💦Decoding Prop AMMs: Humidifi
What Is a Prop AMM?
In the world of Solana trading, a brand-new market-making model has emerged: the Proprietary AMM (Prop AMM). This model is fundamentally different from traditional, publicly funded liquidity pools:
The market-making logic is embedded directly inside on-chain programs and controlled by institutions or professional teams, rather than relying on external LP capital.
Pricing is not derived from a fixed formula, but from actively updated strategies designed to closely track centralized-exchange quotes.
Leveraging Solana’s high throughput and low latency, Prop AMMs minimize delay and slippage by updating internal prices at extremely high frequency, using highly optimized instructions and very low fees. Prices are computed internally by contract algorithms, allowing these systems to rapidly dominate volume across popular trading pairs.
In other words, a Prop AMM is not a “passive liquidity pool” in the traditional sense—it’s an on-chain trading engine where professional market makers deploy automated algorithms, private strategies, and their own capital directly on-chain.
Humidifi
Humidifi is a textbook example of this approach. At its peak, it accounted for nearly half of Solana’s total trading volume.This article breaks down Humidifi’s internal logic to show how it actually works.
XOR State Obfuscation
As a Prop AMM, Humidifi has no interest in making its data structures or call patterns easy to reverse-engineer. The protocol hardcodes a set of 64-bit XOR keys directly into the program.The real state is never stored in plain form—every value is bitwise XOR-encoded.
Q48 Extreme Precision
At the core of any AMM lies math—but on Solana, math has hard constraints:
Floating point ❌ (non-deterministic)
Standard integers ❌ (insufficient precision)
Humidifi’s solution: build a custom math system.
48 bits are reserved for the fractional part.
Precision is roughly 1 / 2⁴⁸, about 14 decimal places.
Notably, Humidifi does not rely directly on u128, ensuring that in tiny, high-frequency trades, the curve does not accumulate “dust-level” errors.
No Formula: Table-Driven Market Making
This is the most fundamental difference between Humidifi and systems like Raydium or Meteora DLMM.Traditional AMMs
A formula defines the curve.
All prices are derived from that formula.
Humidifi
No formula.
Only tables.
After decoding the pool state, you’ll find three arrays:
Table A (18 integers)
Table B (30 integers)
Table C (10 integers)
These are not parameters—they are the skeleton of the entire pricing curve.How does pricing work?Instead of solving equations, the core logic:
Determines which interval the trade size falls into.
Performs linear interpolation between two points to compute the price.
In short, Humidifi’s curve is “connect-the-dots.”What does this enable?
Stablecoin-like curves
Concentrated liquidity profiles
Arbitrary curve reshaping at any time
Curve updates without upgrading the contract—just modify the tables
Built-In MEV Defense
As a Prop AMM, Humidifi takes a hard stance on MEV—especially atomic arbitrage.During execution, the contract scans other instructions in the same transaction, searching for specific byte patterns.If certain conditions are met—such as calls from non-whitelisted programs—the final output may include additional fees or penalties.
Summary
Humidifi makes its philosophy crystal clear:❌ Not optimized for transparency✅ Built for on-chain adversarial environmentsIt’s not a pool—it’s a tightly sealed financial black box:
XOR-locked state
Q48 extreme precision
Lookup tables instead of formulas
Execution-layer MEV defenses
If you truly understand Humidifi, you’ve already stepped into the real Prop AMM arena.In the next pieces, I’ll break down other protocols in the same category.What lies beneath the surface is far deeper than anything you see in the UI.
Last updated