Skip to main content
Volatility is the rate at which an asset’s price changes over time. In a lending protocol, volatility directly determines the risk of a healthy loan becoming liquidatable — and how quickly that transition can happen. A loan at 75% LTV with SOL collateral and USDC debt will be liquidated if the SOL price drops enough to push the LTV above the liquidation threshold (e.g., 85%). How likely is that 10% move? How quickly could it happen? These questions are answered by volatility analysis.

​
Why Volatility Matters for LTV Calibration

The gap between a token’s Max LTV and its Liquidation LTV is the safety buffer. This buffer must be wide enough to account for how far the collateral price can drop — and how fast — before liquidators execute. If the buffer is too narrow, prices can gap through the liquidation threshold, leaving positions deeply underwater before liquidators can act. Consider two tokens:
  • Token A has annualized volatility of 40%. On an average day, its price moves 2-3%. A 10% safety buffer provides several days of runway before liquidation.
  • Token B has annualized volatility of 150%. On a volatile day, its price can move 10-15%. A 10% safety buffer could be consumed in hours.
Token B requires a wider safety buffer — which means a lower Max LTV. Setting the same LTV parameters for both tokens would expose lenders to unacceptable risk on Token B.

​
Parkinson’s Volatility Measure

Kamino uses Parkinson’s Volatility rather than simple close-to-close volatility. Standard volatility measures — like the standard deviation of daily returns — only capture the price at two points (open and close). If a token’s price drops 20% intraday and then recovers, standard volatility might show a calm day. But that 20% intraday drop could have triggered liquidations. Parkinson’s measure captures the intra-period range — the highest and lowest prices within each time window: 
σ = √( (1/4N·ln2) · Σᵢ [ln(Hᵢ/Lᵢ)]² )
Where:
  • Hᵢ = highest price in period i
  • Lᵢ = lowest price in period i
  • N = number of periods
  • ln = natural logarithm
By using the high-low range rather than open-close returns, Parkinson’s measure captures volatility that intraday-recovering price movements would otherwise hide. This is particularly important for cryptocurrency markets, which are:
  • 24/7: No market close means there are no overnight gaps. Parkinson’s measure works especially well when there are no gaps between periods, because the high-low range fully captures the price action.
  • Flash-crash prone: Brief but severe price dislocations occur regularly on crypto markets. These are invisible to close-to-close measures but captured by Parkinson’s.
  • Liquidation-relevant: What matters for liquidation risk is the maximum intraday drawdown, not whether the price recovered by close. Parkinson’s measures exactly this.

​
Calculation Methodology

Volatility is calculated using hourly price data across multiple time windows:
WindowPurpose
Short-term (7 days)Captures current market regime — is the asset in a volatile or calm period?
Medium-term (30 days)Smooths out short-term noise while still reflecting recent conditions
Long-term (90+ days)Establishes the baseline volatility regime for the asset
All three windows are considered together. A token might show low short-term volatility during a calm market, but its long-term window reveals periodic spikes. LTV calibration must account for the spikes, not just the calm.

​
Monitoring and Response

Volatility is not static — it clusters. Calm periods are followed by volatile periods, and volatile periods tend to persist. When volatility shifts, protocol parameters must respond:

​
Gradual Volatility Increase

If an asset’s rolling volatility increases over days or weeks — perhaps due to shifting market sentiment, approaching token unlocks, or increased speculation — the Risk Council may proactively lower Max LTV or reduce supply caps before the volatility manifests as liquidation events.

​
Sudden Volatility Spike

Abrupt onset of extreme volatility — such as the February 2026 event when SOL dropped 18% in 48 hours — tests whether existing parameters were set conservatively enough. In that event, Kamino’s LTV parameters held: 55,649 liquidations executed profitably with $0 bad debt. Post-spike, the Risk Council reviews whether the event was within expected parameters or whether recalibration is needed.

​
Volatility Compression

When an asset’s volatility compresses over an extended period, there may be room to increase Max LTV — giving borrowers more capital efficiency. This is done cautiously, with the understanding that low-volatility periods often precede high-volatility events.

​
Volatility Across Asset Classes

Different asset classes on Kamino exhibit structurally different volatility profiles:
Asset ClassTypical Volatility ProfileLTV Implication
USD Stablecoins (USDC, USDT)Very low (~1-2% annualized)Highest LTV, E-Mode eligible for stablecoin pairs
SOLModerate-high (60-100% annualized)Standard LTV (70-80%), primary collateral asset
SOL LSTs (JitoSOL, mSOL)Low relative to SOL (stake-rate priced)E-Mode eligible for SOL pairs, higher LTV
ETH, BTCModerate (50-80% annualized)Standard LTV, similar to SOL
Newer tokensHigh (100-200%+ annualized)Lower LTV, potential isolation mode
Note that LST volatility is measured relative to SOL when priced via stake-rate oracles — the stake rate increases monotonically, so LST/SOL volatility is near zero. This enables E-Mode with higher LTV for LST/SOL positions.

​
Relationship to Other Market Risk Metrics

Volatility analysis does not operate in isolation. A token could have moderate volatility but extremely thin liquidity — meaning that even moderate price moves make liquidation difficult because selling the collateral causes massive price impact. Conversely, a token could be highly volatile but deeply liquid — meaning liquidators can execute quickly at reasonable cost even during price swings. The Liquidity & Price Impact analysis is the complementary dimension: volatility tells you how fast prices can move; liquidity tells you whether liquidators can execute when they do. The KRAF Dashboard provides real-time Parkinson’s volatility time series for all listed assets, enabling continuous monitoring and comparison across tokens and time periods.