Case Study 2: Fee Structure Design for a New Platform

Overview

A startup called "ForecastX" is launching a new prediction market platform and needs to design its fee structure. The founding team must balance three competing objectives:

1. Revenue: Generate enough fee income to sustain the business
2. Liquidity: Attract market makers who provide tight spreads and deep order books
3. Trader participation: Attract active traders who generate volume

This case study walks through the quantitative analysis of different fee structures, modeling their impact on platform economics, trader behavior, and market quality.

Background: ForecastX Platform Specifications

Required monthly revenue (to cover costs + margin): $$R_{\text{required}} = \frac{\$130{,}000}{1 - 0.30} = \$185{,}714$$

At 500,000 contracts per month, this requires: $$\text{Revenue per contract} = \frac{185{,}714}{500{,}000} = \$0.371$$

At 5,000,000 contracts per month (Year 2): $$\text{Revenue per contract} = \frac{185{,}714}{5{,}000{,}000} = \$0.037$$

Part 1: Fee Structure Options

The team evaluates five potential fee structures, each modeled after real-world platforms:

Option A: Flat Taker Fee (Kalshi Model)

• Maker fee: $0.00
• Taker fee: $0.02 per contract
• Resolution fee: $0.00
• Withdrawal fee: $0.00

Option B: Percentage of Notional

• Trading fee: 2% of notional value (both sides)
• Resolution fee: $0.00
• Withdrawal fee: $0.00

Option C: Profit Fee (PredictIt Model)

• Trading fee: $0.00
• Profit fee: 8% of profits at resolution
• Withdrawal fee: 3% of withdrawal amount

Option D: Maker-Taker with Rebate

• Maker fee: -$0.005 (rebate)
• Taker fee: $0.015
• Resolution fee: $0.00
• Withdrawal fee: $0.00

Option E: Hybrid Model

• Maker fee: $0.00
• Taker fee: $0.01
• Resolution fee: $0.005 per contract (on winning side only)
• Withdrawal fee: $0.00

Part 2: Revenue Modeling

2.1 Volume Assumptions

Volume depends on fees — higher fees discourage trading. We model this using an elasticity parameter $\epsilon$, where a 1% increase in effective trading cost reduces volume by $\epsilon$%.

Based on industry research, we estimate: - Price elasticity of demand for prediction markets: $\epsilon \approx -0.8$ (moderately elastic) - Baseline volume at zero fees: 700,000 contracts/month

The volume model: $$V(c) = V_0 \times \left(\frac{c_0}{c}\right)^{|\epsilon|}$$

where $c$ is the effective cost per contract and $c_0$ is the reference cost.

More practically, we estimate volume for each fee structure:

Effective cost for Option C is estimated based on average win rate and profit margins. *Option D's lower effective cost attracts more volume due to the maker rebate attracting liquidity providers.

2.2 Revenue Calculations

Option A: Flat Taker Fee

Assuming 60% of volume is taker (40% maker): $$R_A = 0.02 \times 550{,}000 \times 0.60 = \$6{,}600/\text{month}$$

Option B: Percentage of Notional

$$R_B = 0.02 \times 0.50 \times 500{,}000 \times 2 \text{ (both sides)} = \$10{,}000/\text{month}$$

Wait — the 2% applies to each side's notional: entry and exit. If average price is $0.50 and each contract trades an average of 2 times (entry + exit): $$R_B = 0.02 \times 0.50 \times 500{,}000 = \$5{,}000/\text{month (entry only)}$$ $$R_B = \$5{,}000 \times 2 = \$10{,}000/\text{month (entry + exit)}$$

Option C: Profit Fee

Average profit per winning contract: $(1.00 - 0.50) \times 0.50 = \$0.25$ (half the contracts win at an average price of $0.50$)

Revenue from profit fee: $0.08 \times 0.25 \times 450{,}000 \times 0.50 = \$4{,}500/\text{month}$

Revenue from withdrawal fee (assuming $100,000 monthly withdrawals): $0.03 \times 100{,}000 = \$3{,}000/\text{month}$

$$R_C = 4{,}500 + 3{,}000 = \$7{,}500/\text{month}$$

Option D: Maker-Taker with Rebate

Maker rebate cost (40% of volume): $0.005 \times 620{,}000 \times 0.40 = \$1{,}240$ Taker revenue (60% of volume): $0.015 \times 620{,}000 \times 0.60 = \$5{,}580$

$$R_D = 5{,}580 - 1{,}240 = \$4{,}340/\text{month}$$

Option E: Hybrid

Taker revenue: $0.01 \times 600{,}000 \times 0.60 = \$3{,}600$ Resolution fee (winning contracts, ~50% of outstanding): $0.005 \times 600{,}000 \times 0.50 = \$1{,}500$

$$R_E = 3{,}600 + 1{,}500 = \$5{,}100/\text{month}$$

2.3 Revenue Summary

Critical finding: At initial volumes (500,000-620,000 contracts/month), none of the fee structures generate enough revenue to cover costs. This is typical for platform businesses — they require scale to become profitable. The question is which structure best supports growth to the required volume.

Part 3: Impact on Market Quality

3.1 Spread Analysis

Fee structures affect the equilibrium spread through two channels: 1. Direct cost to market makers: Fees they must pay reduce their profit, requiring wider spreads 2. Indirect incentive effects: Maker rebates encourage tighter quotes; high taker fees discourage aggressive taking

We model the equilibrium spread as:

$$S_{\text{eq}} = S_{\text{base}} + f_{\text{maker}} - f_{\text{rebate}} + \frac{f_{\text{resolution}}}{2}$$

where $S_{\text{base}} \approx 0.03$ (the spread due to adverse selection and inventory risk alone).

Option B: 2% of $0.50 = $0.01 per side for market maker. *Option C: No trading fee for market makers, but they face the profit fee on net positions.

3.2 Liquidity Provider Attractiveness

We score each option's attractiveness to market makers on a 1-10 scale:

Options A and D are most attractive to market makers, while Option C (profit fee) is least attractive due to the complexity of tracking profits and the asymmetric cost on winning positions.

Part 4: Trader Behavior Modeling

4.1 Trader Segmentation

ForecastX's user base consists of:

4.2 Impact by Segment

Casual bettors are relatively insensitive to fee structure. They trade for entertainment and are unlikely to optimize across platforms. All five options are acceptable to this segment.

Active traders care about per-trade costs and will compare platforms. They prefer transparent, low fees. Option D (maker-taker with rebate) is most attractive if they learn to use limit orders. Option C (profit fee) is least attractive because of the psychological pain of paying a large fee on wins.

Professional/quant traders will precisely calculate costs and only trade where the net edge after fees is positive. They strongly prefer: - Zero or negative maker fees (Options A, D, E) - Low taker fees for when they need immediacy - Predictable cost structures (not profit-based, which adds variance)

Market makers are the most critical segment. Without them, spreads will be wide, liquidity will be poor, and no other segment will trade. They strongly prefer Option D (rebate) and will avoid Option B (2% on their high-frequency trades would be devastating) and Option C (profit fee on net positions creates tracking nightmares).

4.3 Churn Risk Analysis

Part 5: Long-Term Scenario Modeling

5.1 Growth Trajectories

We model volume growth under each fee structure over 24 months:

Assumptions: - Organic growth rate: 10% per month (common for marketplace startups) - Growth modifier based on market quality (liquidity → attracts traders → more liquidity): - Excellent liquidity (Options A, D): 1.2x growth modifier - Good liquidity (Option E): 1.1x growth modifier - Moderate liquidity (Options B, C): 0.9x growth modifier

5.2 Revenue Trajectories

5.3 Breakeven Timeline

*Breakeven defined as monthly revenue exceeding $185,714.

5.4 Key Trade-Off

Options B and C generate more revenue per contract but suppress growth. Options A and D generate less revenue per contract but attract more volume, leading to faster absolute revenue growth.

Option B breaks even first (month 19) due to higher revenue per contract, but Option D generates the most revenue by month 24 due to its superior volume growth.

Part 6: The Recommended Fee Structure

6.1 Recommendation: Phased Approach

Phase 1 (Months 1-12): Option D — Maker-Taker with Rebate - Maker fee: -$0.005 (rebate) - Taker fee: $0.015 - Rationale: Maximize market maker attraction and liquidity quality. Accept lower initial revenue for faster growth.

Phase 2 (Months 13-24): Modified Option E — Hybrid - Maker fee: $0.00 (remove rebate as liquidity is self-sustaining) - Taker fee: $0.01 - Resolution fee: $0.005 (on winning contracts) - Rationale: Reduce subsidies to market makers once the flywheel is spinning. The resolution fee is less visible and adds revenue without significantly impacting trading decisions.

Phase 3 (Month 25+): Optimized Hybrid - Maker fee: $0.00 - Taker fee: $0.008 - Resolution fee: $0.005 - Volume discounts for top traders - Rationale: Reduce taker fee slightly to remain competitive. Revenue is sustained by high volume.

6.2 Revenue Projection Under Phased Approach

Total pre-profitability investment: approximately $3.2 million.

6.3 Sensitivity to Key Assumptions

Part 7: Additional Considerations

7.1 Dynamic Fee Adjustments

ForecastX should consider dynamic fees that adjust based on market conditions:

• Wider spreads → higher maker rebates: During periods of low liquidity, increase the maker rebate to attract quotes
• High volume → lower taker fees: Volume discounts for the most active traders
• New market launch → temporary zero fees: Waive all fees for the first 48 hours of a new market to bootstrap liquidity

7.2 Non-Fee Revenue Opportunities

• Data licensing: Sell prediction market data to media, researchers, and financial firms
• API access fees: Charge for high-frequency API access (free for basic, paid for premium)
• Premium features: Advanced charting, alerts, portfolio analytics
• Market creation fees: Charge for creating custom markets

7.3 Regulatory Considerations

Fee structures may face regulatory constraints: - CFTC-regulated platforms have specific fee disclosure requirements - Some jurisdictions may cap fees or require fee parity between participants - Maker rebates may face scrutiny for creating conflicts of interest (payment for order flow concerns)

Discussion Questions

1. The analysis assumes a price elasticity of -0.8. How would you empirically measure the true elasticity? What experiments could ForecastX run?

2. Option C (profit fee) has the lowest appeal but is the most capital-efficient for the platform (no revenue until contracts resolve). Under what conditions might this be the optimal choice?

3. How should ForecastX handle the chicken-and-egg problem of needing market makers before traders will come, but market makers only come if there are traders?

4. If a well-funded competitor (e.g., backed by a crypto exchange) enters with zero fees and is willing to operate at a loss indefinitely, how should ForecastX respond?

5. Design a fee structure that is progressive — lower effective fees for smaller traders and higher fees for large traders. What are the trade-offs?

Code Reference

See code/case-study-code.py for the complete Python implementation of the fee structure modeling, revenue projections, and scenario analysis.