Market forecasting is a tough job: Less than 10% of those who do it may actually survive until the end of the year

Original Article Title: "Only 10% of Prediction Markets Can Survive Until Year End, Not an Exaggeration"

Original Article Author: Azuma, Odaily Planet Daily

Over the past two days, there has been a lot of discussion on Twitter about the formula for prediction markets: Yes + No = 1. This discussion was sparked by a prominent figure, DFarm (@DFarm_club), who wrote an article dissecting Polymarket's shared order book mechanism, triggering a collective emotional resonance regarding the power of mathematics: "A Comprehensive Explanation of Polymarket: Why Must YES + NO Equal 1?".

In the subsequent discussions, many prominent figures, including Lan Hu ( @lanhubiji ), mentioned that Yes + No = 1 is another simple yet powerful formula innovation after x * y = k. It is expected to unlock a trillion-dollar information trading market. I completely agree with this point, but at the same time, I feel that some discussions are overly optimistic.

The key lies in the issue of liquidity provision. While many people may think that Yes + No = 1 solves the barrier to entry for ordinary people to provide liquidity, therefore, the liquidity of the prediction market will surge like an AMM with x * y = k, the reality is far from that.

The Difficulty of Providing Liquidity in Prediction Markets

In practical terms, the ability to enter the market as a liquidity provider and build liquidity is not just a matter of the participation threshold but also an economic question of profitability. Horizontally comparing the AMM market based on the x * y = k formula, the difficulty of providing liquidity in the prediction market is actually much higher than the former.

For example, in a classical AMM market that fully follows the x * y = k formula (such as Uniswap V2), if I want to provide liquidity for the ETH/USDC trading pair, I need to deposit ETH and USDC into the pool in a specific ratio based on the real-time price relationship between the two assets in the pool. When this price relationship fluctuates, the amount of ETH and USDC I can withdraw will fluctuate accordingly (known as impermanent loss), but I can also earn trading fees. Of course, the industry has made many innovations around this basic x * y = k formula, such as Uniswap V3 allowing liquidity providers to stack liquidity within a specified price range to pursue a higher risk-adjusted return, but the fundamental model remains unchanged.

In this kind of Automated Market Maker (AMM) model, if transaction fees over a certain time range can cover impermanent loss (often requiring a longer time to accumulate fees), then it is profitable — as long as the price range is not too volatile, I can lazily provide liquidity, only checking in occasionally. However, in a prediction market, if you approach liquidity provision with a similar mindset, you will likely end up losing.

Take Polymarket as an example. Let's consider a basic binary market where, for instance, I want to provide liquidity in a market where the "YES real-time market price is $0.58." I can place a buy order for YES at $0.56 and a sell order for YES at $0.6 — as DFarm explained in the article, this is essentially placing a buy order for NO at $0.4 and a sell order for NO at $0.44 — that is, using the market price as a reference to provide order support at slightly wider specific price points.

Now that the orders are placed, can I just leave them be? The next time I check, I may encounter one of the following four scenarios:

· Both orders remain unfilled;

· Both orders are filled;

· One side's order is filled, and the market price remains within the original order range;

· One side's order is filled, but the market price has moved further away from the remaining order — for example, buying YES at 0.56 while the sell order at 0.6 is still there, but the market price has dropped to 0.5.

So, when can you make money? I can tell you that different scenarios may lead to different results in low-frequency attempts, but if you consistently operate with such laziness in a real environment, the ultimate outcome is likely to be a loss. Why is this the case?

The reason is that the prediction market is fundamentally different from the AMM liquidity pool mechanism; it is more akin to a Centralized Exchange (CEX) order book liquidity provision model, with entirely different operating mechanisms, operational requirements, and risk-reward structures.

· In terms of operating mechanism, AMM liquidity provision involves depositing funds into a liquidity pool to collectively provide liquidity, spreading liquidity across different price ranges based on the x * y = k formula and its variants; order book liquidity provision requires placing buy and sell orders at specific price points, requiring orders to provide liquidity support, and achieving trades through order matching.

· In terms of operational requirements, AMM liquidity provision only requires depositing both tokens into the pool within a specific price range, and it remains effective as long as the price stays within the range; order book liquidity provision necessitates active and ongoing order management, continuously adjusting quotes to respond to market changes.

· In terms of risk-reward composition, AMM liquidity provision mainly faces impermanent loss risk, earning fees from the liquidity pool; order book liquidity provision, on the other hand, needs to deal with inventory risk in a one-sided market, with profits coming from the bid-ask spread and platform subsidies.

Building on the assumption from the previous section, if I know that the main risk I face as a liquidity provider on Polymarket is inventory risk, and that my profits mainly come from the bid-ask spread and platform subsidies (Polymarket provides liquidity subsidies to orders close to market price, see official website for details), then the potential profit and loss scenarios for the four situations are as follows:

· In the first scenario, you are unable to capture the bid-ask spread but can benefit from liquidity subsidies;

· In the second scenario, you have profited from the bid-ask spread but will not receive further liquidity subsidies;

· In the third scenario, you have taken a YES or NO position, establishing a directional position (i.e., inventory risk), but in some cases, you can still receive certain liquidity subsidies;

· In the fourth scenario, you have also taken a directional position, incurred unrealized losses, and are no longer receiving liquidity subsidies.

Two additional points to note here. Firstly, the second scenario actually always evolves from the third or fourth scenario because often only one side of the order will be executed first, temporarily resulting in a directional position. However, the risk does not manifest, as the market price subsequently moves in the opposite direction to trigger the orders on the other side. Secondly, compared to the relatively limited gains from liquidity provision (spread profits and subsidy amounts are often fixed), the risk from directional positions is often unlimited (the upper limit being that your YES or NO holdings can go to zero).

Therefore, if I want to continuously profit as a liquidity provider, I need to actively seek out profit opportunities and avoid inventory risk—which means I must actively optimize strategies to maintain the first scenario as much as possible, or swiftly adjust order ranges after one side of the order is filled to convert it to the second scenario, avoiding prolonged exposure to the third or fourth scenarios.

Achieving this consistently is not easy. Liquidity providers must first understand the structural differences of different markets, comparing subsidy levels, price volatility, settlement times, arbitration rules, and more; then, they need to more accurately and rapidly track or even predict price changes based on external events and internal fund flows; subsequently, they must promptly adjust orders proactively in response to changes, all the while designing and executing risk management for inventory risks... all of which clearly surpass the capabilities of an average user.

A Wilder, Jumpier, Less Sportsmanlike Market

If that were the only issue, it would still be fine, after all, the order book mechanism is not a new invention. In CEXs and Perp DEXs, the order book is still the mainstream market-making mechanism. Liquidity providers active in these markets can easily migrate their strategies to prediction markets to continue making profits while injecting liquidity into the latter. However, the reality is not that simple.

Let's think about this issue together. What is the market maker's biggest fear? The answer is very simple—one-sided market movements because such movements often continuously increase inventory risk, leading to a breakdown of the asset allocation balance and resulting in huge losses.

However, compared to the traditional cryptocurrency trading market, the prediction market is inherently a wilder, more erratic, and less polite place. One-sided markets always appear more exaggerated, abrupt, and frequent.

By "wilder," it means that in the regular cryptocurrency trading market, if you were to extend the timeline, mainstream assets would still show a certain oscillating trend, with price fluctuations often cycling in periods. In contrast, the trading assets in the prediction market are essentially event contracts, each contract having a clear settlement time. Additionally, the formula Yes + No = 1 determines that ultimately, only one contract's value will become $1, with all other options going to zero—this means that bets in the prediction market will eventually end at some point with a one-sided market movement. Therefore, market makers need to more rigorously design and implement inventory risk management.

By "more erratic," it means that the volatility in the regular trading market is determined by the continuous game between emotions and funds. Even if the fluctuations are severe, the price changes are continuous, providing market makers with room to adjust inventory, control spreads, and dynamically hedge their positions. However, the volatility in the prediction market is often driven by discrete real-world events, leading to price changes that are more abrupt—the price could be at 0.5 one second, and then in the next second, a real-world development could instantly move it to 0.1 or 0.9. Many times, it's very difficult to predict at which time the order book will experience a dramatic change due to which event, leaving very little time for market makers to react.

By "less polite," it means that there are many insider players in the prediction market who are close to the information source or are the information source themselves. They are not engaging in transactions with counterparties based on their future market predictions; instead, they enter trades with a clear outcome in mind. Market makers are inherently at an informational disadvantage against these players, and the liquidity they provide ironically becomes the channel for these insiders to cash out. You might ask, don't market makers have insider information? This is a typical paradox. If I had insider information, why would I bother market-making? Betting directly on the outcome would yield more profits.

It is precisely because of these characteristics that I have long been more inclined to agree with the statement that "the design of prediction markets is not very friendly to liquidity providers," and I strongly advise against ordinary users easily participating in liquidity provision.

So, is providing liquidity in prediction markets not profitable at all? Not exactly. Buzzing founder Luke (@DeFiGuyLuke) has revealed that, based on market experience, a relatively conservative estimate is that a Polymarket liquidity provider can earn fees equivalent to about 0.2% of the trading volume.

So, in simple terms, this is not an easy money-making opportunity. Only professional players who can accurately track market changes, promptly adjust order statuses, effectively implement risk management, can sustain operation over a longer period of time and make money based on real skills.

The Challenging Path of Prediction Markets, Where Variety May Be Scarce

The liquidity provision challenge in prediction markets not only places higher demands on liquidity providers' abilities but also presents a liquidity-building challenge for platforms.

The difficulty of providing liquidity implies restricted liquidity building, which directly impacts users' trading experience. To address this issue, leading platforms such as Polymarket and Kalshi have chosen to invest substantial sums to subsidize liquidity to attract more liquidity providers.

Focusing on the prediction market landscape, analyst Nick Ruzicka cited a Delphi Digital research report in November 2025, stating that Polymarket has invested approximately $10 million in liquidity subsidies, once paying over $50,000 per day to attract liquidity. With its leading position and brand effect consolidated, Polymarket has significantly reduced the subsidy intensity, but on average, it still needs to subsidize $0.025 for every $100 of transaction volume.

Kalshi also has a similar liquidity subsidy program and has allocated at least $9 million for this purpose. In addition, Kalshi leveraged its regulatory advantage in 2024 (Kalshi is the first prediction market platform to receive CFTC regulatory approval; in November 2025, Polymarket also received approval) and signed a market-making agreement with Wall Street's top market maker, Susquehanna International Group (SIG), significantly improving the platform's liquidity situation.

Whether for treasury reserves or compliance thresholds, these are the moats that powerhouse platforms like Polymarket and Kalshi have concretely established — just a few months ago, Polymarket secured a $2 billion investment from ICE, the parent company of the New York Stock Exchange, at an $8 billion valuation, with rumors of another funding round at a valuation exceeding $10 billion, while Kalshi, on the other hand, has raised $300 million at a $5 billion valuation, showcasing ample ammunition in the hands of these two key players.

The prediction markets have now become a focal point in the entire market, with various new projects constantly emerging, but I am not very optimistic. The reason being that the network effect of the prediction market is actually stronger than many people realize. Faced with the substantial subsidies of established players like Polymarket and Kalshi, as well as partnerships stemming from the compliance world, what do new projects have to compete head-on? And how much capital do they have to endure? While some new projects may have the backing of a true powerhouse and strike gold, clearly not all of them do.

A few days ago, Haseeb Qureshi, the bald-headed partner at Dragonfly, posted his forecast for 2026, where he wrote, "The prediction market space is growing rapidly, but 90% of prediction market products will go completely unnoticed and gradually disappear by the end of the year." I don't know what his reasoning is behind this prediction, but I agree that it is not an exaggeration.

Many are eagerly anticipating a flourishing prediction market race and envisioning profiting from past experiences, but such a scenario may be hard to come by. Instead of spreading bets around, it might be more effective to focus directly on the key players.

Original Article Link