how many markets it offers” or “how popular a particular event is,” but whether the platform can seamlessly connect these three elements:

The urge to generate real trades;

Maintain sufficient order book liquidity;

Convert taker trading orders into fees.

This is why the business model of prediction markets is far more complex than simply "collecting fees on opening bets." On the surface, it may appear to be nothing more than a series of yes/no wagers, but the true foundation of platform revenue lies in the underlying trading structure, liquidity mechanisms, fee differentials, and user behavior.

In particular, after the leading platform Polymarket began systematically introducing taker fees, the narrative around prediction markets has shifted from being an "information tool" to entering a "revenue validation" phase.

This article will take a business perspective to deeply analyze the underlying mechanics of prediction markets:

How do prediction market platforms make money?

Why does the maker/taker fee structure determine the platform's survival?

What is the fundamental difference in fee structures between mainstream platforms such as @Polymarket, Kalshi, @opinionlabsxyz, and @predictdotfun?

Why is the trading volume highest not necessarily the most profitable sector?

💡 Key Insight: Prediction markets don't sell answers—they sell disagreement.

The closer the price is to 50/50, the greater the market divergence and the stronger the trading impulse, making it easier for the platform to convert active trades into fee revenue; the closer the price is to 0 or 100, the more certain the outcome becomes—although informational value still exists, the corresponding fee weight decreases significantly.

Therefore, the real business barrier in prediction markets is not turning "events" into betting markets, but turning "disagreements" into trades, and then reliably converting those trades into revenue.

To break down the cash flow of a prediction market, start by identifying its four core revenue drivers. They are interconnected and together form a closed loop from traffic to monetization.

1️⃣ Trading Fee - Direct Revenue Source

Most prediction markets charge a fee to the side that takes the trade, known as the Taker, because the Taker consumes liquidity, while the Maker provides it.

This means that not all trades on a prediction market generate revenue. The trades that truly contribute fees to the platform are typically those where users are willing to actively execute trades and pay for speed and certainty.

2️⃣ Liquidity - The Foundation of Continuous Trading

The hardest part of prediction markets isn't setting the odds—it's creating deep liquidity.

If there are no orders on the order book, users cannot buy or sell, so even if the market is trending, it will struggle to form a valid price.

Therefore, many platforms reduce the cost of making or even incentivize makers.

This is not a direct "source of revenue," but it determines whether trading fees can sustainably exist.

Without liquidity, there can be no consistent trading, and fee income naturally cannot be stable.

3️⃣ Information Value – Mindshare Capture

Prediction markets differ from regular trading platforms in that they are not just trading tools—they also generate information.

Once a order book has sufficient trading volume and liquidity, its price becomes a probabilistic signal. Media outlets will cite it, KOLs will interpret it, traders will monitor it, and ordinary users will use it to gauge market sentiment.

This may not directly become a fee, but it generates attention, user awareness, and external exposure for the platform. Over the long term, this informational value will enhance trading demand in return.

4️⃣ User Engagement and Discount System – Converting Activity into Revenue

In addition to base trading fees, different platforms enhance trading frequency through discounts, referrals, promotions, points, and rebates. These measures may not directly generate revenue but can impact the platform’s long-term monetization potential. For example, Opinion offers user discounts, trading discounts, and referral discounts; Predict.fun employs a simpler model based on basic fees and discounts; Polymarket focuses on differentiated fee structures across various markets and Maker rebates. The essence of discounts and incentives is not merely subsidies, but rather trading a portion of profit to secure user retention, gradually converting engagement into revenue.

Looking at the fee structures of several leading prediction markets, the industry’s strategic direction is highly aligned: encouraging limit orders to provide liquidity and turning active trades into revenue. However, in terms of tactical execution, platforms exhibit clear strategic differentiation due to their differing positioning.

1️⃣ Polymarket: Precise Pricing by Category

Polymarket's taker fee logic极致地结合了“赛道差异化”和“分歧度定价”。其官方核心公式为：

fee = C × feeRate × p × (1 - p)

Among these, C is the traded quantity, p is the trade price, and feeRate is determined by the market segment.

This mechanism includes two core variables:

Market Segmentation by Fee Rate: Based on the current verified fee rate口径, the fee rate for the Crypto market is 0.07, Sports is 0.03, Politics / Finance / Tech is 0.04, Culture / Weather is 0.05, and some Geopolitics markets are 0. This means Polymarket does not apply a uniform fee rate across all markets; instead, it implements differentiated fee rates based on trading volume, sensitivity, and user willingness to pay for each market category.

Disagreement Pricing: Perfectly aligned with the mathematical curve p × (1 - p). Fees are highest when prices are closest to 50/50 (maximum market disagreement) and lowest when outcomes are more certain (close to 0 or 100).

https://docs.polymarket.com/trading/fees

2️⃣ Kalshi: Closer to a compliant exchange model

Kalshi's fee structure is designed to align more closely with traditional financial derivatives exchanges under a regulatory framework, and its standard taker fee formula is also tied to price dispersion:

Fee = ceil(0.07 × C × P × (1 - P))

Where C is the number of contracts, P is the contract price, and fees are rounded up to the nearest cent. This structure is very similar to Polymarket’s C × feeRate × p × (1-p).

Kalshi’s fee structure is similar to Polymarket: its trading fees are tied to the contract price, with higher fees near 50¢ and lower fees near 1¢ or 99¢. According to Kalshi’s fee schedule, the taker fee for 100 contracts ranges approximately from $0.07 to $1.75.

However, an important distinction between Kalshi and Polymarket is that Kalshi applies maker fees to certain markets, and these fees are charged only if the limit orders are eventually filled—cancelling orders incurs no fee. This indicates that Kalshi’s fee structure more closely resembles that of a regulated exchange: rather than offering perpetual free maker fees, it implements more complex, market-specific two-sided fee rules.

https://kalshi.com/docs/kalshi-fee-schedule.pdf

3️⃣ Opinion: Emphasize discounts and user tiering more

Opinion has introduced an extremely complex "multi-dimensional discount system," with the effective fee formula:

Effective fee rate = topic_rate × price × (1 − price) × (1 − user_discount) × (1 − transaction_discount) × (1 − user_referral_discount)

In other words, Opinion's fees are influenced not only by market price and topic_rate, but also by factors such as user discounts, transaction discounts, and referral discounts.

Opinion also set a $5 minimum order and a $0.25 minimum fee to prevent excessively low fees from small transactions.

This indicates that Opinion's fee structure is more oriented toward user acquisition:

topic_rate is used to distinguish between different markets.

user_discount is used for user segmentation.

Therefore, unlike Polymarket’s “category-based pricing,” Opinion treats fees more like an operational tool: using a discount system to encourage trading, retention, and referrals on one side, while lowering the barrier to placing orders through free maker fees to maintain market liquidity on the other.

https://docs.opinion.trade/trade-on-opinion.trade/fees

4️⃣ Predict.fun: A Minimalist Flat Fee

Predict.fun's fee structure is relatively simpler, helping to reduce user comprehension costs.

According to its current public statement, its fee calculation formula is:

Raw Fee = Base Fee % × min(Price, 1 − Price) × Shares

The base fee is currently 2%. The actual fee varies with the execution price: below 50%, the fee remains approximately fixed at 2%; above 50%, the actual fee decreases as the price approaches 1.

In addition, Predict.fun supports fee discounts, which further reduce the transaction fees.

This design is more intuitive: users don’t need to first determine which side of the order book they’re on—they can simply focus on the execution price to understand fee changes.

https://docs.predict.fun/the-basics/predict-fees-and-limits#limits

It can be seen that a common feature of prediction market platforms is that they all aim to convert active trading behavior into revenue.

This also shows that the commercialization of prediction markets is not limited to a single path. Ultimately, they all answer the same question: Are users willing to pay for trading?

Although various platforms offer diverse approaches, Polymarket remains the most suitable platform for observing the real monetization efficiency of prediction markets.

There are mainly two reasons:

Its fee structure is the clearest: starting with Crypto trials, expanding to Sports, and gradually implementing fees across nearly all categories.

Its data is also more comprehensive: official feeRate, 7D/30D fees can all be used to further break down revenue structure.

So next, let’s use Polymarket as an example to answer a more specific question: Is the market segment with the highest trading volume also the most profitable?

3.1 From Free to Paid: Polymarket’s Monetization Timeline

January 2026: Crypto becomes the first paid section

Polymarket is returning to U.S. users and is the first in the Crypto section to introduce Taker Fees. With short settlement cycles, high price volatility, and trading behavior similar to secondary market short-term trading, users prioritize speed of liquidity over sensitivity to friction costs, making it an ideal testing ground for fees.

February 18, 2026: Sports becomes the second paid section

On February 18, 2026, the Sports section became the second paid section. Sports betting inherently features high frequency and short cycles, providing continuous trading opportunities. Therefore, charging for Sports is a natural extension of this model.

Therefore, Polymarket is initially charging for Crypto and Sports to validate its revenue model using the two sectors with higher user adoption.

March 30, 2026: Fees extended to additional sectors

On March 30, 2026, Polymarket expanded its taker fee to additional categories including Politics, Finance, Economics, Culture, Weather, Tech, Mentions, and Other/General, increasing the total number of fee-bearing categories to 10.

After implementing comprehensive fees, Polymarket did not simply charge the same fee across all markets; instead, it adopted a more granular fee structure. This step marks a key milestone in Polymarket’s monetization, as the platform begins to extend its fee model to a broader range of markets.

The effectiveness of its fee structure is remarkable. According to the latest data, Polymarket has demonstrated exceptional revenue-generating capacity: 7D fees reached $9.27M, and 30D fees reached $36.3M. Its 7-day revenue has now ranked within the top six among all crypto projects, officially entering the ranks of revenue-generating projects.

3.2 Breakdown of Core Track Single-Type and Price Distribution

To estimate Polymarket's true revenue across its various sectors as accurately as possible, we analyzed Polymarket trading data from 2021 through February 2026 to calculate fees for five major categories.

Looking at the proportion of market orders, there are clear differences among the five sectors:

Crypto has the highest market share at 75%, which strongly aligns with the volatile nature of crypto assets, as users prefer to use market orders to lock in profits and losses immediately. The Weather sector, driven by real-time,突发 weather data, is similarly highly valued by users for its speed of response.

Second, the amount of fees heavily depends on the price range of trades on the order book.

The reason is that transactions subject to fees do not incur the same charges. Polymarket's fees are related to p × (1 - p): the closer the price is to 50/50, the greater the market uncertainty, and the higher the fee weight; the closer the price is to 0% or 100%, the more certain the outcome, and the lower the fee weight.

Based on data from the five main categories, most trading activity is concentrated in the 30–50 range, particularly between 40 and 50:

This data shows that Polymarket’s primary trading activity does not occur in ranges where outcomes are nearly certain, but rather concentrates in areas where significant market disagreement still exists.

3.3 Revenue Projection: Who Are the Profit Drivers?

We estimate Polymarket’s fee revenue across five markets by combining market trading volume with the corresponding fee rate, applying a p × (1-p) weight based on different price intervals. Additionally, we account for the fact that after implementing fees, some fee-sensitive users may shift from Taker orders to Limit orders—particularly users engaged in end-of-market trading, low-odds arbitrage, or frequent short-term trading, who will more carefully calculate their return rates.

Therefore, we can establish a more conservative assumption based on the original estimate: assuming that after the fee implementation, the market order trading volume in each segment decreases by 20%.

The adjusted formula becomes:

Estimated fee after adjustment ≈ Market turnover × 80% × feeRate × (1 - p)

Based on 7-day total trading volume and the proportion of trading volume across each sector, we estimate the 7-day market order trading volume for the five main sectors.

The order volume for each赛道 has already been calculated. Next, we will estimate the fees by combining each赛道's feeRate and price range weight. To ensure greater accuracy, we use the median of the price range as the approximate price:

(Note: Due to differences in statistical methodology, lag in historical order type proportions, and dynamic changes across sectors, this estimation model aims to reconstruct the contribution ratios of each sector; the total may differ slightly from the system’s actual settled Fees.)

What does the data indicate?

1️⃣ Crypto is currently the highest-revenue-generating segment, with estimated fees over the past 7 days totaling approximately $4.39 million, making it a "profit engine."

This is somewhat counterintuitive, as based on trading volume share, Sports is the leading category, with a 7-day trading volume of approximately $401 million, higher than Crypto’s $174 million. However, in terms of fees, Crypto ranks first, primarily due to two reasons:

Higher proportion of market orders: Market orders account for approximately 75%, significantly higher than Sport’s 60%. Since Polymarket primarily charges for market orders, more Crypto trades fall within the fee scope.

The fee rate is at its maximum: feeRate is 0.07, while Sport is only 0.03. Even if both market orders have the same trade value, the fee per unit traded on Crypto will be significantly higher.

2️⃣ Sport is the second-largest revenue source, with a 7-day estimated fee of approximately $3.31 million, serving as the "volume base."

Sport's advantage lies in its high trading volume. Its 7-day trading volume is approximately $401 million, ranking first among the five categories. However, its weakness is also clear: its fee rate is the lowest at just 0.03%.

3️⃣ If Politics and Trump are merged into a single political market, the estimated 7-day cost is approximately $3.14 million, very close to the Sports category, acting as a pulsed traffic funnel.

Political markets are characterized by strong event-driven activity. Unlike sports, which offer consistent daily matchups, or crypto, which experiences continuous price fluctuations, political markets see concentrated trading whenever elections, polls, policy changes, or candidate statements occur. Although the trading rhythm in political markets may not be steady, they generate substantial fee revenue during peak event cycles.

4️⃣ Weather's 7-day estimated cost is approximately $400,000, the lowest among the five tracks.

Therefore, Polymarket’s revenue structure can be simply summarized as: Crypto drives platform revenue, Sport drives trading volume, and Politics/Trump drive viral events that attract new users to the platform.

Polymarket's successful closed loop offers transformative insights for the entire prediction market sector:

1️⃣ Comprehensive overhaul of evaluation metrics

In the past, people tracking prediction markets focused on trading volume and trending topics. In the commercial era, the metrics for success will shift entirely to: real fees, taker ratio, order book depth, and bid-ask spread. Trading volume generated solely through circular trading will no longer be sustainable under a fee-based model.

2️⃣ Different event types correspond to different revenue roles

Future prediction market platforms will not rely on a single order book to dominate the market, but will instead move toward specialized division of labor.

Crypto markets are more akin to financial trading, with rapid price fluctuations and short feedback cycles, making users more sensitive to trade execution speed and thus more likely to achieve high income efficiency.

Sports resemble a steady stream of transactions—frequent events, clear outcomes, and continuous trading scenarios—making them ideal for generating daily trading volume.

Markets related to politics or Trump are more event-driven; they may not be stable on a regular basis, but they tend to experience significant volume spikes during key moments such as elections, polling updates, or policy changes.

Markets like weather demonstrate that, as long as events are sufficiently standardized and outcomes are clearly defined, even if they are initially small in scale, they still have the potential to develop their own trading environments.

3️⃣ The fee structure will incentivize higher order book quality.

During the free phase, the platform can open many trading pairs; after introducing fees, both users and market makers will become more cost-conscious, and the fee structure will naturally filter for higher market quality.

A good prediction market needs not only engaging topics but also the ability to meet several criteria simultaneously:

Clear results for easy settlement

Information is updated frequently and can cause price changes.

Market divergence is significant enough to give users a motivation to trade.

Liquidity is sufficient, and users are willing to place trades actively.

The results are difficult to manipulate.

4️⃣ The barrier to prediction markets lies in "continuous pricing power"

Opening a YES/NO market is not difficult; what’s challenging is sustaining active orders, matching trades, regular price updates, and encouraging participants to take on risk. Only when a market has sufficient depth and trading volume does its price become meaningful as a reference—and only then can the platform potentially generate revenue.

Therefore, the real barrier in prediction markets isn't "who can spot trends faster," but rather: turning trends into tradable markets → ensuring long-term market liquidity → making prices a signal external parties are willing to reference.

Projects that can articulate grand narratives are countless, but those that can turn those narratives into real, tangible revenue are few and far between.

Polymarket was once the most glamorous traffic representative in the entire sector, and now that it has transitioned from a “traffic narrative” to a “systematic money-making machine,” it aims to prove one thing to the entire industry:

The ultimate value of prediction markets lies not merely in "how accurately they predict the future," but in their success at transforming the uncertainty of the real world into a supermarket where risks can be standardized, priced, frequently traded, and consistently profited from.

In the past, prediction markets proved they could attract massive traffic; now, they are proving themselves to be an unrivaled business.

1. Calculation Method: Break down the ratio of market orders to limit orders for each trading pair, then estimate the impact of p × (1 - p) on fees across different price ranges, and finally combine this with the corresponding fee rates for each trading pair to calculate the approximate fee contribution from each pair.

Original authors: Changan, Amelia, Biteye Content Team