Monte Carlo Simulation with value at risk (VaR) and conditional value at risk (CVaR) in Python

[@gian_piano]

awesome video. I am even struggling reading your code and understanding it, it is just admirable how you came up with this code

[@TheChievovr]

you are back! Great!

[@sebastianbrudersohn5021]

Very well done, many thanks!

[@louisdesmarestz1919]

Thanks for this great video ! Would be awesome if you could also explain the differences in interpreting Historical vs Parametric vs MC VaR and CVaR. From what I understood they measure the same thing but don't have exactly the same meaning.

[@QuantPy]

Thanks, will definitely make a video summarising this one. Essentially these are three different models to asses the possible distribution of future returns. The difference between the models are the assumptions on the asset distributions. Historical VaR, makes no assumptions about the distribution (uses historical distribution). Parametric VaR uses a distribution defined by you, and is parametrised in terms of mean and covariance matrices. MC VaR, is a flexible methodology, whereby you could combined various assumptions for each asset.

[@anthonypeters7030]

What would we do if we wanted to calculate min, median, max, and avg return from the 100 simulations?

[@2255.]

very helpful thank you

[@Suvra2961984]

Helpful video. Please make a video of calculating VaR using Genetic Algorithm in Python. Thanks!

[@QuantPy]

Great suggestion! Haven't heard of common use cases for Genetic Algorithm (GA) approach for producing a static portfolio. Just having a quick search in the literature, seems like GA can be used to reduce computation time compared to Linear Programming (LP) methods, over large time horizons for allocation problems.

[@Suvra2961984]

@@QuantPy The code isn't working. It says no data could be fetched using YahooDailyReader. Any idea how to make it work? Thanks.

[@QuantPy]

pip install -upgrade pandas-datareader

[@superuser8636]

Gillespie's Algorithm is a great example

[@mathysferriere3987]

@@Suvra2961984 from pandas_datareader import data as pdr import yfinance as yf yf.pdr_override()

[@stevealmond7460]

hello. aren't we assuming returns are normally distributed tho?

[@giulioc.6066]

Thanks for your videos, they are very useful! Could you tell me how to calculate the contribution to the VaR of each individual asset? For example, if the portfolio is made up of 2 assets and the portfolio VaR is 10%, the first asset contributes 3% of VaR to the second asset 7%.

[@QuantPy]

Thanks for your question. The way I would visualise the contribution to VaR as you are describing is the same as major trading/market making companies visualise risk contribution. 1. This is by separating VaR by particular classes, so in your example Asset A and then Asset B. 2. Run VaR separately. 3. Run VaR together. 4. The reduction/increase is called 'Diversification Effect' Please see page 98 on Goldman Sachs 2020 Form 10-K for an example where they have separated VaR into asset classes. www.goldmansachs.com/investor-relations/financials/current/annual-reports/2020-annual-report/multimedia/2020/annual-report-2020.pdf

[@frodewilkensen]

Hi - is there a quick adjustment for me to replicate this code but only for a single security with a specified estimate on volatility?

[@frodewilkensen]

And thank you for the great video!