Prediction markets have struggled for a decade to escape the "niche toy" label, and the deepest reason is not elegant protocol design, not contract safety, not lack of hot events, but the absence of deep liquidity. A market so thin that ten thousand dollars can push the implied probability from 0.6 to 0.05 cannot serve a hedge fund, cannot provide credible price signals to news organizations and sportsbooks, and cannot convince a retail user that a market order will execute as expected. After building market making and liquidity systems for dozens of prediction market operators, sportsbook platforms, media companies, and financial infrastructure clients, SoonTech has observed that prediction market market making is a distinct engineering discipline: its "underlying" is a binary or multi-outcome conditional token, its "price" is the market-implied probability, its counterparties are event insiders, arbitrage bots, and emotion-driven retail, and its "settlement" is not T+2 cash delivery but post-event token redemption. These characteristics mean market makers must quote using a logic neither identical to spot CEX nor identical to perpetual swaps. This article begins with why liquidity determines success, then walks through outcome probability pricing, LMSR/AMM market making, order book market making, hybrid models, spread and inventory management, maker incentives, holding costs, cold starts, institutional APIs, risk control and position caps, and settlement uncertainty, before closing with the SoonTech stack and an operator implementation checklist.

1. Why Liquidity Determines Prediction Market Success or Failure

The essence of a prediction market is the "financialization of collective intelligence" — aggregating the judgments of thousands of people into a tradable price signal that predicts elections, sports outcomes, crypto milestones, corporate earnings, even wars. For this aggregation mechanism to function, participant judgments must be absorbed, transmitted, and priced. Liquidity is the lifeblood. Without it, prediction markets face three fatal problems simultaneously.

First, price discovery becomes distorted. When the order book is thin, a small order can push the price to an extreme, and the market signal no longer reflects collective judgment but is merely noise distorted by a few large orders and bots. Sophisticated users can immediately tell which markets are "usable" and which are "decorative," so capital flows toward a few deep markets, long-tail markets shrink further, and a death spiral emerges.

Second, user experience collapses. The most common frustration for retail users in prediction markets: they see an attractive odds, click in, and either the execution price is 5% worse or they wait endlessly for a counterparty. AMM slippage is acceptable for small trades, but for medium and large orders it is nearly unusable. Pure order book markets frequently present "apparent quotes with phantom depth."

Third, institutions cannot participate. Hedge funds, sportsbooks, and media data buyers need large execution, deterministic settlement, and stable spreads. A prediction market with only tens of thousands of dollars of depth cannot justify the compliance cost of integration for an institution. Institutional absence further weakens liquidity, creating a vicious cycle.

SoonTech repeatedly emphasizes to clients one judgment: the endgame of prediction markets is a liquidity war. Event design, user growth, UX, and compliance all matter, but any operator that builds solid market making infrastructure and a healthy maker ecosystem will, within three to six months, build a lead competitors cannot easily catch.

2. Fundamentals of Prediction Market Pricing

To understand prediction market market making, one must first understand what "price" actually represents. In a binary prediction market, an event has two outcomes, Yes and No. The system mints a conditional token for each outcome. After settlement, the winning token redeems for one dollar and the losing token expires worthless. Before settlement, the market price of each outcome token is the implied probability of that outcome — for example, the Yes token quotes at 0.62 dollars and the No token at 0.38 dollars, the market implying a 62% probability of the event.

Understanding this is essential for market making. A prediction market maker is not "betting on whether the price will rise or fall"; he is maintaining a two-sided quote around a reasonable probability curve. His counterparties may be "informed traders" with information advantage, or simply emotional retail bettors; the maker's goal is to capture the bid-ask spread while keeping inventory risk within a hedgeable range.

Beyond binary outcomes, there are multi-outcome markets (one winner among multiple candidates), range markets (an asset price within a range), combinatorial markets (compound propositions involving multiple events), and scalar markets(numerical results settled by a linear function). Each has its own pricing mechanism and market making challenges, but the core logic is consistent: market price is the collective pricing by participants, and the maker earns risk premium for providing liquidity.

In a binary market, a clean mathematical property holds: at any moment, the sum of all outcome token prices should approximately equal one dollar (ignoring fees and funding). This is the foundation of LMSR and similar curve designs. In multi-outcome markets, the sum of prices must equal exactly one (completeness constraint), which brings engineering challenges that binary markets do not have.

3. LMSR and AMM Market Making

LMSR (Logarithmic Market Scoring Rule) is the classic prediction market curve proposed by Hanson in 2003. Its core idea: describe the maker's "price sheet" across all outcome tokens using a logarithmic function. Prices shift along a log curve as inventory changes, guaranteeing bounded maximum loss in any direction for the maker, at the cost of an information costfor some orders. The more participants bet on a direction, the more the odds shift toward that direction; the maker effectively "sells insurance" to all participants.

The key parameter of LMSR is the liquidity parameter b. A larger b means smaller price impact per order, deeper depth, higher fill probability, but the maker must reserve a potential loss on the order of b dollars for the maximum inventory of any outcome token. A smaller b means a steeper curve and shallower depth, but smaller potential loss. Operators must select b based on expected market volume, event importance, and making budget.

In engineering, LMSR is typically deployed as an AMM (Automated Market Maker) on chain: users swap stablecoins or collateral for outcome tokens, the AMM maintains a quote curve derived from the log function, and orders fill instantly without waiting for a counterparty. This "algorithm-as-maker" model has several advantages.

First, long-tail markets always have a price. Once the AMM is deployed, regardless of whether anyone posts orders, users can trade immediately. This is critical for cold-starting markets.

Second, no matching engine or order book needed. On-chain AMMs use constant-function formulas to quote; engineering is simple and audit cost is low.

Third, price automatically slides with trade size. This is itself a "dynamic spread" — large orders naturally move the price, signaling that "informed traders are knocking."

But AMM has three fundamental limits.

First, limited depth. Usable AMM liquidity equals the parameter b. To support a 500,000-dollar single trade, b must be set to several hundred thousand, tying up massive making capital.

Second, inventory risk is significant. When the market increasingly favors one outcome, the AMM accumulates large inventory in the unfavorable direction. If the final result lands on that side, the AMM loses. This is exactly analogous to Uniswap accumulating depreciated tokens when a token price falls.

Third, professional maker strategies cannot be directly expressed. Professional makers are accustomed to order books, spread adjustments, and inventory hedging; the AMM's "passive quote" mechanism does not let them fully apply their capabilities.

4. Order Books and Professional Market Makers

In contrast to AMM sits the order book model: the operator runs a Central Limit Order Book (CLOB), professional market makers post two-sided quotes, and users trade against the book. This mirrors traditional stock exchanges and crypto spot CEXs.

The order book model is highly attractive to professional makers.

First, strategies are expressible. Makers can freely set quote prices, quote sizes, cancel conditions, layer multiple price levels, and use iceberg orders to hide true intent.

Second, maker obligations and rewards are quantifiable. The platform can specify minimum quote size, minimum quote time, maximum spread, and maximum response latency, paying rebates and rewards against these metrics. Professional makers excel at this "pay for performance" model.

Third, risk control is more granular. Makers can adjust each quote in real time, price each counterparty individually, and manage exposure per event.

Fourth, institutional integration is easier. Institutional makers use FIX and low-latency APIs; order book platforms can integrate seamlessly. AMM platforms force institutions to understand an extra concept of "curve parameters."

But order books have their own challenges.

First, cold start is extremely difficult. New markets have no counterparty quotes; the book is empty; users arrive to find "no one to sell to."

Second, long-tail markets have no maker. Professional makers only serve markets with high flow, controllable volatility, and small information asymmetry; long-tail and controversial markets go un-served.

Third, the matching engine and latency are critical. Order book platforms require high matching performance, market data distribution, and order routing; the build cost is far higher than AMM.

Fourth, counterparty risk and fairness concerns. Makers may use latency, wash trades, or cancellations to capture undue profit. Platforms need matching priorities and last-look rules to prevent abuse.

5. Hybrid Market Making Models

Since AMM and order books each have strengths and weaknesses, the industry has settled on hybrid market making as best practice. The core idea: AMM as floor, order book takes priority.

Specifically, the hybrid flow works as follows.

First, when a user submits an order, the Smart Order Router (SOR) first queries the order book for fillable opposite-side orders and executes at the best price.

Second, where the order book depth is insufficient, the AMM fills the remainder. The AMM plays the role of "last-resort maker."

Third, order book makers receive price improvement over the AMM — they can quote at prices better than the AMM, attracting users to fill against them first.

Fourth, professional makers access the book via API and can also arbitrage against the AMM in both directions: when the AMM price deviates from the maker model, the maker posts on the book to attract arbitrageurs, or trades directly against the AMM.

The hybrid model simultaneously solves the pain points of AMM and order book. Long-tail markets get baseline depth from AMM, avoiding the "no price" problem. Hot markets get tight spreads and deep books from professional makers. Makers have both service opportunity and arbitrage space. Users get both instant fill and tight spreads.

Across multiple client deployments, SoonTech has observed that after launching the hybrid model, market depth typically increases by an order of magnitude within two to three weeks, spreads narrow 30% to 50%, and competition among makers prevents any single party from abusing dominance.

The hybrid model also creates a useful side effect: AMM and order book makers naturally supervise each other. If order book makers quote too wide, users flow to the AMM. If the AMM quotes too wide, makers quote on the book and arbitrage against the AMM, driving convergence. This "competitive market making" is hard to achieve with a single model.

6. Spread Models and Inventory Management

Whether order book makers or AMM "smart quoters," they all need a spread model to determine bid-ask width. Spread is the sum of cost, risk, and target profit.

A professional spread model typically considers the following inputs.

First, basic hedging cost: matching fees, withdrawal fees, gas, bridge fees — at minimum the maker must cover these in the spread.

Second, volatility: implied vol, historical vol, and external reference vol. Higher vol, wider spread.

Third, time to expiry: as expiry approaches, prices converge toward 0 or 1, spreads can narrow; longer-dated events require wider spreads to compensate for holding time.

Fourth, current inventory: the more AMM inventory deviates from neutral, the more the corresponding direction quote must skew to invite rebalancing.

Fifth, information asymmetry risk: certain events (celebrity health, internal decisions) carry materially higher insider risk, requiring wider spreads.

Sixth, settlement risk: if the event is contentious or settlement is unclear, the maker must price in "invalid settlement" or "delayed settlement" risk.

Inventory management is the central application of the spread model. When a maker accumulates excessive inventory in one outcome, he must actively adjust quotes: post lower offers and lift lower bids in the over-inventoried direction, inviting counterparties to help the maker reduce exposure. In the under-inventoried direction, do the opposite. SoonTech's inventory module supports multi-dimensional limits: per market, per event, per maker, per counterparty, and platform total. All limits can be tiered by time and event.

Another key point of inventory management is hedging. A prediction market maker cannot simply "reverse trade in external markets" the way a spot maker can, because external markets (perps, spot, ETFs) may not perfectly correlate with prediction market outcomes. Professional prediction market makers must build a correlation matrix: when an event outcome correlates with certain asset prices (e.g., "Fed cuts rates" and US Treasury yields), the maker can hedge through those assets. When an event has no tradable correlation (e.g., "will a celebrity divorce"), the maker can only rely on in-platform reverse traders to hedge, which means spreads must be wider.

7. Market Maker Incentives, Rebates, and Tokenomics

Prediction market makers face not only market risk but also the question of whether incentives are sufficient. A 50,000-dollar capital pool can service millions of dollars of two-sided quotes in spot CEX, but in prediction markets it can only service comparable volume — because event markets are naturally narrow, naturally low in flow, and naturally high in inventory risk. Without sufficient platform incentives, professional makers prefer to deploy capital into spot or perps.

In designing client incentive programs, SoonTech typically uses the following layers.

First, maker rebate: rebate part of the trading fee based on the maker's volume, quote time, minimum quote size, and other metrics; some platforms use negative fees (makers receive 0.5 to 2 basis points per fill).

Second, event sponsor subsidy: for high-value events, the event provider (sports league, crypto project, media) can sponsor a maker pool, specifically incentivizing that event.

Third, liquidity bootstrapping pool (LBP): a dedicated incentive pool that pays rewards based on the maker's quality-adjusted depth contribution.

Fourth, platform token incentive: tie maker rewards to the platform token, incentivizing long-term commitment and creating real demand for the platform token.

Fifth, tiered maker obligations: tier makers by capital, quote quality, and stability; higher tier makers receive higher rebates, priority execution, and exclusive event access.

The key to incentive design is avoiding low-quality "post and collect" makers. Paying purely on quote size incentivizes makers to post "ghost quotes" far from the market. The system must combine spread, response time, cancellation rate, and other metrics. SoonTech's maker scoring engine computes each maker's "quality score" in real time and dynamically adjusts rebate rates.

The flip side of incentives is penalties. Makers who fail to meet quote obligations, quote too wide, deliberately create false depth, or fail to cancel in time may have rewards docked, be downgraded, or have cooperation terminated. Clear, transparent assessment mechanisms filter out partners who truly want to make markets.

8. Funding, Holding Costs, and Expiry Convergence

A prediction market maker holds a basket of outcome tokens until event settlement, during which he faces holding cost. Cost sources include.

First, capital occupation: maker capital is tied up in outcome tokens and cannot earn returns in external markets.

Second, opportunity cost: that capital could have earned higher returns in more active spot or perpetual markets.

Third, settlement uncertainty: until settlement, the event always carries "cancellation," "invalid," and "delay" risk; the maker must price these.

Some platforms introduce funding rate mechanisms to balance maker and speculator costs: when long exposure to a token significantly exceeds short, longs pay shorts funding, and vice versa. This mirrors traditional perpetual funding, but with a slightly different meaning: in prediction markets, funding is more about balancing directional imbalance than true funding cost.

Expiry convergence is one of the most critical features of prediction market making. As the event approaches settlement, outcome token prices gradually converge to 0 or 1, and the maker can narrow spreads to extremely tight levels. In the last day or two before expiry, the book often shows "0.98 vs 0.99," and almost every maker fill is essentially risk-free. But the last moments before expiry are also when informed traders are most active: they hold the latest information, know the outcome, and pose the greatest adverse selection risk to the maker.

SoonTech's making engine automatically narrows spreads as expiry approaches, while enforcing strict risk control on large orders near settlement: limit single fill size, require sufficient counterparty margin, monitor abnormal fill patterns. This "dual adjustment of spread and risk control" lets the maker safely capture "near-risk-free" returns before expiry while preventing insider trading.

9. Cold Starts and Long-Tail Market Liquidity

The hardest part of prediction market making is the cold start. A new platform, a new batch of events, a long-tail theme — no counterparties, no makers, no users. How does the making infrastructure go from zero to one?

SoonTech's client practice has produced a "five-step method" for cold start.

Step one, platform proprietary making launch: deploy an internal making engine with platform capital, posting initial two-sided quotes on new markets. Even a simple spread strategy builds basic "there's a price to buy" confidence for users.

Step two, liquidity bootstrapping pool: set up a dedicated incentive fund, paying rewards based on making contribution to attract external makers.

Step three, event sponsor subsidy: partner with event providers (sports leagues, crypto projects, media) to sponsor making pools for their events, reducing platform-side making pressure.

Step four, time-limited rebates: offer time-limited high rebates for new markets, new makers, and new users, stimulating early trading behavior; gradually reduce to normal levels as the market matures.

Step five, tiered launch: first launch to invited users, makers, and community in a small scope, test liquidity, refine spreads, identify issues, then open to the public.

Long-tail market liquidity is an extension of cold start. Prediction markets may host thousands of events, but only a few dozen are "high activity" markets, while most see only a few fills per day. Professional makers have no incentive to serve these markets, and the platform cannot afford human market makers for every long-tail event.

SoonTech's solution is "AMM floor + smart making": long-tail markets are served by AMM for baseline depth, with the platform layering a "parametric maker" on top. The parametric maker auto-adjusts the AMM's b value and skew parameters based on event flow expectations, volatility, and controversy. When a long-tail market suddenly goes viral (e.g., a niche event trends on social media), the system detects the flow spike and invites professional makers to enter rapidly.

10. Institutional Market Making, APIs, and Custody

The institutionalization of prediction market making is a clear trend since 2024. More and more hedge funds, sportsbooks, and family offices are treating prediction markets as an alternative asset class, while institutional makers (Citadel Securities, Susquehanna, Jump Crypto, Wintermute, etc.) are actively evaluating entry.

Institutional making demands a much higher bar than retail making, mainly across the following dimensions.

First, low-latency API: institutional maker algorithms written in C++/Rust are highly latency-sensitive, requiring WebSocket market data, REST order entry, FPGA acceleration, and co-location.

Second, FIX protocol support: traditional financial institution making systems are built on FIX; the platform must provide a FIX gateway or FIX-to-API bridge.

Third, custody and settlement: institutional capital must be held by qualified custodians (banks, licensed trusts, qualified crypto custodians); the minting, transfer, and redemption of outcome tokens must follow clear custody flows.

Fourth, compliance and KYC: institutional makers must meet AML, KYC, and Market Abuse Regulation (MAR) requirements; the platform must provide full compliance interfaces.

Fifth, credit lines and counterparty management: institutional makers want credit lines, fill-then-settle; the platform must build counterparty risk assessment and dynamic margin models.

Sixth, risk limits and hedging convenience: institutional makers want to hedge inventory risk within the platform or across platforms; the platform must support portfolio margin, cross-market hedging, and internal transfers.

Seventh, audit and reporting: institutional makers are regulated by PCAOB, SEC, FCA, and others; the platform must provide full trade records, statements, tax reports, and incident audits.

SoonTech provides a complete institutional access kit: FIX 4.4 gateway, low-latency WebSocket market data, REST order entry, multi-signature custody integration, KYC/AML integration, cross-market hedging API. Institutional makers can complete integration and go live within weeks.

11. Risk Control, Manipulation Prevention, and Position Caps

One of the most severe risks in prediction market making is manipulation. Unlike spot CEX, the "price" in a prediction market represents a judgment about an event, and the event itself may be influenced or fabricated by a few. Typical manipulation patterns include.

First, event-provider manipulation: an event provider (e.g., a project) buys large amounts of "Yes" tokens to make itself appear "popular." The maker may not lose, but the market signal is corrupted, eroding user trust over time.

Second, informed trader manipulation: traders with insider information (advance knowledge of election results, match scores, merger announcements) bet heavily on the market; when the outcome confirms, the market reverses sharply and the maker is caught.

Third, bot collusion: a few accounts, through coordinated quoting, wash trading, and cancellation, create false depth, luring users to fill at disadvantageous prices.

Fourth, last-moment manipulation: in the final minutes before settlement, a single large order pushes the price to an extreme, then reverts after settlement; arbitrators profit while the maker loses.

Preventing manipulation requires the platform to act on institutional, technical, and economic levels simultaneously.

Institutional: explicitly prohibit event providers from trading on events they provide; require all makers and maker accounts to complete KYC; set settlement observation periods for major events (large prizes or sensitive topics).

Technical: SoonTech's risk control system monitors each order's source account, counterparty relationships, and historical behavior in real time, identifying suspicious coordinated quoting and wash trades. Large orders in the last moments trigger trade pauses and cool-down periods. On-chain behavior analysis identifies bot accounts.

Economic: position caps are the core tool. Set a per-account max holding on each outcome token to prevent single-account price manipulation. Set a per-direction max volume per event. Set a max inventory per maker. Set a max single fill in the final hours before settlement.

SoonTech's risk control system supports a "hard limit + soft limit + temporary limit" three-tier structure: hard limits are system-enforced and directly reject orders; soft limits require maker manual confirmation; temporary limits are dynamically adjusted for hot events or abnormal periods.

12. How Settlement Uncertainty Affects Market Making

One of the biggest differences between prediction markets and traditional financial products is settlement uncertainty. Stock contracts cash-settle at closing price on expiry, with unambiguous results. Prediction market settlement depends on one or more oracles reporting the event outcome, and the oracle itself may err, delay, or interpret the event in multiple ways.

Settlement uncertainty affects market making along three dimensions.

First, cost of settlement delay. From event confirmation to actual outcome token redemption, there may be hours to days of delay. During this time the maker's capital is locked in a "nearly settled" state, neither redeployable nor transferable. The longer the delay, the higher the additional compensation the maker requires.

Second, risk of settlement dispute. The "outcome" of certain events (especially political, sports, and corporate events) is itself controversial: elections may have recounts, matches may have appeals, corporate announcements may be revised. Makers must price in "invalid settlement" or "revised settlement" risk. SoonTech's solution lets makers set a "settlement risk premium" per event, automatically folded into the spread.

Third, risk of settlement source. If settlement depends on a single oracle, an oracle attack or failure can paralyze the entire market. SoonTech recommends clients adopt a "primary-backup multi-source" settlement architecture: the primary oracle reports the result, the backup independently verifies. If primary and backup agree, settle. If they disagree, enter the dispute process, during which makers maintain wide spreads.

Settlement uncertainty also affects the "invalid probability" of outcome tokens. In the last stage before settlement, the maker must leave probability space for "the event may be ruled invalid." For example, before settlement the market shows "Yes 0.95, No 0.04," but the maker may cap the highest bid on the Yes token at 0.90 — because he reserves the remaining 0.05 for the chance of "invalid settlement." Once settlement is confirmed valid or invalid, the price converges to 0 or 1, and the maker captures the 0.90 to 1.00 (or 0.00) difference as compensation for hedging this risk.

SoonTech's solution automatically enters a "settlement observation period" before settlement: makers are forced to widen spreads, single fill limits are tightened, all unfilled quotes are frozen, and the platform suspends new large orders, avoiding violent moves when settlement disagrees with market expectation.

13. SoonTech's Prediction Market Liquidity Stack

Across all the modules above, SoonTech provides prediction market operators with a complete liquidity infrastructure.

First, hybrid AMM + order book engine: the platform deploys the LMSR curve by default as the floor maker, while supporting order book priority fill, professional maker integration, and maker-AMM two-way arbitrage. The b value and spread parameters are dynamically configurable by event type, flow expectation, and controversy.

Second, maker API and FIX gateway: complete low-latency APIs for professional makers — WebSocket market data, REST order entry, FIX 4.4 gateway, batch order, cancel, modify interfaces. Co-location and hardware acceleration supported.

Third, institutional custody integration: supports bank-grade custody, qualified crypto custody, and multi-signature custody; the full lifecycle of outcome token minting, transfer, and redemption is on-chain auditable.

Fourth, configurable maker obligations and rebates: operators can customize maker obligations and rebate tiers based on capital, quote quality, stability, and response speed. Rebates can be paid in fiat, stablecoin, or platform token.

Fifth, collaboration with oracles and arbitration layers: SoonTech integrates deeply with UMA, Chainlink, Kleros, and other leading oracle and arbitration protocols, supporting primary-backup multi-source settlement, with automatic "settlement observation period" mode during disputes.

Sixth, risk control and position caps: built-in five-tier position limits across account, event, direction, maker, and platform; hard, soft, and temporary limit structures; real-time monitoring of abnormal fill patterns, with automatic circuit breakers.

Seventh, cold start and long-tail market tools: liquidity bootstrapping pool configuration, event sponsor subsidy module, time-limited rebate engine, and parametric maker (auto-configuring AMM parameters for long-tail markets) to help new and long-tail markets quickly gain baseline liquidity.

Eighth, data and monitoring: full-dimensional data interfaces including maker performance dashboard, depth heatmap, spread distribution, inventory risk exposure, counterparty analysis, and settlement timing monitor.

The core philosophy of the SoonTech stack: let operators focus on event design, user growth, compliance, and brand, and leave making and liquidity to professional infrastructure. Multiple client cases show that after deploying the SoonTech liquidity stack, market depth increases 5 to 20x on average, maker count grows by 2 to 4, and institutional onboarding time shrinks from 6 months to 6 weeks.

14. Operator Implementation Recommendations

If you are a prediction market operator looking to move liquidity from "zero" to "deep," the following five steps can guide you.

Step one, tier events and plan liquidity budget. Divide all events into three layers: core high-activity events (80% of flow) served by professional makers; medium-activity events (15%) covered by AMM floor + smart maker; long-tail events (5%) defaulted to AMM. Allocate making budget accordingly.

Step two, deploy hybrid making infrastructure. Do not choose between AMM and order book; the combination covers both depth and tight spreads. Onboard 2 to 3 professional makers in the first batch, signing clear maker obligation and rebate agreements.

Step three, build maker assessment and incentive system. Score by "quality score = quote size × quote time × spread compliance rate × response speed." Higher score, higher rebate. Continuous low-score makers get downgraded or exited.

Step four, design cold start and long-tail strategy. Before each new event launches, complete the "five-step method": platform proprietary making, liquidity bootstrapping pool, event sponsor subsidy, time-limited rebate, and tiered launch. Long-tail markets use the parametric maker for auto-management.

Step five, incrementally onboard institutional makers. After flow and risk control mature, onboard institutional makers, open the FIX gateway and credit lines, lifting liquidity to institutional grade.

Step six, establish settlement observation period and dispute resolution flow. Interface with oracle and arbitration layers, clarify settlement source, dispute period length, and final ruling mechanism. During the observation period, makers are forced to widen spreads and tighten single-fill limits.

By executing these five to six steps diligently, transforming a prediction market from "niche toy" to "institutional-grade risk pricing market" is no longer wishful thinking. Making infrastructure determines the endgame of prediction markets. Every unit of energy operators invest in this area will return tenfold in depth and flow within three to six months.

🌐 Build secure and scalable Web3 platforms with SoonTech.

Explore our solutions for White Label Crypto Exchanges, Prediction Markets, MPC Wallets, Matching Engines, Liquidity Integration, and Compliance.

FAQ

Q1: How do prediction market makers actually make money?

A: The main revenue source for prediction market makers is the bid-ask spread: posting bids and asks on the book and capturing the spread on each fill. Beyond that, makers earn rebate income (the platform rebates part of the trading fee based on making contribution), event sponsor subsidies (certain high-value events are subsidized by the event provider), and inventory management returns (by actively managing outcome token inventory, makers capture returns as positions converge to neutral before expiry). It is important to emphasize that makers do not make money by "judging event outcomes" — that is the speculator's job. Makers are liquidity providers who earn long-term spread income in exchange for short-term inventory risk.

Q2: Can LMSR be "bought out"? Will the maker lose money?

A: Yes. An inherent feature of LMSR is that the maker may accumulate large inventory in an extreme direction, and if the event outcome lands on that side, the maker loses. But LMSR is designed to guarantee bounded maximum potential loss for the maker (determined by the liquidity parameter b), and will not be completely drained in extreme markets the way a traditional AMM might. In practice, makers use spread widening, hedging, and limit management to keep losses within acceptable ranges. The SoonTech solution also supports "adaptive b": when inventory in one direction accumulates too much, the AMM automatically raises the b value for that direction, encouraging counterparty reverse trades.

Q3: Why are some prediction market books so wide?

A: Wide spreads usually have four causes: flow is too small (long-tail markets have no professional makers willing to serve), high controversy (event outcomes have multiple plausible interpretations, so makers demand higher risk premium), unclear settlement (depends on a single oracle or the event provider has manipulation risk), and far from expiry (makers must charge for longer holding time). The SoonTech hybrid model auto-floors long-tail markets with AMM, but the AMM's "dynamic spread" also widens on large orders, so long-tail spreads are naturally wider. The most effective ways to narrow spreads are adding more makers and increasing event flow.

Q4: What should makers do during a settlement dispute period?

A: The settlement dispute period is the toughest stage for makers. The event outcome is undefined, outcome tokens cannot redeem, and capital is locked. Makers need to do several things. First, automatically enter "settlement observation period" mode, with the platform forcing makers to widen spreads, tighten single-fill limits, and suspend large order intake. Second, restrict new quoting until settlement is clear, avoiding taking new inventory on uncertain outcomes. Third, prepare for arbitration — if the primary and backup oracles disagree, the arbitration process must be initiated, and makers must decide per platform rules whether to participate or withdraw in advance. Fourth, prepare for post-settlement re-entry — once the outcome is confirmed, makers quickly resume normal quoting and capture compensating returns from the spread reversion.

Q5: Can retail users provide liquidity?

A: Yes, with the right tools. The SoonTech solution provides two retail participation paths. First, "copy market making" — retail users authorize their capital pool to a professional maker to quote on their behalf, earning returns based on contribution. Second, "parametric maker pool" — retail users participate in a maker pool using platform-preset parameter templates (grouped by event type, risk level, expected return). Retail returns are typically lower than professional makers (because professionals have scale, technology, and relationship advantages), but for users seeking "passive making returns" it is a viable path. Before participating, retail users must fully understand the settlement risk of outcome tokens and the potential impact of inventory loss.

Q6: Should the platform make markets itself?

A: Yes, but only "seed making." Platform proprietary making has three values. First, providing initial liquidity — in cold-starting new markets, professional makers are not yet willing to enter, so the platform must make markets itself. Second, anchoring maker incentives — the presence of a platform proprietary maker gives external makers a price reference, avoiding extreme quotes. Third, hedging risk and inventory adjustment — when professional makers develop arbitrage imbalances, the platform can stabilize via proprietary making. But platform proprietary making cannot replace professional makers: professionals have more sophisticated strategies, broader hedging tools, and finer risk control, and are the main force of liquidity in the long run. SoonTech recommends platform proprietary making budget at 10% to 30% of total making budget, with the rest attracted from external professional makers via incentives and rebates.

Conclusion

Prediction market making is a young discipline, but its importance is growing rapidly as the asset class matures. Operators that treat liquidity as a strategic asset — not a side project — will be the ones that build institutional-grade risk pricing markets. SoonTech's full-stack solution — hybrid AMM + order book, institutional APIs, custody, configurable incentives, oracle and arbitration integration, risk control, cold start tooling — gives operators everything they need to turn illiquid markets into deep, healthy, defensible businesses. The endgame belongs to operators who invest in making infrastructure today.

🌐 Build secure and scalable Web3 platforms with SoonTech.

Explore our solutions for White Label Crypto Exchanges, Prediction Markets, MPC Wallets, Matching Engines, Liquidity Integration, and Compliance.