Reward Model
Top-of-book eligibility, the fixed NVDA/USDG allocation, side clocks, and quantity-adjusted DEEP rewards.
Deepstate has one launch rewarder, attached to the NVDA/USDG pool. It pays only the canonical best bid and best ask, gives each side its own immutable budget and clock, and transfers rewards from a fixed DEEP balance funded at deployment.
The rewarder depends on the engine's sold-token hook input and authoritative topOrder view. Both components must use the same callback semantics and be tested as one integration.
Eligible position
An order accrues only while it is the canonical top of its side. Quantity away from the top does not earn. When a new order improves the price, a fill displaces the order, or cancellation removes it, the rewarder settles the outgoing top through that transition timestamp and starts the incoming top's interval.
Price priority remains absolute. Quantity changes the fraction of the scheduled ceiling that the top order earns; it does not allow an inferior price to earn.
Fixed launch allocation
| Pool | Bid maker sells | Ask maker sells | Per-side allocation | Pool allocation | Duration |
|---|---|---|---|---|---|
| NVDA/USDG | USDG | NVDA | 500,000,000 DEEP | 1,000,000,000 DEEP | 395 days |
The deployment mints the complete 1 billion DEEP allocation into the rewarder before transferring administration to governance. The rewarder has no MINTER_ROLE. Claims transfer from this prefunded balance, and unearned rewards remain locked in the rewarder after the schedules expire.
"Bid" and "ask" follow displayed market notation. Internally the engine sorts token addresses; integrators must map the pair before interpreting the engine's isBid flag.
Immutable configuration
The rewarder fixes these values at construction:
- Deepstate engine and DEEP token;
- NVDA/USDG pool identifier and sorted token pair;
- 500 million DEEP allocation for each side;
- 395-day duration for each side;
- starting and maximum full-reward quantity for USDG and NVDA; and
- logarithmic emission and exponential quantity-ramp parameters implied by those values.
Governance can disable or replace the pool hook, but it cannot rewrite this rewarder's pool, schedule, cap, duration, or quantity targets.
Independent side clocks
The bid and ask schedules start separately when Deepstate first reports a nonzero top order for that side. Deploying or prefunding the rewarder does not start either timer.
Once activated, a side's clock runs continuously with block.timestamp until its 395-day duration ends. An empty book does not pause or extend it. Scheduled capacity that is not earned expires; it does not move to the opposite side or a later interval.
Reward calculation
For an eligible interval, the rewarder:
- computes the side's scheduled reward ceiling between two timestamps;
- measures the top order in the token the maker is selling;
- compares that amount with the time-varying full-reward quantity;
- integrates the linear quantity factor over the interval; and
- caps aggregate accrued rewards at the side's 500 million DEEP allocation.
An order at or above the target earns the full interval ceiling. A smaller order earns proportionally less. Adding size above the target never exceeds the schedule.
Why top-only rewards
The mechanism pays for the state the protocol wants: the most competitive immediately executable bid and ask with meaningful collateral behind them. It does not pay for nominal TVL or an order waiting behind a better price.
The tradeoff is explicit. Makers can compete one tick at a time, small orders can chase priority, and deeper secondary liquidity is uncompensated. The quantity target limits rewards for a tiny winning quote, but it does not remove those strategic incentives.
See Emission Schedule, Full-Reward Quantity, and Claiming Rewards.