Has anyone attempted to build a monte carlo simulation using Coda?
I haven’t, but would be very interested to see if anyone has! Technically I don’t think it would be too difficult but I suspect the sticking point will be storing data, e.g. if you want a table with a row for each iteration, then that could be one massive table. Can Coda cope with 100,000s of rows?
I was wondering if this would be a good advantage.
Monte Carlo simulations - usually - require quite a lot of computational power and I don’t think this is the main benefit of Coda.
What is the use-case you were thinking about?
Hi, @Federico_Stefanato, you may be right. I am using Coda to build flexible financial models. The advantage over excel is that I can make changes to the model drivers without breaking the model more easily than in excel. Plus, my models are available through any web browser and my mobile devices.
The use-case for the monteo carlo simulation is forecasting financial statement data within a specified range of potential outcomes. I have seen something similar in excel with 5,000 simulations (link to excel model here: https://github.com/ARKInvest/ARK-Invest-Tesla-Valuation-Model/blob/master/Tesla%202024%20Valuation%20Extract_3.25.20_v4.1.xlsx).
I could run this in python and use the API to feed the data back to Coda if the computational power is a gating item.
Not exactly what you asked for, but might be a starting point: https://coda.io/@brandon-trew/template-time-series-scenario-modeling
You could automatically create pretty large timeseries outputs like this, and you could use random numbers in the formula inputs (columns in the time series) to simulate those kinds of random events.