TokenByte

We're investigating the correlation between staking and project value, using this relationship to predict and set benchmarks for new staking patterns and reward systems. The initial model relies on polynomial regression, with possibilities for enhancement via more complex statistical methods or machine learning, up to deep learning. Our metric for comparing reward mechanisms is the AUC over time, which can be adjusted using historical data to cater to specific projects. Currently, we're using Balancer V2's circle staking data as our proof-of-concept, comparing three unique staking mechanisms provided by the founding team.

The goal of our simulation is to identify which design best maintains staker stability amidst protocol revenue fluctuations. We've outlined three possible staking scenarios:

1. Alternative 1: A staker's power is determined by the number of locked tokens and the remaining lock duration. For example, if a holder stakes 100 $BRNT for five years, they initially gain the power of 175 (100 * 1.75). After one year, the power decreases to 150 as the multiplier drops to 1.5. In the following year, the power further decreases to 130 as the multiplier reduces to 1.3.

2. Alternative 2: A staker's power is calculated based on the number of locked tokens, remaining lock duration, and current staking time. For instance, if a holder stakes 100 $BYNT for four years, they initially gain the power of 100. After a set period (e.g., 3 or 6 months), the multiplier increases to 1.15, giving the staker a power of 115. This continues until the maximum multiplier is reached, after which the power decreases annually.

3. Alternative 3: A staker's power is determined by the number of locked tokens and the current staking time. As an example, if a holder stakes 100 $BRNT, they initially gain no power during a three-month warm-up period. After this period, their power becomes 100, and the multiplier increases annually (e.g., 1.15, 1.3, 1.5, 1.75) for up to 10 years, with a maximum multiplier of 3.5. When the staker decides to unstake, their power reduces to 0, and they must wait a three-month cooldown period before they can withdraw their $BRNT.

In this Proof of Concept (PoC) project, we utilized a range of tools and technologies to gather, analyze, and present staking data. Specifically, we used AirStack to obtain the desired project staking data and CoinGecko to access additional data.

To analyze and visualize the data, we used Jupyter notebook, which allowed us to write and execute code, document our work, and create interactive dashboards. To share our work with others, we hosted the notebook using Jupyter Binder, which enables users to run and interact with the code online.

To perform our data analysis, we utilized a variety of Python's data science and statistics libraries, such as Pandas, NumPy, and SciPy, as well as visualization libraries. And, of course, we also relied on good old-fashioned mathematics to help us draw insights from the data.