Skip to main content

Whitepaper & research

These are the canonical research artifacts the protocol is built on. They are kept under papers/ in the protocol's source repository (currently private). This page summarizes what each is; we do not reproduce derivations here.

The Liquity whitepaper​

Liquity Whitepaper rev. 0.3: papers/whitepaper/Liquity Whitepaper rev. 0.3.tex.

The foundation. Defines the trove model, the Stability Pool, redistribution, and the redemption-as-floor mechanism that RAI Dollar builds on. Everything in Borrowing, Liquidation paths, and the base portion of Redemption mechanism traces back to this paper.

If you want a primer on:

  • ICR / TCR / MCR / CCR;
  • the offset-vs-redistribution liquidation choice;
  • why the Stability Pool's depositors get collateral at a discount;
  • why redemption creates a price floor and what the arbitrage looks like.

Read the relevant section of this paper.

Efficient order-preserving redistribution of troves​

papers/Efficient_Order-Preserving_Redistribution_of_Troves.pdf (LaTeX source in same directory).

Describes the algorithm Rewards.sol uses to track per-trove pending redistribution rewards in O(1) per trove. The key trick is a small set of accumulators (L_Coll for collateral, L_baseNormRDDebt / L_shieldNormRDDebt for debt) plus per-trove snapshots, so settlement is a single multiplication. This is what makes redistribution scalable in a multi-thousand-trove branch.

Read this if you're auditing Rewards.sol, working on a subgraph that needs to mirror trove state, or trying to understand why redistribution doesn't cost gas proportional to active-trove count.

Scalable reward distribution with compounding stakes​

papers/Scalable_Reward_Distribution_with_Compounding_Stakes.pdf (LaTeX source in same directory).

The math behind reward distribution in the Stability Pool. Same conceptual pattern, accumulator + snapshot, but extended to handle stake balances that change as rewards accrue (SP deposits shrinking through liquidation). MOAIStaking and LPStaking use the simpler fixed-stake form of the same pattern (running sums F_RD and F_MOAI per unit staked).

Read this if you're auditing StabilityPool.sol, MOAIStaking.sol, or LPStaking.sol, or implementing an off-chain reward calculator.

Debt and supply rounding proofs​

Four Markdown documents in papers/ prove that rounding never lets recorded debt exceed RD supply:

FileWhat it covers
System_Supply_Debt_Conservation.mdThe umbrella proof that the system invariant systemDebt <= supply holds across every operation that moves debt or supply, with the rounding direction of each.
Supply_Debt_Rounding_Cheatsheet.mdA one-page reference of the same rounding map.
Shield_Toggle_Migration_Rounding_Closed_Form.mdThe closed-form rounding guard that ensures moving a trove between the base and Redemption Shield debt cohorts never raises branch debt, in O(1).
Redistribution_Branch_Debt_Conservation.mdThe matching O(1) guard for liquidation redistribution across the two cohorts.

Read these if you're auditing debt accounting, Redemption Shield toggles or redistribution.

Scenario PDFs​

Four public scenario walkthroughs of TCR-drop attack/stress paths, all in papers/:

FileWhat it covers
SCENARIO 1_ TCR drop via price sandwich + opening huge Trove (public).pdfA specific manipulation pattern using a price sandwich plus a large new trove to drive TCR down.
SCENARIO 2_ TCR drop via underwater debt + redemption + redistribution (public).pdfTCR drop via the interaction of underwater debt, redemption, and redistribution.
SCENARIO 3_ TCR drop via underwater debt + opening huge Trove + redistribution (public).pdfSame family of stresses with a different ordering.
SCENARIO 4_ TCR drop via underwater debt + opening huge Trove + redemption (public).pdfUnderwater debt plus opening a huge trove plus redemption pressure.

These are useful if you're modeling how the system responds to adversarial activity, designing a keeper bot that should react to these patterns, or auditing recovery-mode and shutdown logic.

How this site treats the papers​

  • We point at them whenever the topic shows up.
  • We summarize implications in plain language so users don't need to read the math.
  • We do not reproduce proofs or simulation walks here. If you want them, read the paper.

For the operating parameters of the deployed protocol (controller gains, deadbands, dwell times, etc.) the source of truth is the contract code, not the papers. The papers establish the math and motivate the parameter ranges; the code commits to specific values inside those ranges. See Risks and the contract-level pages for current values.