TL;DR

Initialization

Data Provider creates Pool request to Factory (opens in a new tab) (owned by AzuroDAO). If approved, Factory deploys a Pool (opens in a new tab) for the Data Provider and gives access to Azuro LP (and etc.) via Access (opens in a new tab).

This allows the Data Provider to create new prediction markets, via a set of Conditions (opens in a new tab) making up each prediction market. For example (for a football match): FullTimeResult, BothTeamsToScore, etc.

Data Provider sets Reinforcement (opens in a new tab) for each Condition. Reinforcement represents the initial liquidity seeded from the Azuro LP to serve that Condition (e.g., $200k), distributed across multiple outcomes (e.g., $30k on Team 1, $70k on X, $100k on Team 2). This will also set the *Initial Odds for each outcome within a Condition ($200k/$30k = 6.67 odds for Team 1, etc.).

Data Provider sets margin (i.e., 2.5%) on each Condition. Displayed odds to users will be post-margin, not the 'true' odds derived from Reinforcement seeding. Post-margin odds will be automatically applied onchain, based on 'true' Initial Odds (and Live Odds) of an outcome.

Active markets

User bets will add to the Condition’s initial liquidity (aka Reinforcement). If the Condition’s initial liquidity is $200k and it sees $50k in bets, its total seeded liquidity becomes $250k.

Live Odds will deviate from Initial Odds depending on betting flows, divided by the Condition’s total seeded liquidity at time of bet. The more flows to one side of the Condition, the lower the outcome’s odds will be (e.g., if this $50k flows entirely to Team 1, its Live Odds will become $250k/$80k = 3.125, massively down from its Initial Odds of 6.667).

The above bet entails a potential loss of $106.25k ($50k * 3.125 - $50k) to the Condition, hence the bet will still be accepted as $106.25k is still below the Condition’s $200k Reinforcement.

As bettors won’t bet on just one side of a Condition, on aggregate it’s unlikely for potential loss to exceed Reinforcement. However, if more bets are disproportionately placed on one side and push the Condition’s potential loss to exceed Reinforcement, the bets will be rejected.

Hence, apart from seeding initial liquidity (and setting Initial Odds), Reinforcement also caps the amount of max loss that could be incurred to the Condition assuming the worst-case scenario (borne out of bad oddsmaking by the data provider, or unforeseen changes relating to the event post market creation that severely tilts the probability of one outcome happening over the other).

Singleton LP

As all active Conditions share the same singleton LP, each active Condition needs to ensure there’s sufficient liquidity in the LP prior to accepting more bets. The amount denoting the additional liquidity (in excess of Reinforcement) that needs to be booked from the LP for an active Condition before accepting further bets is referred to as the Condition’s Virtual Fund (opens in a new tab) (aka “float”).

Initially, Conditions 'book' liquidity from the LP up to its Reinforcement amount (as it represents the Condition’s max potential loss). Assuming $5M LP TVL (and implied $200k Reinforcement on this active Condition), Azuro LP’s available capacity now stands at $4.8M.

Let’s say this Condition with $200k Reinforcement accumulates $50k total bets with a $100k max loss. In this case, the Condition’s Virtual Fund is $250k ($200k - ($50k - $100k)), which consumes LP capacity from $4.8M ($5M - $200k) to $4.75M ($5M - $250k).

Once the LP’s capacity runs out, it is no longer able to service further bets, that is until capacity frees up (from settled Conditions, LP profits from previously settled bets, new LP deposits, etc.).

Event resolution and rewards distribution

The Data Provider resolves the set of Conditions making up a prediction market, depending on its associated real-world results. Where necessary, AzuroDAO may act as the arbiter of last resort.

Each Condition’s booked liquidity (Reinforcement + Virtual Fund) that is part of the resolved market is immediately released back to the LP, which replenishes its capacity.

Rewards (opens in a new tab) on each Condition is calculated, stored on an onchain vault, then disbursed to relevant protocol actors (i.e., apps, LP, Data Provider, AzuroDAO) on a monthly basis.