Skip to main content
The hook uses a frac^1.5 penalty curve. I covered it here. But there are plenty of curves. Always have been. This page lists alternative curves any developer could swap in — for example, if a version that punishes aggressive bidders more harshly is needed.
None of the alternatives below are implemented in PenaltyFeeLibrary today — this page is a reference for anyone forking the hook, not a description of a toggle in the current code.

​
1. Linear (frac^1)

Frame 47
The simplest option: the penalty grows in direct proportion to how far past the threshold the swap is. Problem: it’s too aggressive right at the threshold. A swap at 3.5x, only slightly over the 2.7x line and arguably still an eager but honest bidder, already gets hit with a 1.1% penalty. Real MEV / sandwich bidding tends to sit far above the median (5x, 10x+), not just barely above it. A linear curve punishes the “slightly aggressive but honest” trader almost as hard, proportionally, as it punishes genuine outliers, which is the opposite of the goal.

​
2. Quadratic (frac^2)

Frame 48
The opposite problem: it’s too forgiving early, then whips up sharply near the cap. At 4.5x the penalty is only 0.61% (barely a deterrent), but between 8x and 10x it accelerates fast. This creates a narrow “danger zone” right before the cap where the fee jumps a lot for a small change in tip: a swap bidding 8.5x pays 6.31%, but 9.5x already pays 8.68%. That’s a steep, hard-to-predict gradient exactly where users are most likely to be bidding competitively. This was actually the curve chosen while the hook was being brainstormed. It’s not bad, but it was considered too cruel in the “danger zone” above.

​
3. Step zones (“big steps”)

Frame 46
Cheapest to compute (just comparisons, no interpolation or sqrt) and easy to reason about. Problem: cliff edges. A swap at 4.69x pays 1%, but a swap at 4.71x, essentially the same tip, suddenly pays 3%. That’s an obvious incentive to game the boundary: bidders cluster their priority fee just under each threshold rather than bidding what they think the swap is worth. Two nearly identical swaps also end up paying very different fees, which feels arbitrary.

​
Why frac^1.5 was chosen

Frame 52
I wrote down a target reference table:
  • ~1-2% penalty around 4-5x (mild deterrent for slightly-aggressive-but-plausibly-honest bidders)
  • ~3-5% around 7x (a real cost starts to bite for clearly aggressive bidding)
  • exactly 10% at 10x (the hard cap)
frac^1.5 happens to hit almost exactly these numbers. It sits between linear and quadratic: steeper than a straight line, gentler than a square.

​
Key takeaway

  • Linear punishes honest bidders too hard right at the threshold.
  • Quadratic is too soft early and too steep near the cap.
  • Step zones are cheapest to compute but create cliff edges that invite gaming.
  • frac^1.5 is smooth - it matches the target penalty shape.
  • You can choose other curve if you have other vision of penalty brutality.
Last modified on September 3, 2026