Lucidworks Fusion Financial Services Demo

This demo showcases Fusion's capabilities for financial services use cases, particularly for analyst and advisor personas. The demo intersects news, publicly available historical equity data, and firm-specific portfolio and client data.

Help on this Demo

Help on this Demo is available in Slack in the #finance-demo channel.

Prerequisites

• Your system should have a minimum free memory of 8-12 GB with a 64-bit processor.
• Enough free disk space to support the writing of index files etc.
• JDK8
• Node/NPM (if you want to add new node modules)
• The zip/unzip utility
• A valid Fusion license file, probably stored in your home directory
• Fusion 4.1.1
• You must be running Fusion's webapps service
• Maven
• External web access to Yahoo Finance
• External web access to the IEX API. Validate that https://api.iextrading.com/1.0/stock/AAPL/chart returns historical prices for Apple.
• A Google Maps API key that you put in app/src/main/resources/conf/app-config-private.conf, with contents googleMapsKey: <YOUR API KEY>.

Setup

Clone the fusion-financial-demo repo

git clone git@github.com:lucidworks/fusion-financial-demo.git

One Time Setup Instructions

• Check that you have https://lucidworks.slack.com/files/U92QXC46S/F9QSN33PG/settings.xml in ~/.m2/settings.xml

Installation

Setup myenv.sh

If you're using a default Fusion 4.x setup, then you can skip this step and pass the path to Fusion (e.g. /opt/fusion/4.1.1) to install.sh, which will create a myenv.sh script with the default settings.

However, if you're using a customized Fusion installation, or are not using the default password, etc, then cp myenv.sh.tmpl myenv.sh and then edit ./myenv.sh to verify all env vars are correct for your environment.

Run

1. Run ./install.sh
2. Run ./redeploy-data.sh
3. Browse to http://localhost:8764/ for the Fusion Admin UI
4. Go to the Datasources and run all. Let the yahoo-news datasource run for a long time. The sp500 and sp500-historical datasources take 1-2 minutes, and the rest are quick.
5. Browse to http://localhost:8780/Finance/login/ for the end user UI

Application Overview

There are three personas supported in this app:

1. Portfolio Manager (users Chloe and Lawson). Manages investment strategy portfolios. Differentiators: portfolio holdings view
2. Sellside Analyst (users Geoffrey and Frank). Covers companies (writes research, issues guidance and earnings estimates). Similar to PM, but no notion of portfolios with weighted holdings.
3. Financial Advisor (users Robert and Cynthia). Covers client relationship, aids in asset allocation / product decisions. Has more client-centric views available, e.g., a Clients tab that allows them to view clients and their accounts.

Development

Most demo development should either be on the Application itself in Fusion (either via the UI or the API) or in the Appkit UI, both of which are covered below.

Updating the UI

Develop the webapp locally:

1. cd app
2. mvn jetty:run

And then redeploy as a Fusion webapp:

1. Run ./redeploy.sh

Updating the Application

1. Make your changes in the Admin UI or via the APIs or by directly editing the objects.json file`.
2. ALL CHANGES TO THE BACKEND APPLICATION MUST BE EXPORTED AND CHECKED IN
3. To Export, run ./export.sh and then review and commit/push to github
4. You probably want to coordinate with whomever else is working on this application, as with other code, if you are both working on the same area, you will get conflicts.

If you wish to redeploy the backend application, run:

1. ./clean.sh -- note, this will delete the application and data (it will first export it to /tmp)
2. ./install.sh (same as above)

Working with Data

This app relies on generated data for clients and portolios. See the Python scripts in /scripts. The scripts will generate these files: company_buckets.csv which includes portfolio holdings, clients.csv, and client_accounts.csv. The latter two files are used to support the financial advisor persona.

1. If you put the data in data/Finance, then it will automatically show up at http://localhost:8780/data/ and can easily be accessed via a Web data source.
2. If the data is less than 100 MB, you can just check it in.
3. If it is greater than 100 MB, use Nexus or S3 or GDrive.
   1. Be sure to add downloading it to prerequisites.sh if it is in Nexus or S3
4. ./redeploy-data.sh will repackage and redeploy the data
5. Be warned that clean.sh will delete any data you've indexed. There are tricks to avoiding this, but they are involved:
   1. The following steps are unverified:
      1. Shutdown Fusion
      2. Copy the data directories out of Fusion
      3. Restart Fusion
      4. Delete/clean your application
      5. Shutdown Fusion
      6. Copy the data directories back into Fusion
      7. Restart Fusion
   2. NOTE: This assumes your application hasn't changed the way it processes the data for indexing.

Parquet file dumping and loading

Sometimes it is more efficient to load and dump data from Parquet (https://parquet.apache.org/) files. For instance, if you want to fairly quickly dump out all the records of an index to parquet, you can use the dump_parquet.sh script to do so and then the load_parquet.sh to load it back in. Before running them, however, be sure to edit the underlying Scala files (dump_solr_to_parquet.scala and load_parquet_data.scala) to properly point at the collections you want to dump/load.

##Troubleshooting

Mismatched Fusion Versions

The setup/app/objects.json is tied, more or less, to a specific version of Fusion. If install.sh fails on the install and says something about mismatched versions, you may need to change this value.

Application or UI is already installed

Run clean.sh first then re-install.