Impermanent Loss Ate My Yield: Real Numbers From 6 Months of LP Positions

Why I Tracked This

Six months ago I deployed five LP positions across Uniswap V3, Curve, and a couple of niche venues, and I committed to logging every fill, every fee collection, every range rebalance, and the underlying asset prices weekly. The reason: every "passive income" pitch on DeFi Twitter has the same shape — APY on the front, no mention of impermanent loss on the back. I wanted my own data.

This article is the data. Five positions, six months, real P&L including gas costs and IL versus a true HODL baseline. The positions:

1. ETH/USDC on Uniswap V3, concentrated range (+/-15% of entry)
2. WBTC/ETH on Uniswap V3, wide range (+/-50% of entry)
3. stETH/ETH on Curve
4. FRAX/USDC on Uniswap V3, concentrated stable range
5. BTC/ETH/SOL on a Curve tri-asset pool

I am not naming the venues for the wide-range pools to avoid implying anything specific about those venues' performance. The pools, fees, and capital sizes are real but anonymized at the venue level.

Important disclaimer: This article reports my personal results from a specific six-month period. Past performance does not predict future performance. Liquidity provision involves risk including but not limited to impermanent loss, smart contract risk, oracle risk, regulatory risk, and total loss of capital. Nothing in this article is financial advice, investment advice, or a recommendation. DeFi protocols are unregulated in most jurisdictions; recovery options are limited if something goes wrong. Consult a qualified financial advisor before deploying capital. Never invest more than you can afford to lose entirely.

Setting the Baseline: HODL

For each position, the HODL baseline is what the same dollar amount would have been worth if I had simply bought the two (or three) assets in 50/50 (or 33/33/33) proportion at the entry timestamp and held them in a wallet untouched.

This is the only fair comparison. Comparing LP returns to "the price of ETH" or to USD is not meaningful — the LP holder is exposed to two assets, not one. The blended HODL of those two assets is the thing the LPing has to beat.

I did not try to time anything. Same entry timestamp for the LP and the HODL baseline. Same exit timestamp. Same dollar amounts. The only difference is whether the dollars went into a wallet or into a pool position.

Position 1: ETH/USDC Concentrated Uniswap V3

• Capital: $10,000 (5,000 ETH-side, 5,000 USDC-side)
• Range: +/-15% of entry price
• Fee tier: 0.05%
• Period: 6 months
• Gas cost: $185 across two rebalances and one entry/exit

Result: -3.1% versus HODL.

The ETH price moved +24% over the period, with two excursions that briefly took the price out of the +/-15% range. During out-of-range time the position earned no fees, and on re-entry I rebalanced (which has its own gas cost and tax implications). Fee revenue was healthy at +5.4% (annualized higher because heavy volume periods coincided with my range), but IL was -8.5%. Net: -3.1% versus HODL.

This is the classic Uniswap V3 trap. The fee revenue looked good in the protocol UI. The actual P&L versus HODL was negative. If I had widened the range to +/-30%, the fee yield would have been about half but the in-range time would have been near 100%; my modeling shows that would have produced about -1% versus HODL — still a loss, but smaller. The honest answer is that ETH/USDC concentrated LPing during a directional ETH move is not a great trade.

Position 2: WBTC/ETH Wide-Range Uniswap V3

• Capital: $10,000
• Range: +/-50% of entry ratio
• Fee tier: 0.30%
• Period: 6 months
• Gas cost: $42

Result: -4.3% versus HODL.

This was my worst position. ETH outperformed BTC by about 18% during the window. The pool rebalanced me into more WBTC and less ETH continuously throughout the period. Fee revenue was modest at +3.2% (BTC/ETH does not see the volume of stable or popular pairs at this fee tier). IL was -7.5%. Net: -4.3% versus HODL.

The lesson here is uncomfortable for the conventional wisdom that "WBTC/ETH is a blue-chip pair." ETH and BTC are correlated but not identical. In the 2022-2024 cycle their correlation was high. In 2024-2026 they have decoupled meaningfully — different ETF flows, different macro narratives, different supply dynamics. Correlated does not mean correlated 1.0, and a small decoupling over six months produces meaningful IL.

If you want exposure to BTC and ETH, holding them in your wallet is straightforwardly better than LPing them, in current correlation conditions.

Position 3: stETH/ETH on Curve

• Capital: $10,000
• Pool: Curve stETH/ETH
• Period: 6 months
• Gas cost: $48 (entry, exit, and one fee claim)

Result: +1.8% versus HODL (excluding stETH staking yield).

This was the position that performed exactly as advertised. stETH and ETH have a peg-style relationship — stETH should redeem 1:1 to ETH minus a small redemption queue discount. The Curve pool absorbs the small fluctuations and earns trading fees on traders who want to swap between the two without going through the redemption queue.

IL was effectively zero (less than -0.2%). Fee revenue plus CRV/CVX rewards (after standard Convex routing) totaled +1.8% versus HODL for the six-month window — about 3.6% annualized. The stETH itself also earned its underlying staking yield (about 3.4% annualized) which I am holding constant in the comparison since the HODL baseline also holds stETH.

This is the kind of LP position that should make up the boring core of a yield-farming portfolio. Modest, reliable, low-risk-of-catastrophe.

Position 4: FRAX/USDC Concentrated Stable

• Capital: $10,000
• Range: $0.998 - $1.002 (very tight stable range)
• Fee tier: 0.01% (Uniswap V3 stable tier)
• Period: 6 months
• Gas cost: $73 across one rebalance and the entry/exit

Result: +3.4% versus HODL.

The cleanest win of the five positions. Both assets are pegged to USD; both have stayed in their range almost the entire period; the volume on FRAX/USDC at the 0.01% tier is enormous because every stable-to-stable swap routes through it. Fee yield annualized about 6.8%. IL was zero to within rounding error.

This is the boring success story. People do not write Twitter threads about FRAX/USDC LPing because it is unsexy. It is also one of the best risk-adjusted yields in DeFi for capital that needs to stay in stables.

Position 5: BTC/ETH/SOL Curve Tri-Asset

• Capital: $10,000
• Pool: Curve tri-asset pool with managed weights
• Period: 6 months
• Gas cost: $61

Result: -2.4% versus HODL.

Curve's volatile-asset pools use a different bonding curve than Uniswap V2/V3 and theoretically reduce IL on volatile pairs. In practice for this pool: the IL was real because SOL outperformed both BTC and ETH meaningfully (about +35% versus the others' moves). The pool rebalanced into more BTC and ETH and less SOL over the period. Fee revenue plus reward incentives totaled about +5.1%. IL was -7.5%. Net: -2.4% versus HODL.

The Curve "v2" math helps versus Uniswap V2 but does not eliminate IL when one asset materially outperforms the others. Tri-asset pools also have correlation considerations across three assets, not two — the IL math compounds.

Aggregate Results

If I had simply put the same $50,000 into a 50/50 split of "correlated stable pairs" and "spot HODL of major coins," I would have outperformed my actual mixed-portfolio result by approximately +1.7% over the six months.

This matches the structural intuition: IL is bounded for correlated pairs and material for volatile pairs. The fee revenue rarely makes up for IL on volatile pairs in any meaningful directional move.

When Concentrated Liquidity Helps Versus Hurts

The Uniswap V3 concentrated liquidity model is sometimes pitched as "an IL reducer." It is not. It is a fee-earning amplifier when the price stays in range and an IL amplifier when the price exits range and you do not rebalance. Specifically:

Helps when:

• The pair is correlated and stays bounded (FRAX/USDC, stETH/ETH)
• You actively manage the range
• Volume is high enough that the amplified fee revenue meaningfully exceeds the gas cost of rebalancing

Hurts when:

• The pair is volatile and goes through a directional move
• You set and forget — the price exits range, fees stop, and you hold the depreciating side
• Capital is small enough that gas cost of rebalancing eats meaningful percentage points

A practical rule: do not use V3 concentrated ranges on volatile pairs unless you are committing to active management. Wide V3 ranges or V2 pools are more forgiving.

Curve Stable Pools as the Workhorse

Across the LP literature and my own data, Curve stable pools (and Uniswap V3 0.01% stable tier) are the most reliable yield venue for capital that can sit in stables. The reasons:

• IL is bounded by the peg-keeping mechanisms of the underlying tokens
• Volume is high because every stable-to-stable swap in DeFi routes through them
• Smart contract risk is concentrated on a small number of well-audited contracts
• Yields are modest but reliable (typically 2-8% APY depending on incentive layers)

The 2024-2025 stablecoin issuer landscape (USDC, USDT, FRAX, crvUSD, sUSDe, GHO) means there is always a healthy pool somewhere. The yield is unsexy. The Sharpe ratio is excellent.

Active Range Management That Actually Helped

The two positions where active management mattered most: ETH/USDC and FRAX/USDC. For ETH/USDC, my rebalances during the period were both reactive (price exited range) rather than proactive (anticipated volatility). With the benefit of hindsight, a more proactive policy — widening the range during expected high-volatility events (CPI prints, FOMC meetings, ETF news) and tightening during quiet periods — would have improved the position by an estimated 1.5-2 percentage points.

For FRAX/USDC, the one rebalance was a precaution that turned out to be unnecessary. Stable pools rarely need rebalancing absent a depeg event.

The honest takeaway: active range management is real work. It requires watching price, watching volatility, paying gas to rebalance, and accepting tax events on each rebalance (in jurisdictions where rebalancing is treated as a taxable event, which includes the US). For positions under $50k, the math often does not justify the effort. For positions over $250k, the time-and-tooling investment pays off.

Tools That Tell the Truth

The single most important LP-related tool: a P&L tracker that nets out IL.

[Revert Finance](https://revert.finance/) — Best-in-class for Uniswap V3. Shows fees earned, IL, gas cost, range coverage time, and net return versus HODL in one view. Free for basic use. Premium tier for advanced features.

[DeFiLab](https://defilab.xyz/) — Stronger range-management features. Lets you simulate range adjustments before executing.

[DeBank](https://debank.com) — Cross-protocol portfolio tracker. Useful for the broader picture but does not deeply track IL on individual LP positions.

[Apy.vision](https://apy.vision/) — Comprehensive multi-DEX LP P&L tracker. Strong for users with positions across many DEXes.

The protocols' own UIs systematically understate IL because they show fees collected (a positive number) without netting against the HODL counterfactual. This is not fraud — they cannot know your entry HODL counterfactual unless you provide entry data — but it does produce a cognitive bias toward overestimating LP returns.

For broader yield-farming context including non-LP strategies, see our DeFi yield farming guide and what is yield farming explainer.

Strategies That Beat HODL in My Data

Three patterns worked consistently:

1. Correlated pairs only. Stable-to-stable, liquid-staking-token-to-underlying, peg-mechanism pairs. The IL is bounded so the fee yield reliably stays in the green.

2. Active management of V3 concentrated ranges on the limited universe of pairs where it makes sense. ETH/USDC during low-volatility regimes with weekly check-ins. Tightly managed.

3. Curve gauge incentives on top of stable pools. The Convex/Aura/Yearn aggregator layer adds 1-3% APY on top of base Curve fees with manageable additional risk. The catch is governance-token volatility for the rewards portion.

Strategies That Did Not Beat HODL

Volatile uncorrelated pairs. ETH/altcoin, WBTC/altcoin. The IL crushed the fees in every directional move.

V3 concentrated set-and-forget on volatile pairs. The price exits range, the position stops earning, and the rebalance you should have done in week 4 you finally do in week 12 after you have lost the equivalent of two months of fees.

Exotic small-cap pairs. The "high APY" was real but came with extreme IL risk and meaningful smart-contract risk. Even when the fees outpaced the IL on a specific position, one rug or exploit elsewhere wiped out the gains.

A Closing Disclaimer

This article reports my personal experience over a specific six-month window. Past results do not predict future results. Liquidity provision is a real economic activity with real risks: impermanent loss (real and frequently permanent), smart contract risk (audits do not catch everything), oracle risk for protocols using oracles, regulatory risk (DeFi regulation in 2026 is in flux), and total-loss risk on the underlying tokens. The information here is educational. Nothing here is investment advice or a recommendation to deploy capital.

If you are new to LPing: start with stable pairs, start small, use a P&L tracker that nets IL, and understand that you are taking on a complex bet with two-sided exposure rather than a pure yield product. If you are experienced: re-check your assumptions about correlated pairs in current correlation regimes — some "blue chip" pairs are not as correlated as they were two years ago.

The DeFi yield-farming pitch usually makes LPing sound like a savings account with a higher APY. It is not a savings account. It is a position. It can win and it can lose. The data above is one window of one operator's experience showing both.

For the wider context of DeFi yield strategies and where LPing fits, see our pillar guide: [DeFi Yield Farming Guide 2026](/blog/defi-yield-farming-guide-2026).