Workstream 1: Ethereum stochastic processes - ETH price

[BenSchZA]

• [ ] Efficiently research the best way to fit a stochastic process for different ETH price scenarios (best case, worst case, base case)
  • Determine the best distributions available for selection
• [ ] Design and implement a mechanism for historical price integration from external API source
• [ ] Design a mechanism for applying ETH price shocks of different forms, to the stochastic process, historical price, and stable price scenarios

[BenSchZA]

We brainstormed investigating the following points:

• Detrending of time-series
• Long-tailed distributions
• Fitting of historical data

[BenSchZA]

From @danlessa:

The quickest way of having samples for ETH price forecast is by using FbProphet, which uses ARIMA under the hoods: https://facebook.github.io/prophet/ It is as simple that it could get

If we want to be more rigorous, I would think about having a stochastic process that makes uses of long-tailed brownian motion, but I'm not sure if there's a ready toolkit for doing that (most people just assume a common GBM) One pointer is that scientific article, which would need to be translated into more actionable items: https://www.tandfonline.com/doi/abs/10.1080/03610920601001766?journalCode=lsta20

If we want to stick with vanilla GBM, this is a good pointer: https://towardsdatascience.com/simulating-stock-prices-in-python-using-geometric-brownian-motion-8dfd6e8c6b18

[BenSchZA]

https://facebook.github.io/prophet/

Prophet is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series that have strong seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend, and typically handles outliers well.