RADIAL

§ motivation · why radial exists

150 days. 6 protocols.
150.8M SOL idle every day.

Every chart on this page is computed from indexed swap data across Meteora and Raydium AMMs — 150 days of historical activity aggregated into daily metrics.

act 01·the problem

150M SOL idle every day.

Across the six largest Solana AMMs, a daily floor of ~150M SOL sits idle — every day, for 150 days running. Pool reserves earning zero yield while users wait.

01 · Daily minimum SOL per protocol

Daily minimum SOL across six protocols. Min = the absolute floor — capital that never moved on any block in the day.

Dec 11Jan 9Feb 7Mar 8Apr 6May 5May 950M100M150M200M250M

On-chain analysis · 150-day window · 6 protocols

  • DLMMmeteora
  • CLMMraydium
  • DAMM v2meteora
  • AMM v4raydium
  • DBCmeteora
  • CPMMraydium

03 · Mean daily minimum, by protocol

Meteora DLMM alone parks 61.8M SOL on average in pool reserves earning nothing.

DLMM62MCLMM60MDAMM v212MAMM v410MDBCCPMM10M20M30M40M50M60M

On-chain analysis · 150-day window · 6 protocols

act 02·the opportunity

4.9M–10M SOL/year at stake.

At blended 6.17% APY (mostly native staking + Marinade), the six protocols would generate 9.30M SOL per year. Conservative lending alone delivers 4.90M SOL. LST-heavy strategies push toward 10.2M SOL.

09 · Annual yield by protocol & strategy

Annual yield per protocol across three APY scenarios. SOL-denominated.

DLMMCLMMDAMM v2AMM v4DBCCPMM1.0M2.0M3.0M4.0M

On-chain analysis · 150-day window · 6 protocols

10 · Cumulative yield over the window

Compounding over the 150-day window. Even conservative deployment alone surpasses 1.2M SOL.

Dec 11Jan 10Feb 9Mar 10Apr 9May 91.0M2.0M3.0M4.0M

On-chain analysis · 150-day window · 6 protocols

  • conservative · 3.25%
  • blended RMM · 6.17%
  • lst-heavy · 6.75%

act 03·the risk

Free money would already be taken.

If idle SOL could be deployed naively, someone would have. The catch: pool drains. A 3-day worst case on DLMM hit 28%. The math has to clear that, with margin.

16 · Observed worst vs σ-buffer models

Worst observed 3-day drop vs the 2σ and 3σ buffer models (each scaled by √3 for the 3-day horizon). Buffer must cover historical worst.

DLMMCLMMDAMM v2AMM v4DBCCPMM10%20%30%40%

On-chain analysis · 150-day window · 6 protocols

17 · Minimum K-σ to cover worst case

Minimum K-sigma to cover every protocol's worst observed drawdown. Green = K ≤ 2 (safe), yellow = K ≤ 3 (margin), red = K > 3 (insufficient).

DAMM v2K=5.38AMM v4K=5.05CLMMK=2.88CPMMK=2.27DLMMK=2.14DBCK=1.00K=2.00K=3.00K=4.00K=5.00

On-chain analysis · 150-day window · 6 protocols

14 · Annual yield across five buffer strategies

Annual yield per protocol across five strategies. Even the conservative 3σ buffer retains 65–88% of the perfect-oracle yield.

DLMMCLMMDAMM v2AMM v4DBCCPMM1.0M2.0M3.0M4.0M

On-chain analysis · 150-day window · 6 protocols

  • no buffer
  • naive
  • perfect oracle
  • 2σ buffer
  • 3σ buffer
observed worst 3-day
2σ buffer (·√3)
3σ buffer (·√3)

act 04·the forecast

Forecast, then deploy.

Walk-forward ML backtest, 5 blocks × 7 days. The Tight model trims over-reserve by ~40% vs the Safe baseline at the same shortfall tolerance. Every protocol gets a strategy that strictly dominates naive 3σ.

19 · Forecast vs actual · 3×2 small multiples

Actual min(D+2) vs Tight and Safe model predictions per protocol. Six small multiples, shared X axis.

Meteora DAMM v25.0M10M15MMeteora DLMM50M100MMeteora Dynamic Bonding Curve2.0M4.0MRaydium AMM v45.0M10MJan 3Feb 4Mar 30May 1Raydium CLMM20M40M60M80MJan 3Feb 4Mar 30May 1Raydium CPMM2.0M4.0MJan 3Feb 4Mar 30May 1

On-chain analysis · 150-day window · 6 protocols

  • actual
  • tight model
  • safe model

21 · Forecast trade-off · shortfall vs over-reserve

Pareto frontier: shortfall % (x) vs over-reserve SOL (y). Each protocol has a tight–safe pair (filled = tight, outlined = safe) connected by a dotted line in its color.

DAMM v2DLMMDBCAMM v4CLMMCPMM10.0%20.0%30.0%40.0%Shortfall days (%)2.0M4.0M6.0M8.0MMean over-reserve (SOL)

On-chain analysis · 150-day window · 6 protocols

§ next · into the engine

Now you've seen the gap.
Help us close it.