Polymarket suck, so i decided to make my own. [Part 1]

In this series, we will follow my learning process on creating an exchange from A to Z, and i will explain how i will do my own prediction market. Maybe a build in public?

Well, if you’re not living in a cave for the past few years, you probably know that prediction markets are growing. Even Coinbase and Robinhood are launching their own prediction markets. And guess how I knew it? I was sniping the mention market in both the Robinhood and Coinbase announcements when they talked about it, and now it’s on their website. We are definitely in the matrix.

So if there are already leaders in this industry like Polymarket and Kalshi, why would I make another one? Honestly, the probability of getting people on it is close to zero, I’ll be honest. I just want to learn, and maybe it will grow somehow. My main motivation, besides learning, is that Polymarket has constant issues, constant crashes, a weird and slow API, and a lot of things that I feel are missing. I’d like to build something that I would personally want to use, with a smooth and fast API and UX, and good features. And they also killed their up/down market with fees and speed bumps.

---

The big deal is liquidity. I think having more prediction markets is healthy for the ecosystem. It creates arbitrage opportunities, brings competition, and generates more flow for market makers to hedge on.

Providing liquidity in prediction markets is fundamentally more complex than in traditional markets. These markets are designed around extreme leverage, binary outcomes, and asymmetric risk, which makes adverse selection much more severe.

As in traditional finance, traders profit from informational or technological edges, but prediction markets push this dynamic to the extreme. Prices move almost exclusively when new information is incorporated, leaving very little room for noise. While traditional markets also absorb information at lightning speed, they still contain a large amount of randomness, which creates space for liquidity providers to earn spreads and people to gamble / extract alpha.

Prediction markets, by contrast, are inherently non-standardized. Each market is unique: different resolution mechanisms, different data sources, different models. This fragmentation creates abundant arbitrage opportunities and explains why AI-driven systems are particularly effective here.

However, this also creates a vicious circle. Information asymmetry is so high that two bots consuming different data feeds can generate signals 30 seconds apart. In such an environment, a market maker can be picked off almost instantly and incur significant losses.

As a result, prediction markets are an extremely toxic environment for liquidity provision, far more so than traditional finance which naturally discourages market makers from participating, it’s why polymarket spend so much on rewards and advertisement.

---

So, what would i bring “new” to the echosystem, here is few ideas i had:

• RPI orders, matching against flow with a huge speedbump and UX trader

• Tighter tick sizes (0.1% compared to 1%).

• Earn passive yield on your account balance (Ethena, YUSD etc..)

• Trade market against other pairs than USDC (e.g BTC, ETH…)

• “Semi” Pro-rata matching engine

• Toxicity based fees structure

• New orders type

• Privacy

• protocol and community vault

• RFQ

• More crypto markets

So, let me introduce to you those ideas and why those ideas. Also note that i am not a EXPERT of prediction market and that i am opened to any suggestion and correction, i am very open about bringing new things to the ecosystem, and if your feature is good, it have lot of chance to be added.

1. So, first of all, as I said above, toxicity is a huge issue on these markets. RPI orders allow makers to match only with non-toxic flow, like UX traders, long-term traders, or trades sent for hedging purposes. The main goal of RPI taker orders would be a significant speed bump, from 500ms up to 1+ second depending on the market type, and much slower rate limits. This would incentivize makers to provide liquidity to less informed flow.

2. Second is tighter tick size. This drastically reduces the fight around queue priority, instead incentivizing makers into a penny-jump competition, which ends up reducing spreads and therefore costs for end users. The current industry standard is around 1% (0.1 for markets near 0). Each buy-sell round trip costs you about 1% of your capital at the tightest spread.

3. Capital efficiency is also very important. Sometimes, the cost of taking an opportunity is simply not profitable because the resolution time is too far away. This is why certain markets, like “Jesus will return in 2025”, trade around 3%. People who do not believe in this are willing to accept a return close to the risk-free rate; they won’t lock capital for a return below that, since it’s irrational. Allowing traders to earn extra yield on their balance lets them arbitrage more markets, making the system more efficient. It also incentivizes more liquidity to stay on the platform rather than locking them into some staking protocol.

4. Should I explain this one?

5. A pro-rata matching engine would incentivize people to post more liquidity, but it also incentivizes spoofing and bad behavior, and does not encourage tighter spreads. My idea would be to have a semi pro-rata and FIFO matching engine, which takes both queue position and provided liquidity into account. This keeps competition on queue priority while still incentivizing larger liquidity provision. Only orders within a certain delta from the mid-price would be eligible.

6. Toxicity-based fees and rebates would be based on the average markout of a taker. This would push toxic taker flow toward other venues, while attracting non-toxic flow to the platform. Traders could take liquidity on another platform and hedge themselves on mine; those hedge orders would be non-toxic and benefit from fee reductions. All this non-toxic flow would also incentivize makers to tighten spreads and provide more liquidity, as they naturally compete to capture this noise flow. The fee structure could be around 3 to 5 bps (roughly half a tick), with 50% rebated to makers, 40% earnable by non-toxic takers, and the platform taking the rest. Meanwhile, Polymarket takes up to 80% of fees that can reach 1.5%.

7. Information privacy is crucial. Iceberg orders are heavily needed. TWAP as well. I’m not sure hidden orders are a good idea, but they could be a variant of iceberg orders. Pegged orders would help reduce the constant cancel-replace behavior of market makers and also preserve some queue priority. Polymarket also lacks reduce-only orders and a “cancel on disconnect” (dead man switch) flag.

8. Privacy is one of the biggest issues. Polymarket, by design, exposes every user’s trades and positions with no delay. This leads to backrunning, and sometimes even frontrunning if you can reverse the signal or know who is backrunning you. It also massively leaks information. Prediction markets are an information game, and no serious entity wants to reveal what they are doing. That said, users could choose to make their profile public, purely for flexing purposes.

9. Protocol and community vaults. People love to follow their guru or deploy strategies with profit sharing, so we can enable that. The protocol vault is more ambitious. It would provide fallback liquidity on the platform by hedging on Polymarket, basically XEMM or some variation of what Vest is doing. This would let users generate extra yield from cross-exchange market making while simultaneously providing liquidity to the platform.

10. This one is going with the other above, maybe instead of quoting our orderbook, we can provide RFQ to api traders and for retails, i have seen someone complaining there is no “1 click trading” on polymarket, we can propose this trough a RFQ, while reducing our exposition to adverse selection. Other market makers could allow to compete on the RFQ by exposing their own quotes too. this will incentive into competition and tighter spread/cost for the end users) (+maybe it will generate a pvp against protocol vault, so maybe non api trader only have the protocol RFQ, and api one have the MM RFQ? idk, let me know what’s best.)

11. Finally, up/down markets for a large list of crypto and tradfi symbols, with more timeframes: 1m, 3m, 5m, 15m, 30m, 45m, 1h, 2h, 3h, 4h, 6h, 8h, and so on.

---

This is everything I had to say for the introduction of my ideas and this series. In the next articles, I’ll probably talk about more technical things, like how I’m designing the matching engine, the infrastructure, and so on. What I learn along the way, etc.

If this project sound cool, your free to join the dedicated discord server: https://discord.gg/D8CjJvmPc2

Thanks for reading me! (Follow and join the discord or u will have to deal with OUPI GOUPI)