shinyapp-alphvantagepf

The Alphavantagepf package also contains a Shiny Application which can be used to query, save, and visualize basic market information without having to navigate the asset-specific functions provided by the Alphavantage API. The app provides an intuitive way to compare small baskets of assets both technically and fundamentally.

This vignette will first go through some overall design goals and conventions before providing an overview of how to start and configure the application then give a brief overview of list management and individual analysis components. Augmenting the data is also describes. Future plans are discussed in an appendix.

Overall design goals and conventions

The app is designed to

  • Integrate four of the main asset classes into the analysis. For example, an equity, an index and FX exchange rate can all be plotted together.
  • Minimize the amount of price information downloaded by caching (to the degree possible) older price data.
  • Provide interfaces for capturing any data requested and adding external data to the internal cache.
  • Provide easy ways to create and retrieve baskets of assets.
  • Use modern design elements as much as possible, in particular the gt package and the dygraphs package. The latter is used via a “sister” package FinanceGraphs

A few conventions which are helpful to know before using the app are

Item Convention Example
Asset Sets Semicolon ; delimited "IBM;NDX;USD/MXN"
Relative Dates (Signed) integers followed by [m|d|y], relative to today "-4m"
Date Ranges Relative dates separated by "::" "-4m::", "-1y::+1y"

App invocation and initial setup.

The app requires a very modest amount of setup before using. It can be run with

library(alphavantagepf)
av_runShiny()

When the app is first run, the following screen will be shown, with the AVOPTS tab selected. The first order of business will be to set the Alphavantage API keys. To do so, just type or copy in both your API key and your entitlement status (Either delayed or realtime), and hit the blue Set Opts button.

Other optional items that can be set are

Item Description
Cache Data Directory Directory to store cached price data. If not specified, a temporary directory will be created and used.
Cached data is in a price file (in fst format), and an inventory file
fgts colorset A named aesthetic set for time series line colors, See fg_update_aes()
Regr Significance p-values below which regression terms will be highlighted
AV dump directory A directory in which to create a file with each downloaded function call. See below for details.
Capture AV Data What Alphavantage calls to capture (e.g. Prices, non-prices, or all data)
Update or Cumulative If Update, then captured data will be keyed and updated only if new, otherwise all data captured
For example, Update will only store the latest price for a particular symbol and date
Data Saving Options Control how often the captured data file is cleared or saved

When capturing is turned on a file called av_download.RD will be created and updated in the specified dump directory. This file will be a named list of data.tables, with a data.table containing all downloaded data corresponding to an Alphavantage function (e.g. TIME_SERIES_DAILY). If “cumulative” is selected, all data will time-stamped and added to the relevant data.table. If not, then data will be replaced for each relevant key (usually SYMBOL). The user can then use that file to integrate search histories into other analytics, such as a rolling Markdown file.

Asset List management

Securities are seldom analyzed in isolation. It is easy to create and use ad-hoc groups of securities in this app.

  • To create a list, First, type in a new name for the list in the list box as shown below, and Hit Enter. Second, type in the components (separated by semi-colons) and hit the set button.

  • To get the assets in a list, select the desired list from the dropdown and hit get button.

You can see a table of current asset lists by running the “Gen:Inventory” command, and can add asset lists outside of the app using av_add_assetgroups()

Running Basic Analyses

To run analyses, select the desired item from the dropdowns and press the blue RUN button. One of more graphs or tables will be shown in the MAIN tab, with additional data possible in the DETAIL tab.

Graph events and decorations can be specified in the first two boxes, with annotations for the last level and 5 turning points used as defaults. The options possible are detailed in the fgts_dygraph() documentation. Empty out the boxes for clean graphs. The date ranges plotted come from the “dts” box (see above for conventions), and more recent data can be focused in on using the FocusPct slider. For example, if the slider is set to 50pct, and the date string is "-2y::" then the graph will only show the last year of data, but can be slid back to see the entire 2 year period.

Note that status messages are sent to the console if the “Status Msgs” box is checked and output data for each analysis is copied to the clipboad if the “copyTable” box is checked.

Plotting options

Data is plotted in raw form unless the Rebase options are used. Rebasing at the start will normalize all series to 100 at the start of the window. (In the the example above, the rebase would occur at the first point plotted 2 years prior to today). If Rebasing is set to the focus point, the rebasing occurs at that point set by the slider.

An alternative to rebasing is to plot the first asset on it’s own axis, which can be done by selecting "splitts" in the "TSGraphOpts" box. Other options that can be selected in that box are

  • Highlighting the first series with a thicker line
  • Adding Labels showing the last value or series title
  • Adding hi-low shading, if the data is available.
  • Changing the line colors (via the "fgts colorset" setting in the "AVOPTS" tab.)

Event codes can also be specified in the "events" box. See fg_get_dates_of_interest() and Dygraphs vignette These can be single day events or highlighted ranges of dates. Also, time series that do not appear to be daily will be plotted with step functions.

Analyses that can be run

The table below shows the current analyses that can be run. Select what you wish and hit the RUN button.

Analysis Explanation/Notes
Gen:Inventory Price inventory and Asset Lists
Gen:LivePx Live Price Table
Gen:NameSearch Search for an Equity or Index [1]
TS:PriceTS Time Series of Prices [2],[3]
TS:ActiveTS Time Series of Total Return of Line 1 Assets less Line 2 [4]
TS:HistVolTS Time series of Rolling volatilities and correlations [5]
EQ:DES Table of descriptive data for Line 1 Assets
EQ:News Table of recent news items for Line 1 Assets [6]
EQ:DivEarn Table of Dividends and Earnings for Line 1 Assets
EQ:OptSearch Option Lookup and Search [7] (See Section Below)
Gen:Movers US Market Movers

Notes

1 Due to Alphavantage limitations, the search returns tickers that begin with the search term.

  1. Graphing options are described above. THe graph can interactively be zoomed in on and reset with a double click.
    Two synchronized graphs can be used by selecting this option for both Line 1 and Line 2.

  2. Total return data is used unless the box is unchecked. The data is augmented by live data if the box is checked.

  3. Graphs show active returns, i.e. Total return of each asset in Line 1 less the first asset (typically a benchmark index) in Line 2. Rolling correlations and full period betas are also shown.

  4. Volatilitity parameters in the Histvolparams box are passed to the [TTR::volatility()] function.

  5. Results will be displayed in the OPTIONS tab.

  6. Results will be displayed in the NEWS tab.

As an example, suppose we wish to look at active returns of three popular chip manufacturers against the Nasdaq 100 broad index. Put the tickers of interest in Line 1, and an index in Line 2. (If there are more than one tickers in Line 2, only the first is used.) Select TS:ActiveTS and hit run. The app will get whatever data it needs, typically from the cache, but augmented with most recent (live) data and produce the following output:

Managing data outside of the app.

The Shiny App also has facilities for adding and retrieving time series data externally using other R scripts or documents. All of the downloaded price history, as well as an inventory of prices are stored either in a temporary directory created when the app is first run, or a directory specified in the "AVOPTS" tab. The price data is stored in fst format, which has been the performance winner for several years. Time series data stored in that cache augments that returned by (e.g.) av_get_pf("IBM","TIME_SERIES_DAILY") function wih split and dividend data returned by other av_get_pf() calls. The full set is shown below. Note that a timestamp for each data retrieval is also kept.

symbol timestamp  open  high   low close adjusted_close dividend_amount split_coefficient                  ts origclose
<char>  <IDat>    <num> <num> <num> <num>          <num>      <num>             <num>              <POSc>     <num>
ABBNY 2001-04-06  7.37  7.37  7.27  7.29           7.29          0             0.433 2026-05-31 08:06:48      16.8
ABBNY 2001-04-09  7.32  7.50  7.32  7.50           7.50          0             0.433 2026-05-31 08:06:48      17.3

WAdding data and integrating different asset classes.

The data can also be added to externally using the [av_add_data()] function. Any time series can be added, but not all time series can be guaranteed to be updated by Alphavantage calls. One difficulty with Alphavantages’ API platform is that there are different function calls (and returned data sets) for each major asset class. This app integrates them into one price table, but at the expense of having to know each asset type for the data added.

The app tries to determine what asset class a symbol is to update it using av_get_pf() calls as best as it can. Specifically, each new ticker is checked against current Index lists, cryptocurrency lists, and then with Alphavantage’s "SYMBOL_SEARCH". If several tickers are passed at once, this can be quite expensive, as Alphavantage tends to limit data usage if too many calls are made in succession. If possible, it’s much better to predetermine the asset type (one of c("Equity","ETF","Index","FX","Crypto")) in advance and pass it to [av_add_data()] call at runtime.

Data which is completely exogenous to Alphavantage can also be added to the app’s cached data set. If the asset type isn’t specified or cannot be determined, it’s assumed to be a "user" asset type which isn’t updated, but can still be plotted or used. As an example, suppose we wish to plot the prices of a common credit ETF against FRED’s FEDFUNDS data. We will download both and add them programmatically. Generic data needs to be manipulated into the form the app expects, but data downloaded from Alphavantage is already in that form.

# av_add_data can be used simply, but
av_add_data(av_get_pf("IBM","TIME_SERIES_DAILY_ADJUSTED"))

# For many assets, use this
asset_df <- data.frame(symbol=c("HYG"),type=c("ETF"),currency=c("USD"), name=c("HY ETF"))
av_add_data(av_get_pf("HYG","TIME_SERIES_DAILY_ADJUSTED"), assettypes=asset_df)

# External Data
suppressMessages(require(quantmod))
ffdta <- as.data.table(quantmod::getSymbols("FEDFUNDS",src="FRED",auto.assign=FALSE))
ffdta <- ffdta[,.(DT_ENTRY=index,close=FEDFUNDS,adjusted_close=FEDFUNDS,symbol="FEDFUNDS")]
av_add_data(ffdta)

dump_inv() |> gt()

The data added can be plotted using the "TS:PriceTS" analytic, as seen below. Note that the monthly time series needs to be used with care, since it doesn’t correspond to the actual Fed Meetings.

Future plans

Planned improvements currently include

  • Earnings and dividends incorporated into plots including E/P analysis.
  • Ability to plug in user-defined functions which have access to the internal data store.
  • Correlation and hedging analyses.
  • Interface to events and aesthetics sets relevant to FinanceGraphs