# You are not earning the fee tier any more

> Uniswap's fee switch is on. We read all 25 pools we track from the chain: the protocol takes a quarter of the low tiers and a sixth of the 0.3%.

- Source: https://defiloops.com/blog/you-are-not-earning-the-fee-tier-any-more
- Published: 2026-09-16
- Category: Strategy
- Tags: uniswap, liquidity, fees, unification
- Author: DeFiLoops

---
If you provide liquidity in a 0.05% pool, you are not paid 0.05% any more. On the pools where
it has been switched on, a quarter of that goes to the protocol before it reaches you.

This is not obscure. It passed by near-unanimous governance vote in December 2025 and was
covered widely — as a story about UNI, buybacks and token burns. Almost nobody wrote the
sentence that matters to a liquidity provider, which is that **the number in the pool's name
is no longer the number you earn.**

## What we measured

Not from an announcement. Read from the chain on **1 September 2026**, across all 25 pools we
track on Ethereum, Base and Arbitrum:

<Spec rows={[
  ['0.01% and 0.05% tiers', 'The protocol takes one quarter. You keep 75%'],
  ['0.30% tier', 'The protocol takes one sixth. You keep 83.3%'],
]} />

Uniswap's own documentation states the 0.3% case plainly: with fees off an LP earns 0.3%, and
once activated the LP earns 0.25% while 0.05% goes to the protocol.

<Callout type="warn" title="These are the pools we track, on the date we read them">
  It is a snapshot, not a constant. More pools have been switched on since the vote and more
  will be. Any figure in this article is true of a specific set of pools on a specific day, and
  the next section is how to check a pool yourself rather than trusting this one.
</Callout>

## The asymmetry is a strategy change

A quarter against a sixth is not a rounding difference. It hands the 0.3% tier an **11%
relative advantage** over 0.05% that did not exist before.

That is not enough to make a wide tier beat a narrow one where the narrow one was genuinely
better — across our corpus it flipped no winner. But it narrows every gap, and any tier
comparison you did before December 2025 was run under different arithmetic. If two tiers were
close then, they are not close in the same way now.

## The rollout is per pool, and assuming otherwise is silent

This is the part that will catch people, because it fails quietly rather than loudly.

It did not arrive chain by chain. It arrived **pool by pool**:

<Spec rows={[
  ['Ethereum WETH/USDC and WBTC', 'Switched 27 December 2025'],
  ['Ethereum USDC/USDT', 'Waited until 6 March 2026'],
  ['Base and Arbitrum', 'Along with that later wave'],
  ['Uniswap v4', 'Extended July 2026, across seven networks at once'],
]} />

So two pools on the same chain, on the same day, can pay differently. Any tool that models
this as a per-chain or per-version flag will be wrong for months at a time, and wrong silently
— nothing errors, the yield figure is just too high.

## How to check a pool yourself

The truth is on the chain and takes one call:

```
cast call <pool> "slot0()(uint160,int24,uint16,uint16,uint16,uint8,bool)"
```

The `feeProtocol` field is a nibble per token. A value of `n` means the protocol takes **1/n**,
and `0` means it takes nothing. So `4` is a quarter and `6` is a sixth.

If you want the date it changed rather than the current state, read the event:

```
cast logs --address <pool> $(cast keccak 'SetFeeProtocol(uint8,uint8,uint8,uint8)')
```

<Callout type="note" title="Take the event whose old value is zero">
  The data word is old0, old1, new0, new1. There are later re-emissions that change nothing, so
  the switch-on is the one moving away from zero — not necessarily the most recent.
</Callout>

## What this invalidates, including ours

Every LP yield figure computed from the Uniswap subgraph and covering any period after a pool
switched is **too high**, because the subgraph never had a protocol-fee term. It derives a
day's fees as volume times the tier and stops there, which was correct for the whole of v3's
life until it suddenly was not.

That includes a figure we published. Our post on [why half of LPs lose
money](/blog/half-of-lps-lose-money) quotes a **57.9% fee APR** on the Ethereum USDC/WETH 0.05%
pool. That pool switched on 27 December 2025, that tier is a quarter, and the post went out on
25 August 2026 — a week before we read any of this from the chain. The figure is gross.

<Callout type="note" title="The conclusion survives, which is the awkward part">
  That post's argument is that a 57.9% fee APR still finished the year down 6.4%, because
  impermanent loss was larger. Cutting the fee side makes the gap wider, not narrower. The
  number is wrong and the article was, if anything, understating its own case.
</Callout>

We are re-running it rather than adjusting it by hand, because a corrected figure you
arithmetic your way to is a guess with a decimal point on it.

## What to actually do

- **Re-check any tier comparison made before December 2025.** The arithmetic underneath it
  changed.
- **Do not trust a fee APR without asking whether it is gross.** Most published ones are, and
  most do not say.
- **Read the pool, not the article.** Including this one. It has a date on it for that reason.