Introducing Azuro’s Liquidity Tree: A New Liquidity Pool Design for Specialized Prediction Markets

7 min readJun 4, 2022

General Prediction Markets are doomed

Prediction markets have been a part of blockchain since 2015 when Augur became Ethereum’s 1st ever ICO. Whilst holding so much promise for the space as one of the first crypto primitives, it wasn’t long before they were fraught with issues including limited markets, insufficient liquidity, and poor UX.

Following on, more projects, including Gnosis and Polymarket decided to enter the space, to try to solve the issues with prediction markets. Attempts were also made to bring sports betting to the decentralized space, but the issues remained and prediction markets proved unpopular with bettors. This was particularly disappointing given that sports betting is by far the biggest application for prediction markets today.

Gnosis decided to pivot away from prediction markets and built great products like crypto’s de-facto multi-sig solution — Gnosis Safe, and more recently launched GnosisChain.

Polymarket is still operating today, and decided to go down a less conventional route, with markets being created around politics and cryptocurrency, though yet extremely niche with a few tens of markets usually live. In an article from 2020 UncommonCore’s Hasu concludes that:

“If prediction markets want to survive, they need to realize that all markets need an overlay in the form of a) commercially viable use cases, or b) for a lack of a better word, fish. Most markets will never have either, as most people have not shown an inclination to bet on events other than sports, finance, and politics, and that is unlikely to change.

So they can either optimize for the markets that are proven to work, in which case they will discover that competition in these markets is strong and entrenched and optimized to offer the best experience for particular types of markets (e.g. sports betting.)”

Or, in other words:

  1. General prediction markets are doomed without incentives, which no one is willing to provide as of yet
  2. Specialized prediction markets (e.g. for sportsbetting) need to deliver a much better product to be able to compete with established industries like Online Betting

The problem with sportsbetting prediction markets

Sportsbetting is the biggest application area for PMs. But the problem of scaling prediction markets and making them fit for decentralized sportsbetting remained unresolved. Prediction markets offer peer-to-peer matching of Market Makers (create market, price market, and seed initial liquidity behind the market) and Bettors (trade on the market). This approach allocates liquidity inefficiently and is limited by simple YES/NO markets, plus a trading UX, instead of the popular UX in betting based on odds.

Curiously, many new decentralized betting projects today still employ Peer-to-Peer architecture even though its limitations from a sportsbetting perspective are obvious. Peer-to-Peer architecture has been used in sportsbetting for 20+ years (by betting exchanges), but remains a niche product, with just one notable betting exchange (Betfair) in a sea of bookmakers & sportsbooks (bet365, William Hill, DraftKings, FanDuel, 1xBet, and so on), which >99% of bettors worldwide flock to.

Solving the liquidity problem in prediction markets: from Peer-2-Peer towards Peer-2-Pool

Peer-to-Peer comprises two parties, a Market Maker, and a Bettor. Market makers are required to create the markets, price the markets and provide liquidity while taking 100% of the risk. Relying upon a single party to provide the liquidity inevitably leads to a scarcity in events and markets being put on-chain. Changing to Peer-to-Pool meant that the market maker’s role could be broken into 3 separate roles: Market Pricing (Odds Feed Providers), Market Creation (Oracles), and Liquidity Provision (Liquidity Providers).

Moreover, LP-ing is shared amongst many liquidity providers contributing to a common pool.

The idea goes as follows: Odds Feed Providers provide a feed with initial odds, Oracles import the priced markets on-chain, then these markets receive parts of the liquidity from the Liquidity Pool, where multiple LPs have put money.

However, existing Liquidity Pool designs couldn’t solve the problems associated with this model…

New Life for Specialized Prediction Markets: The Liquidity Tree

Azuro faced issues that required custom solutions.

Firstly — a Classic Liquidity Pool (as utilized in e.g. the AMM) holds asset X + asset Y. In this case — it is clear exactly what amount of asset X and asset Y there is in the pool at all times.

But Azuro’s liquidity pool couldn’t benefit from that clarity as the liquidity in the pool gets allocated to multiple betting markets, and until those are resolved — it is unclear what amount of $ they will return to the pool. Even though Azuro’s liquidity pool holds only 1 asset (as it is not a trading pool), say asset X — it is unclear exactly what amount of asset X can be utilized, as asset X is used to seed liquidity into multiple betting markets many of which are yet unresolved.

Secondly, the exposure to outside events (e.g. real-world sporting events) means that LPs would have opportunities to add liquidity to the pool exactly when profits are to be realized (i.e. an event resolving in a certain way) thus cheating the other LPs out of potential gains.

Imagine if each swap on Uniswap took say 10 blocks, and LPs can put money after the swap has started — it would mean some LPs could cheat and realize gains without any impermanent loss!

These were tough problems to solve which required a proprietary solution. A new liquidity pool design, which we called the Liquidity Tree (or LT).

The Liquidity Tree name is inspired by a well-known data structure in Computer Science known as Segment Tree, which we utilize for the data in the new liquidity pool design.

Azuro aims to allow any LPs to provide liquidity to the pool and get exposed to all betting markets happening on the protocol, thus reducing the risk for LPs dramatically and leading to a consistent APY. All the while utilizing all that money efficiently for seeding initial liquidity behind thousands of betting markets.

To achieve that however — there is immense complexity arising from the fact that LPs can add or withdraw liquidity at any point while betting markets receive parts of the overall pooled liquidity at the same time. Solving this by following where each single LPs’ liquidity goes would be a heavy computational burden (including timestamps, tracking, etc.) and therefore not suitable for a blockchain-powered solution.

Тhis is where the Segment Tree data structure comes into place allowing for an extremely significant increase in efficiency, making the protocol lean, while at the same time solving the issues with addition and withdrawal of liquidity

The result is a unique peer-to-pool design: the Liquidity Tree pool. It allows for:

1. Thousands of liquid betting markets with competitive pricing (odds)

2. Rich feature set including critical bet types like accumulators/parlays (many bets in the same slip) and multiple outcome bets (e.g. Which player (out of unlimited amount of players) will win tournament X?)

3. Odds-based UX which shows potential gains from the bets clearly

The above approach puts a blockchain-based PM extremely close to delivering the experience a traditional sportsbook/bookmaker does.

The Liquidity Tree is a new piece of open-source code that Azuro hopes will become a popular household name similar to the AMM and will fuel the birth of specialized prediction markets.

Bug Bounty

Delivering bleeding-edge smart-contract innovation is not easy. Azuro’s Liquidity Tree pool has undergone several audits, including enlisting vital help from partners and investors from Gnosis whose experience is second to none (it is a little-known fact that the Gnosis team partook in the creation of one of crypto’s recent biggest innovations. As acknowledged by Uniswap’s Hayden Adams — the initial steps towards the AMM (which fuelled DeFi summer) were made by the Gnosis team who experimented with the new design first).

Still — any help and additional eyes over is appreciated! That is why we also have an open Bug Bounty Program with ImmuneFi.

If you want to contribute to making the Liquidity Tree pool safe & sound for open use — make sure to check out the campaign!

What’s next?

Azuro’s Mainnet Launch is expected soon! The new design means that we can not only provide a solution for prediction markets but also bring a sportsbook experience to decentralized betting. The protocol will soon be live on mainnet and ready for the first betting dApps to be built on Azuro.

Follow the links below to join Azuro’s communities and stay up to date with our progress.

Twitter | Discord | Telegram | Website




🌊 Dive into the Azure: Empowering decentralized sports markets & games with deep liquidity & prediction contracts. 🌊