Skip to content

Latest commit

 

History

History
235 lines (158 loc) · 6.63 KB

README.md

File metadata and controls

235 lines (158 loc) · 6.63 KB
Raw

Monte Carlo Simulations for Value at Risk

This tutorial demonstrates how to run Monte Carlo simulations on Google Cloud's Batch and how to analyze the data using Vertex AI Notebooks.

Monte Carlo simulation is a mathematical technique that can be used to estimate the value at risk (VaR) of a financial portfolio. The basic idea is to generate a large number of random price paths for the assets in the portfolio, and then calculate the VaR as the minimum loss that would be expected to occur with a certain probability.

  1. Choose a time horizon and a confidence level.
  2. Generate a large number of random price paths (random walks) for the assets in the portfolio.
  3. Calculate the portfolio value at the end of each price path.
  4. Determine the VaR as the minimum portfolio value that occurs with the specified confidence level.

Cost to run

The following elements of this tutorial will result in charges to your billing account. Please validate with the Google Cloud Pricing Calculator

  • Batch
  • Pub/Sub
  • BigQuery
  • Vertex AI Notebooks

Tutorial architecture

The overall structure of this tutorial is as follows:

Basic getting started

  1. Get the Cluster Toolkit configured.

Build the gcluster binary:

git clone https://github.com/GoogleCloudPlatform/cluster-toolkit
cd cluster-toolkit
make
./gcluster --version
./gcluster --help

2. Run gcluster on the blueprint fsi-montecarlo-on-batch.yaml

./gcluster create community/examples/fsi-montecarlo-on-batch.yaml \
   --vars "project_id=${GOOGLE_CLOUD_PROJECT}"

Where GOOGLE_CLOUD_PROJECT has been set via an export command

export GOOGLE_CLOUD_PROJECT=my_project_id

If successful, you will see output similar to:

To deploy your infrastructure please run:

./gcluster deploy fsimontecarlo

3. Deploy the blueprint as instructed:

./gcluster deploy fsimontecarlo

If successful, this will prompt you:

Summary of proposed changes: Plan: 22 to add, 0 to change, 0 to destroy.
(D)isplay full proposed changes,
(A)pply proposed changes,
(S)top and exit,
(C)ontinue without applying
Please select an option [d,a,s,c]:

Respond with apply, "a". You may be asked to respond twice.

When the job is complete it will indicate Succeeded. You can then proceed to the next section.

At this point, all the required infrastructure has been deployed.

Open Vertex AI Workbench

1. Go to the Vertex AI Workbench Notebooks instances in the Google Cloud Console:

https://console.cloud.google.com/vertex-ai/workbench/user-managed

2. Open JupyterLab on the Notebook instance listed.

Click on `OPEN JUPYTERLAB` link

3. In the JupyterLab UI, you will see a list of directories:

Select `fsi`

Under fsi all the files required to run the demo have been pepared.

4. Open a terminal window by clicking on the terminal icon.

5. Update the local Python requirements:

python3 -m pip install --user -r requirements.txt

There may be some incompatibilities listed, but it will not affect this demo.

6. Run the batch.py Python script to ensure it is working.

python3 batch.py --help

You will see a listing of the help messages

7. To start the VaR simulation, run batch.py with the config file and --create_job

python3 batch.py --config_file mc_run.yaml --create_job

You should see output without any errors, listing information about the job.

To see if the job is running, use the --list_jobs options.

python3 batch.py --config_file mc_run.yaml --list_jobs

If you want to see the jobs listed in the Cloud Console, you can click:

https://console.cloud.google.com/batch/jobs

View the data in BigQuery

The batch job runs Monte Carlo simulation for the VaR calculation. The output from each run is stored in BigQuery. To view this data in it's raw form, you can view BigQuery in the Cloud Console:

https://console.cloud.google.com/bigquery

Navigate to the fsi_table.

There you can see the schema.

To see the data, click on PREVIEW

For an advanced user, you can run queries directly in the BigQuery UI.

Visualization in the Notebook

Finally, you can select the FSI_MonteCarlo.ipynb from the left navigation in the JupyterLab window.

To run the cells in the notebook, select the cell, then click the play button, or Alt-Enter.

Run the cells in the order they appear.

After running the second cell, you should see output.

Finally, when you run Cell #4, you will see graphs and table summaries.

Summary

In this tutorial, you accomplished the following:

  • You created Cloud infrastructure
    • Vertex AI notebooks
    • BigQuery Tables
    • Pubsub BigQuery Subscription
    • Batch jobs
  • You ran a MonteCarlo simulation for VaR on several stock tickers.
  • You reviewed the data in BigQuery
  • You visualized the data in Vertex AI notebooks.

Shutting down

The best way to clean up your workspace is to delete the project. This will ensure you are not billed for any of the Cloud usage.

Alternatively

The other choice is to run a gcluster destroy command.

./gcluster destroy fsimontecarlo