DATA PROVENANCE & SWEEPS

Every constant, traced to a sweep.

All experiments replay real on-chain swaps at sqrtPriceX96 precision against an oracle ground truth. This page shows the windows, the sweeps that picked each value, and the honest limits of what was measured.

A / EXPERIMENTAL SETUP
16-anchor alignment

Piecewise real-time alignment between pool events and the reference feed.

±300 s rolling median

De-trends the pool-vs-fair spread. 62% of raw variance is slow basis — not charged.

Binance 1 s candles

Calm-window reference. Volatile uses Pyth ETH/USD on a 12–60 s adaptive cadence.

Pyth benchmark prices

Historical benchmark feed used as the oracle ground truth for precision.

CALIBRATION WINDOWS
WINDOW POOL / REFERENCE RESULT
Calm
2026-05-30 · 20:04–22:30 UTC
Arbitrum WETH/USDC 0.05%
Binance ETHUSDC 1 s
629 swaps · median Δ 0.82 bps · capture 17%
Volatile
2026-05-23 · 17:00–21:00 UTC · 430 bps
Same pool
Pyth ETH/USD · 12–60 s adaptive
6,444 swaps · median Δ 3.70 bps · capture ~24–26% · LP bleed 22×/hr

Of these, 4,511 priced swaps (629 calm + 3,882 volatile) were replayed as the regression-test corpus behind every precision and capture figure.

B / PREMIUM SLOPE α SWEEP
TRUTH-CAPTURE vs α capture is first-order linear in α
60% 40% 20% 0% α = 0.5 0.25 0.75 1.0 1.25 25.56% 22.98%
Volatile
Calm

Why α = 0.5.

Capture rises almost perfectly linearly with the premium slope — more α harvests more, but charges more. Precision stays put: ~92.4% calm and ~98.9% volatile across the whole sweep.

PegGuard pins the conservative midpoint — enough capture to return value, low enough to keep the fee gentle on real LPs.

C / DEADBAND & STALENESS KNEES
Deadband sweep τ = 450 s

Selected 0.5 bp (50 e2) — the precision/capture knee in each window.

DEADBANDCALM P / CVOL P / C
0.0 bp90.83 / 28.2198.45 / 26.98
0.5 bp ▸92.42 / 22.9898.86 / 25.56
1.0 bp93.60 / 18.2399.15 / 24.23
P = precision %, C = truth-capture %. Higher deadband trades capture for precision.
Staleness guard

Older oracle data can inflate revenue while precision collapses — so the live Pyth-pull guard is a hard rule, not a tunable.

Calm precision retainedguard 5 s
2 s98%
5 s ▸95%
10 s91%
30 s77%
Volatile precision retainedguard 2 s
2 s ▸95%
5 s87%
30 s72%
D / SENTINEL CALIBRATION SWEEP

40 bps / 180 s.

Swept on the corrected volatile fixture (1,108 swaps, 418 bps) and a 250-event calm sentinel. The selected setting is the tightest trigger with zero calm false positives.

It arms the volatile regime 74 minutes before the measured heavy-bleed window. The looser 60 bps trigger only fires in-window — after $2,399 has already bled.

SHA-256
4eb34aa0…bbae8348
TRIG WINDOW CALM FP VOL TRIG 1st VOL BLED
201800141825 s
203001151825 s
206003221825 s
301800104332 s
303000124332 s
306001164332 s
40 ▸180064332 spre-window
40300064332 s
40600084332 s
601800312656 s$2,399
603000412656 s$2,399
606000412656 s$2,399
E / NET LP BY SWAP SIZE

Net LP outcome and premium precision per bucket. Live forward-test P&L is deliberately omitted — these are the calibrated calm and volatile windows only.

Calm window
BUCKETNET bpsPREC.n
<$1k+1.2984.0%295
$1k–$10k+2.0093.2%331
$10k–$50k+0.43100%2
≥$50k0
Every populated bucket returns net-positive value.
Volatile window
BUCKETNET bpsPREC.n
<$1k+0.6799.1%958
$1k–$10k−0.2698.6%2,663
$10k–$50k−3.1098.7%254
≥$50k+19.69100%7
Amber n = small sample; read as directional, not precise.
F / ARCHITECTURE CONCLUSIONS
01
Hot path uses Pyth pull

The race clock is the pool's ~14 s update interval, not the arbitrageur's sub-2 s execution. Pyth at 2 s staleness retains 88% of edge.

02
Staleness is a trap

Older data raises measured revenue 25%→41% while precision collapses 95%→72%. Revenue can reward bad operation — so the guard is hard.

03
Consensus ceiling = best member

An AMM-only basket captured 3%; adding CEX only reached the pure-CEX 44%. Pairs with CEX feeds use Pyth directly.

04
Reactive only in control plane

Regime changes, breaker signals and baseline upkeep have minute-scale budgets where cross-chain delivery latency is acceptable.

G / HONEST CAVEATS & OPEN ITEMS

Capture is a same-swaps upper bound. Route-away elasticity is not modeled — in production, charging a premium may deter some of the very flow it prices.

No forward-test P&L is claimed. Every figure is a calibrated backtest. PegGuard makes no “profitable in production” claim.

RSC false-positive coverage is partial. Validated on the supplied 250-event calm fixture; broader coverage is an open item.

Guard-mode dual-feed is a skeleton. The stable-pair depeg replay (e.g. USDC/USD ÷ USDT/USD) remains separate; the current path uses a single feed.

Stable-pair audit

Why GUARD charges no directional premium — a provenance audit, not a depeg replay.

Normal stable deviation0.00–0.35 bps
Feed noise~0.2 bps
AMM basket capture0.60%
AMM + CEX capture6.30% · corr 0.58
GUARD trip margin142.9×

Signal sits at feed-noise level — no directional fuel. The breaker is cleanly separated from normal stable noise.