Introduction

The expected return for yield opportunities in DeFi have been on a steady downtrend with the decline in AMM dominance and the amount of hacks that have been happening in DeFi. Now spot volumes route through many RFQs or offchain makers compressing yield.

For reference, DEXs on DeFi LLama had 80B in TVL end of 2021 which was 32% of total DeFI TVL and now DEXs have 13.5B in TVL which is 4.8% of total DeFI TVL. That's a 27% drop in dominance.

Additionally, users are finding it more difficult to overcome the risk of providing liquidity into DeFi protocols. There was nearly 1B hacked in just the past two quarters in DeFi - $320M hacked from Liquid, ~292 from Kelp DAO and ~300M from Drift Protocol.

Are there yield strategies actually worth the squeeze in 2026? Yes. The yield is market making on prediction markets.

Market Making Order Books on Prediction Markets

The top prediction market makers on Polymarket are making insane profits.

Account: X-MoneyForWhiskas

• Profile: https://polymarket.com/@x-moneyforwhiskas?tab=activity
• PnL: +$424k since Mid June
• Total Position Value Snapshot: $521.18
• Avg Trade Value: ~$14.92
• Median Trade Value: ~$11.50

Account: BoneReaper

• Profile: https://polymarket.com/@bonereaper
• PnL: +$1.4M since end of March
• Total Position Value Snapshot: $1,514.97
• Avg Trade Value: ~31.20
• Median trade value: ~$12.69

Account: OhioRiskManagement

• Profile: https://polymarket.com/@ohioriskmanagement
• PnL: +$262k since January
• Total Position Value Snapshot: $1,092.71
• Avg Trade Value: ~$10.68
• Median trade value: ~$4.18

Account: FairValueMachine

• Profile: https://polymarket.com/@fairvaluemachine
• PnL: $47k since mid July
• Total Position Value Snapshot: $25.51
• Avg Trade Value: ~$17.33
• Median trade value: ~$16.24

The average trade value across accounts was $18.53 and avg total position value was $788.59. Average and median trade value were taken from last 1000 trades, not since first trade.

The majority of trading across accounts was in the BTC, ETH, SOL 5 minute markets. The reason they can earn such high PnLs is due to how capital efficient these binary option contracts are. Every 5 minutes they can redeem and recycle their capital.

Take OhioRiskManagement. The average monthly profit was $25,000 with about $1,000 in position value at risk at any given time. That's a monthly return of 2,500% or annualized 30,000% without compounding.

Millions of dollars are on the table for market makers if they find the right strategy. There is little to no capital requirements to start. The yield is higher than anything you can find in DeFi, comparable to LPing in AMMs during DeFi Summer.

Market making on order book exchanges used to be impossible for the average person. Now, anyone can use Claude to create a bot to test and then execute a trading strategy.

Anyone can become a market maker now and since we plan to launch markets tracking unique assets there will be opportunities to start without the competition you find on incumbent exchanges.

Market Making Bot

To help our community members who are interested in getting started with market making, we have open sourced our market making bot: https://github.com/gammaswap/v2-tradingbot

This bot was built to trade on GammaSwap V2, the crypto prediction market. About GammaSwap V2 here: https://docs.gammaswap.com/

It can run as either a maker or a taker, but not both in one instance. It can be repurposed for other prediction market platforms like Kalshi which is also a margin exchange. Polymarket and HIP-4 use conditional tokens so it would require more refactoring.

How Maker Mode Works

Maker mode is the passive market-making bot. It connects to GammaSwap exchange API + orderbook websocket.

Maker Mode Tracks:

• current market/asset/epoch
• orderbook bid/ask
• account balance
• current inventory

It computes a reference price from the oracle-derived fair value, and/or current orderbook midpoint.

Then it adjusts that reference price for inventory. If long inventory, quote lower to encourage selling/reducing. If short inventory: quote higher to encourage buying/reducing.

It then creates a bid/ask ladder:

• Multiple passive levels on both sides.
• Orders are ALO / add-liquidity-only, so they should not cross the spread.
• Ladder can be equidistant: levels spread evenly between calculated quote and book/reference range
• Levels expand farther away using exponential spacing

It continually cancel/replaces stale quotes when fair value changes or price on book changes.

How Taker Mode Works

Taker mode is not passive. There is no quote ladder it just attempts to take liquidity when there is an edge.

Supported models:

• edge: buy if fair value is above best ask; sell if fair value is below best bid.
• mean-reversion: buy more when price is below a midpoint, sell more when above.
• edge-with-fallback: uses edge model first, falls back to mean reversion.
• It

It sizes IOC orders based on how many book levels to wipe, a slippage buffer, and max aggression quantity.

How to Run

To run, npm install the GammaSwap SDK and the V2 trading bot.

V2 SDK: https://github.com/gammaswap/v2-exchange-sdk

npm install @gammaswap/v2-exchange-sdk

V2 Trading Bot: https://github.com/gammaswap/v2-tradingbot

npm install @gammaswap/v2-tradingbot

Then follow the rest of the instructions in the V2 trading bot readme on github.

In Conclusion

You can make thousands, even hundreds of thousands of dollars market making on prediction markets with average trade sizes of $50. That's how capital efficient the 5 minute binary crypto markets are.

We open sourced our market making bot to help you get started.