Table of Contents
Do not index
Introduction
In the world of DeFi, liquidity providers (LPs) often look at APR as the main metric for assessing returns. However, calculating and interpreting APR in concentrated liquidity models (CLAMM / V3 DEXes) like those powered by Algebra requires nuance. In this article, we’ll walk through our methodology, the limitations of APR, and more robust approaches to evaluating pools.
How We Calculate APR
The standard formula we use is:
APR = (Last Day Fees ÷ TVL) × 365 × 100
In practice:
- We take active positions in the pool.
- Calculate the current TVL in one of the tokens.
- Measure how many fees were collected over the last 24 hours in that same token.
- Divide fees by TVL, annualize it, and convert to percentage.
This gives us an approximate APR for the pool.
Important Nuances
- Static assumption: This method assumes active positions didn’t change over the last day. With volatility, APR may be overstated or understated.
- Range differences: In v3 pools, LPs choose custom ranges. APR for individual positions may differ significantly from the average. Our method implicitly averages across active positions, which may not reflect a new LP’s outcome.
- Token denomination: Whether you count TVL and fees in token X or token Y doesn’t matter — APR is dimensionless.
- USD anchor: For pools with a USD stablecoin, we calculate based on fair market value in USD. For non-stable pools, growth is measured in another token, isolating the APR but exposing LPs to additional risks (both tokens can lose value vs USD while the pool price remains constant).
Alternative Approach
Another way is to divide fees by total TVL of all positions (not just active ones).
- Pros: More stable during volatile periods.
- Cons: Systematically lower APR, since it counts out-of-range positions that earn nothing.
Core Problems with APR
- Instability: Yesterday’s fees ≠ tomorrow’s. Short-term snapshots are volatile.
- Liquidity distribution blindspot: Pools with tighter ranges show higher APR, but the metric ignores how liquidity is actually distributed.
- Risk-blind: A pool may show high APR, but position value can decline (impermanent loss, token devaluation).
- Volatility link: High fees usually come during high volatility. But volatility = higher impermanent loss. So the highest APR pools often carry the highest risks.
Conclusion: APR is useful for comparing average yields across pools, but might be insufficient for deciding which pool to join.
How to Choose a Pool
Let’s compare some of the pools’ setups:
1. Stablecoin Pools (Stable vs Stable)
- Main risk: depeg.
- Hard to quantify, requires individual research into each stablecoin.
- Metric: APR ÷ depeg risk.
- Example (Camelot):
FDUSD–USDC: 150%USDT–USDC: 5.7%2. Fee Tier Comparisons (same pair, non-stable)
- Volatility is roughly the same across tiers.
- What matters: concentration. Higher concentration = higher APR.
- We use normalized TVL: what TVL would look like if all positions were full-range. This allows fairer APR comparison across fee tiers.
3. Stable–Volatile Pools
Unlike fee tiers, pools can contain different tokens, which means different prices and different risk levels. The higher the volatility, the higher the expected impermanent loss (IL) in the pool. The most straightforward way to account for this risk is to divide by volatility — but this assumes a linear relationship between volatility and IL, which is not accurate. Instead, it’s necessary to calculate the expected value of IL in the pool. This can be done for v2 pools, since we normalize TVL (The formula proposed by Daniel Alcarraz — source)
- Different risks: stable depeg vs impermanent loss.
- To capture risk-adjusted yield, we subtract expected impermanent loss (E[IL]) from APR:
- E[IL] is calculated via volatility (σ) and expected log-return (ν) of the token pair, annualized from historical data.
4. Volatile–Volatile Pools
The classic calculation of impermanent loss (IL) assumes that one of the tokens in the pool is stable (the one used to measure the position’s value). However, in volatile–volatile pools this is not the case. Losses relative to USD may occur, even if no IL is observed within the pool itself.
To account for this risk, it’s necessary to recalculate both the pool position and the hold position in terms of external USD prices for the tokens:
It is possible to use the classical IL metric, but this would ignore the additional risk specific to volatile–volatile pools, resulting in an overestimated metric.
Pools where one token overlaps (e.g., ETH–volatile pairs) can be compared without recalculation. In such cases, IL is measured relative to the second token, while ETH is treated as the base (quasi-stable). This means profits and losses are calculated in ETH, and ETH’s own price movement (up or down) does not matter for the IL calculation.
- Trickier, since neither token is stable. Both may lose value relative to USD, even if IL inside the pool is zero.
- Requires recalculating position and hold values in USD.
- Pools sharing one token (e.g., ETH–volatile) can be compared without conversion by treating ETH as the base.
Key Takeaways
When it comes to providing liquidity there are many factors at play. In CLAMMs, APR depends heavily on volatility and the share of active liquidity in the pool — making it often highly unstable.
Different approaches to measuring efficiency can be applied to different token pairs — so some discrepancies and inaccuracies are only natural.
