seamlik / ibkr-toy

Toolkit for playing with my account on Interactive Brokers

IBKR Toy: Toolkit for playing with my account on Interactive Brokers

The main feature right now is to rank all my stocks with a simple (and probably stupid) algorithm. It also generates a table that help me decide which stocks to invest in once I get my salary.

ibkr-toy's Issues

Check existence of Node.js before running it

Otherwise, the error message at the moment is now intuitive.

First request to IBKR market data always returns empty

This behavior is written in the document already, we just did not notice it.

Add weight to each rankers

• PEMA 200: 0.4
• PEMA 20: 0.3
• P/E: 0.2
• Dividend yield: 0.1

Report column header names in Chinese

Because the whole table is getting too wide.

Long-term change column does not align with what's shown on Yahoo Stocks

It might be even better to remove this column entirely. But first, check if other columns have correct data, and search for some other indexes that measure the long-term success of a stock.

Long term change period set to 5 years

It is only fair because getting data from some stocks (e.g. Swiss ones) always fails, and I must manually get their data from Yahoo Stocks mobile app, which limits the long-term price change to 5 years.

Also, setting the scope to too long will give some stocks an excessively high score while other stocks will never see the light of day.

Unit test on summing of factors into scores for a single stock

Find more stock measurements to replace the short-term and long-term price change in ranking factors

Rationale of replacement

At the moment, when we give a score to a stock, we these 2 ranking factors: short-term and long-term price change. However, they pose some serious problems.

First of all, to obtain these measurements, we need to download the historical market price of a stock on the day 1 month ago and 5 years ago. Unfortunately, IBKR does not provide an API to download the data on a particular day. Instead, we need to download all data from a particular day to the current day. This adds complexity in API usage and calculation. What's worse, in order to get a precise calculation, we must use a short interval (like 1 day) when downloading data for 5 total years, which results in a huge increase in download time (and might hit usage limit one day).

Secondly, I accidentally found that the calculation of long-term price change is incorrect when compared to Yahoo Finance (#5).

It is better to get rid of these 2 measurements and opt for ones that are already provided by IBKR.

Candidates of replacement

All these measurements are provided by IBKR:

• Exponential Moving Average