This post is ported over from a foreword written by Manny Rincon-Cruz and Evan Kuo in reference to Gauntlet Network's forthcoming report about the Ampleforth network. This foreword and report is also hosted on our website
Link to full report


In this foreword, we seek to capture our thoughts and reactions to the Gauntlet Network’s forthcoming report on the Ampleforth network’s performance, which has given us an opportunity to reflect on Ampleforth’s founding hypotheses and intellectual motivations. The Gauntlet Network is an independent third party, and so the opinions expressed in this foreword remain our own, and are not a reflection or endorsement of their work, which stands on its own. We nonetheless would like to express our thanks to the Gauntlet Network team for producing a stimulating piece of research.


The Ampleforth protocol was designed to create a new type of synthetic commodity (AMPL). Like Bitcoin, it would be uncollateralized. Unlike Bitcoin, it would eschew a fixed supply schedule and instead automatically adjust supply in response to price-exchange rate. Further analysis had led us to hypothesize that these rule-bound supply changes might lower the correlation of the AMPL market capitalization with those of BTC and ETH. Two significant consequences could follow if AMPL were to become such an uncollateralized asset with a volatility signature unlike other digital assets:

  1. AMPL could add value to baskets of digital assets, reducing aggregate volatility, if it were lowly correlated with existing digital assets.
  2. AMPL could be a useful building block in broader decentralized finance (DeFi) applications, much as physical commodities were building blocks in the historical financial system.

The Gauntlet Network’s recent research validates AMPL’s protocol design and hypotheses, showing that in 1H2020, AMPL’s market capitalization was only correlated with BTC and ETH at 0.048 and 0.020, respectively, which accords with Gauntlet’s agent simulation results as well.


The search for an “ideal” base money has long preoccupied monetary economists, and in 2015 George Selgin published a particularly relevant article in the Journal of Financial Stability titled “Synthetic Commodity Money.” He was investigating the possible use of current-generation cryptocurrencies for monetary reform, and noted that base-monies—which are irredeemable and uncollateralized—conventionally fall into one of two categories: “commodity” money and “fiat” money. Selgin observed that cryptocurrencies like Bitcoin break this conventional dichotomy by resembling both “fiat” and “commodity” monies. Specifically, these digital assets:

  1. Resemble fiat-money in having no non-monetary use.
  2. Resemble commodity-money in being absolutely scarce—that is, no authority has the discretion to “manage” its supply.

He concluded that the categorization to-date was incomplete, and that a superior classification would treat non-monetary value and absolute scarcity as independent properties. Selgin named one of the missing quadrants of his new classification system “Synthetic Commodity Money.” He suggested that synthetic commodity monies may be “especially capable of supplying the foundation for monetary regimes that are both macro-economically stable and constitutionally robust.”

The Ampleforth protocol was created with the goal of producing a new, and hopefully more ideal, synthetic commodity money—instead of following a fixed supply schedule, the protocol would perfectly enforce supply adjustments in response to demand. This operationalizes, in a way, the long-standing thesis by Nobel-laureate James M. Buchanan that rule-bound “predictability”—–as opposed to human discretion—–might allow for more effective financial institutions. Further analysis led us to hypothesize that these rule-bound changes in supply might lower the correlation of the AMPL market capitalization with those of BTC and ETH. Our May 2019 whitepaper asked, “can we create a synthetic commodity that is not highly correlated with either Bitcoin or other asset classes?”

While BTC and ETH were correlated 0.72, AMPL was correlated with BTC at only 0.048 and with ETH at only 0.20. Thus far, it seems that the Ampleforth protocol is working as designed.

The Gauntlet Network also conducted an agent simulation, which produced a market value dynamic independent from BTC and ETH, and sheds some light on how user incentives are playing out in the market for AMPL. Below, we summarize the main results, but we hope you will nonetheless read the main report here.

Historical Correlations

Gauntlet relied on on-chain pricing and supply information, and also used other empirical data obtained from the third-party provider, AmberData. Their analysis spans the 171 days from January 3 to June 22, 2020, and they look at time series per minute, hour, and day.

When looking at historical correlation in USD prices per day, Gauntlet found that BTC and ETH were correlated at .933224, BTC and AMPL were correlated at .567950, and ETH and AMPL correlated at .546765. In other words, in the span of an average trading day, BTC and ETH moved together a large majority of the time, and AMPL moved with them most of the time.

However, the rebase mechanism at the core of AMPL means that the supply of AMPL is dynamically changing in response to price movements. These supply changes can be observed in measures of the AMPL market capitalization, but not in BTC or ETH’s market capitalization, which have well-defined monotonic functions for increasing their token numbers.

The historical daily correlation of market capitalization, BTC and ETH stand at .724373. But AMPL’s correlation to BTC’s capitalization is only .048274, and with ETH’s capitalization, just .020120.

In other words, while the BTC and ETH market capitalizations still move in tandem, the supply fluctuations engendered by the Ampleforth protocol means that at least on a daily basis AMPL’s market cap marches to the beat of its own drum.

Even more tellingly, the 14400m rolling correlation of AMPL and ETH token prices show a massive spike around March 12—when the COVID-19 crisis impacted financial markets worldwide. The 14400m rolling correlation of AMPL and ETH market capitalization, in contrast, shows no such spike, but a lot of oscillation and noise around zero.

Agent Simulations

The primary simulation tool utilized by Gauntlet is agent-based simulation (ABS), which consists of defining the incentives and utility functions of a population of agents, setting the terms of their interactions with each other, and then simulating the results. ABS has been used in a variety of contexts including quantitative finance, financial fraud detection, central bank stress testing, and by the Federal Reserve.

Gauntlet defined six types of trader agents. Each of these subpopulations responds differently to outside conditions, which reflects the reality that markets are composed of traders pursuing different strategies at any one point. The six agent profiles are:

  1. Rebase Arbitrage Trader
  2. Mean Reversion Trader (price)
  3. Mean Reversion Trader (market capitalization)
  4. Equilibrium Market Maker Trader
  5. Momentum Trader (price)
  6. Momentum Trader (market capitalization)

The simulation concluded that “the Rebase Arbitrage Trader’s strategy is the only of the modeled trader types that demonstrates significant profitability.” More interestingly, they also find that “Momentum trading, either based on price or market capitalization, does not appear profitable across any of the sampled scenarios.”

They also found that the general drift in AMPL does not significantly affect the profitability of these strategies, meaning that traders should be able to pursue their strategies in a wide range of conditions.

The relative superiority of the Rebase Arbitrage Trader’s strategy means that serious attempts to trade AMPL have to take into account the rebasing dynamic—this is reinforced by the Rebase Arbitrage Trader’s increasing success under conditions of greater volatility.


We think that Gauntlet’s report provides evidence that Ampleforth’s rebasing mechanism introduces novel incentives to its community that do not exist in current-generation cryptocurrencies or traditional assets.

We also think it is possible that as the need to adopt a version of rebase arbitrage becomes clearer, the rebase mechanism might become increasingly effective at keeping AMPL’s market capitalization uncorrelated from the broader digital asset market.


[1] Salvatore, Gu, Xu, Morrow, et al. (2020). “Protocol and Trading Strategy Assessment.” available at:
[2] Selgin, G. (2015). “Synthetic Commodity Money.” Journal of Financial Stability.
[3] Kuo, Iles, Rincon-Cruz, et al. (2019). “Ampleforth–A New Synthetic Commodity” paper available at:
[4] Buchanan, J.M. (1962). “Predictability: the criterion of monetary constitutions.” In:Leland, B.Y. (Ed.), In Search of a Monetary Constitution. Harvard University Press,Cambridge, MA, pp. 155–18