What 'Captured Gains' Actually Means: A May 2026 Walkthrough

DeFi APYs are not still water. On 2026-05-18, the USDC supply rate on Compound (Base) did this in twelve hours:

00:00 — 3.16%
04:00 — 6.08%
08:00 — 3.15%
12:00 — 5.16%

Morpho (Arbitrum) had its own move on the same day: 5.82% climbing to 8.09% over a few hours, then mean-reverting.

If your habit is to open your portfolio once or twice a day, you didn't see any of this. By the time you checked, the curve was back to baseline and the dashboard read "still 3-something percent". The Apyee Keeper runs every 5 minutes on every chain. This piece is about what that 5-minute cadence actually catches — and how to read the number it surfaces on your portfolio page.

The 'Captured Gains' card

If you've deposited into any chain's vault, the Portfolio page now shows a card called Captured Gains. It displays three things:

+X.XX%   avg captured APY uplift · <chain>
N rebalances · avg +X.XX% per cycle

Where:

• avg captured APY uplift — average estimatedGainBps across the last N cycles, holding-weighted (each capture's contribution proportional to how long the vault held the new allocation). Converted to a percentage so it reads like APY.
• N rebalances — how many yield-reason rebalances fired during the window.
• avg per cycle — simple average, not holding-weighted. Lets you compare cycle quality without time bias.

Both numbers come from on-chain VaultEvent records. The fields used — apyFromAt, apyToAt, estimatedGainBps, reasonCode — are written by the Keeper at the moment of each rebalance. There is no separate "marketing database".

What gets included

Only reasonCode = 'yield' rebalances count. The other reasonCodes (alloc, hardcap, emergency) exist for housekeeping, not for chasing spikes — including them in the average would smear the headline. Specifically:

• alloc re-deploys idle capital after an emergency exit. It restores normal allocation, not new yield capture.
• hardcap corrects an over-allocated strategy. The move is mechanical, not opportunistic.
• emergency exits a risky strategy. The "uplift" is negative or zero by definition — you're choosing safety over yield.

Filtering these out is the difference between the card showing real predictive power and showing meaningless activity volume.

What gets captured

Two numbers matter on every yield rebalance:

• APY uplift at entry (apyToAt - apyFromAt in bps) — how big the spike was at the moment of action. Dramatic but unrealized.
• Estimated net gain after gas (estimatedGainBps) — what's left once gas is subtracted, normalized to an APY-equivalent over the expected holding window. Conservative but actionable.

Apyee surfaces both — the first as a hint on the rebalance row, the second as the marketing headline. The card averages estimatedGainBps, because anything else would overstate.

A small user, a real capture

The most recent Apyee depositor we have data for held a $4 position on BSC during one rebalance on May 18. Their card reads:

+2.09%   avg captured APY uplift · BSC
1 rebalance · avg +2.09% per cycle

Two things worth sitting with.

First: 2.09 percentage points is not 2.09% of the deposit. The capture is annualized; the realized profit over a 6-hour holding window at +2.09% APY is roughly 2.09% × (6 / 24 / 365) ≈ +0.0014% of principal. For $4, that's a fraction of a cent. The number is useful as a measurement of rate quality, not as a dollar receipt.

Second: the vault paid the gas for that rebalance, socialized across all BSC depositors. A $4 position got the rate uplift without having to justify the gas itself — that arithmetic only works because the vault is the unit charged for gas, not the user.

Scaling this out: a $10,000 position seeing 6 captures per month at an average +200 bps APY for 6-hour holding periods would realize roughly $10,000 × (200 bps × 6 / 24 / 365 × 6) ≈ $1.4/month of additional yield over the un-rebalanced baseline. Small in absolute terms, but it's free relative to the alternative of doing nothing.

Why the number can disappoint

Three reasons the Captured Gains card can show 0.0% or hide itself entirely:

Market was flat for the window. No qualifying spike means no yield rebalance. The card hides itself on N=0 to avoid implying the Keeper failed; it just had nothing to do.

Your holding window doesn't overlap captures. You deposited two days ago, but the captures that drove the 30-day average happened three weeks ago. The card shows your holding-weighted uplift, not the vault's lifetime number. New depositors will see lower numbers until enough cycles accumulate during their tenure.

Your chain is L1. Ethereum gas at ~$8/rebalance disqualifies most spikes from the 3× gas ROI gate. The card is structurally smaller for ETH and structurally larger for L2s. Lowering the gas gate would mean firing rebalances that don't actually beat their cost — the opposite of capturing anything.

Honest caveats

The number is an estimate, not a guarantee:

• estimatedGainBps is computed at entry, using current gas and assumed holding time. If gas spikes the moment after the transaction or the next rebalance fires sooner than expected, realized capture differs from the estimate. Direction (up or down) is correct in the vast majority of cases; magnitude is approximate.
• The card excludes the 15% performance fee. That fee applies to all yield, including captured yield, at the next Keeper harvest. So the displayed uplift overstates what flows to the depositor by about 15%.
• Survivorship of high-yield strategies. A strategy can pay 8% for a week and then get shut down. The capture log records the trade, not the long-tail risk. Read risk rebalances and emergency exits alongside captures to get the full picture.

The honest framing

If you remember one thing: the Captured Gains card measures the slope of your rate, not your rate itself. Most depositors look at "current APY" and stop there. That number is the average. The slope is the additional yield you would miss without continuous monitoring — and continuous monitoring is exactly what an automated vault should do for you.

A small position making 2.09% captured uplift in one day isn't a brag. It's evidence that the cadence works. Scaled, the same cadence is what the 15% performance fee earns.