Steepen Rebase Curve to Tighten Price Distribution

Summary

Steepen rebase curve and tighten equilibrium bands to promote faster equilibrium discovery

Abstract

This proposes an update to the growth, upper asymptote, and lower asymptote configuration parameters of the Sigmoid reaction lag function. Additionally, it proposes tightening the equilibrium band within which no rebases occur.

Motivation

The Sigmoid function (implemented in AIP-5) replaced a linear function with an S-shaped function, allowing for sophisticated supply reaction tuning. However, to ease this transition, AIP-5 kept the steepest segment of the new curve as similar as possible to that of the original linear function. For commentary on the original proposal see: Rebase curve upgrade proposal

There is now sufficient reason to believe that a steeper slope would result in a tighter price distribution around the target, and thus better serve AMPL’s unit of account function. Recognizing that the primary source of price discovery and liquidity exists on constant product AMM’s, we can more precisely estimate the supply change required to bring price into equilibrium as:

  • An X% price deviation requires an X% supply adjustment to restore price equilibrium.

Moreover, tightening equilibrium bands (ie: the threshold of tolerance within which no rebases occur) would allow the system to adjust granularly according to smaller changes in demand rather than waiting until price is 5% away from its target to take effect, causing AMPL To spend more time at its precise target.

Specification

Recall that the sigmoid function curve is defined by:

  • F(x) : fraction of supply added or removed (supply change %)
  • x : normalized price deviation

It has shaping parameters that determine: lower asymptote, upper asymptote, and the steepness of the curve (ie: growth rate).

  • l = lower asymptote
  • u = upper asymptote
  • g = growth rate

Proposed Parameter Values

This update proposes changing the parameter values to:

  • l = -0.077
  • u = 0.05
  • g = 45

The resulting reaction lag curve looks as follows:

The red line, pictured above, represents a linear curve that would offset an X% price deviation with an X% supply change.

The proposed parameter changes produce a sigmoid curve that closely resembles the slope of the red line near the origin, but asymptotes horizontally at a max supply change of +5% above and -7.7% below AMPL’s price target.

This update also proposes changing the deviation ratio values:

  • deviation_ratio = 0.025

Conclusion

The Ampleforth protocol’s rebase reaction curve will keep its existing structure, but will more quickly and precisely adjust supply in response to demand, such that AMPL spends a maximum amount of time near its target.

4 Likes

Excited to see this proposal! As I have been advocating for a change in that direction since 2020 with the earlier version of AIP5, that eventually resulted in implementing the sigmoid curve to able to configure such curve to achieve faster conversion while providing safe caps on the maximum and minimum possible rebase percentages.

Going back to basics, the motivation of rebase is to converge back to the target price. In case of negative rebases. Reducing supply by small amounts while it moves the price in the right direction it can often result in prolonged contraction periods as it’s not pushing the market towards equilibrium strongly enough.

Here is a chart to compare the effect on conversion period for the proposed curve.

For apples to apples comparison including the proposed curve with two different definitions of equilibrium.

  • 5% Deviation which can be directly compared to current as it also defines equilibrium as less than 5% deviation from price target.
  • 2.5% Deviation which would be the new equilibrium definition under the proposal above

Predictably for some value it takes one more step longer to get within 2.5% than 5% with the same curve.

2 Likes

thanks @Naguib think this pretty much makes the case

Updated my graph visualizations of the rebase curves to add the newly proposed configuration, see Desmos | Graphing Calculator. Hope it helps getting a better feeling for the different curves.

4 Likes

Let’s get this out of the door ASAP. We either go to the moon or all die trying.

3 Likes

This is a significant change and affects everything in the ecosystem. I think this will be a good test for Spots durability, if anything else.

I do think we need shorter periods of rebase, especially negative (which feel as though they can take years sometimes)

Im willing to accept that tradeoff for shorter yet more potent cycles.

Lets get this show on the road gents

1 Like

Given the clear support on Discord during the office hours and here, starting the vote

https://signal.ampleforth.org/#/proposal/0xc4b307e33572f12f47bf2d5b4891f29902d5fb01e84c8f634500c938257b23bd

1 Like

I would think that it may be best to activate the implementation of the new curve only once we enter the neutral phase for the first time.

Can you imagine the chaos/panic created if one day you get a 2% negative rebase because VWAP is at $0.93, and the very next day you get negative 7.7% with VWAP at the same $0.93?

1 Like

Is there going to be a new ampl target price with this new implementation? Based on the graph, is target price no $1?

This update makes no changes to the AMPL price target. Those axes are “normalized price”. You can think of that value as a percent deviation away from the price target, currently $1.19 according to the dashboard.