Kevin: checks calender matches the csv for our dataset. build a class derived from this
Anson: on data_loader, fixed some bugs, ensured doc is right.
More robust version of checking the dates
-in practice, we should only remember the latest ticker, finnhub id should not change
On Dataloader, compute_features should be moved to main class, it should not be repeated for each sub-data loader
Further work:
Fundamentals dataset, each json file only contains ticker, need a map to finnhub id (Anson)
(Do a summary)
Features: P/E ratio, EPS ratio.. need to learn formula for financial statements ratio. These are hard to trade upon, more used for stock selection.
Automated web-scrapping from quandl short ratio dataset
Alphavantage - intra-day extended history; minute data for trailing 2-years
both adjusted and unadjusted price for these tickers, can set output to csv.
Unadjusted data - future data can be added straightaway
automate downloads
Data will be downloaded in monthly batches, get it into csv first
Intra-day data is more important for mean-reversion methods in some research
For trading simulator:
Portfolio Dict[str,float] FinnhubID with quantity
Read the data and Return the point-in-time data of stock price history for a particular
generate trades : given the stock price history, generate a list of order you want to make on that day
Rebalance function: given previous days order, rebalance the portfolio
Shu: functions for vol ratio completed
Kevin: checks calender matches the csv for our dataset. build a class derived from this
Anson: on data_loader, fixed some bugs, ensured doc is right. More robust version of checking the dates -in practice, we should only remember the latest ticker, finnhub id should not change
On Dataloader, compute_features should be moved to main class, it should not be repeated for each sub-data loader
Further work:
Fundamentals dataset, each json file only contains ticker, need a map to finnhub id (Anson) (Do a summary) Features: P/E ratio, EPS ratio.. need to learn formula for financial statements ratio. These are hard to trade upon, more used for stock selection.
Automated web-scrapping from quandl short ratio dataset
Alphavantage - intra-day extended history; minute data for trailing 2-years both adjusted and unadjusted price for these tickers, can set output to csv. Unadjusted data - future data can be added straightaway
automate downloads Data will be downloaded in monthly batches, get it into csv first
Intra-day data is more important for mean-reversion methods in some research
For trading simulator: Portfolio Dict[str,float] FinnhubID with quantity Read the data and Return the point-in-time data of stock price history for a particular generate trades : given the stock price history, generate a list of order you want to make on that day Rebalance function: given previous days order, rebalance the portfolio