Description

Learn Computer and Data Science using Algorithmic Trading.

Description

Provide sets of functions and methods to learn and practice data science using idea of algorithmic trading. Main goal is to process information within "Decision Support System" to come up with analysis or predictions. There are several utilities such as dynamic and adaptive risk management using reinforcement learning and even functions to generate predictions of price changes using pattern recognition deep regression learning. Summary of Methods used: Awesome H2O tutorials: <https://github.com/h2oai/awesome-h2o>, Market Type research of Van Tharp Institute: <https://vantharp.com/>, Reinforcement Learning R package: <https://CRAN.R-project.org/package=ReinforcementLearning>.

README.md

cran.r-project.org

lazytrade

The goal of lazytrade is to keep all functions and scripts of the lazytrade educational project on UDEMY. Functions are providing an opportunity to learn Computer and Data Science using example of Algorithmic Trading. Please kindly not that this project was created for Educational Purposes only!

Installation

You can install the released version of lazytrade from CRAN with:

install.packages("lazytrade")

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("vzhomeexperiments/lazytrade")

Several ideas explored in this package

Data manipulation and analysis of performed trades results
Reinforcement Learning for Automated Trading Risk Management
Data manipulation and preparation for Machine Learning (transposing, aggregation, lagging, etc)
Using Deep Learning for prediction of Market Types (Classification)
Using Deep Learning for prediction of future price change (Regression)
Strategy Tests simulations
Utility functions to generate passwords, initialization files, encryption of passwords, etc
Explored idea of building a model using random structures combined with an automated functional (strategy) test to improve model performance
Overall, all functions have working examples with relevant documented sample data included in the package

Example - prepare data for machine learning

This is a basic example which shows you how to solve a common problem:

library(lazytrade)
library(magrittr, warn.conflicts = FALSE)
## basic example code
# Convert a time series vector to matrix with 64 columns
macd_m <- seq(1:1000) %>% as.data.frame() %>% to_m(20)

head(macd_m, 2)
#>      [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13] [,14]
#> [1,]    1    2    3    4    5    6    7    8    9    10    11    12    13    14
#> [2,]   21   22   23   24   25   26   27   28   29    30    31    32    33    34
#>      [,15] [,16] [,17] [,18] [,19] [,20]
#> [1,]    15    16    17    18    19    20
#> [2,]    35    36    37    38    39    40

Why is it useful? It is possible to convert time-series data into matrix data to do make modeling

Example - aggregate multiple log files and visualize results

Multiple log files could be joined into one data object

library(lazytrade)
library(readr)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(magrittr)
library(lubridate)
#> 
#> Attaching package: 'lubridate'
#> The following objects are masked from 'package:base':
#> 
#>     date, intersect, setdiff, union

# files are located in the sample folders
DFOLDER <- system.file("extdata/RES", package = "lazytrade")

DFR <- opt_aggregate_results(path_data = DFOLDER)

This data object can be visualized

library(ggplot2)
opt_create_graphs(x = DFR, outp_path = dir,graph_type = 'bars')

Or just visualize results with time-series plot

opt_create_graphs(x = DFR, outp_path = dir,graph_type = 'ts')

Example - leverage Reinforcement Learning for Risk Management

Example below would generate RL policy based on the trade results achieved so far

library(dplyr)
library(ReinforcementLearning)
library(magrittr)

data(data_trades)
states <- c("tradewin", "tradeloss")
actions <- c("ON", "OFF")
control <- list(alpha = 0.7, gamma = 0.3, epsilon = 0.1)
rl_generate_policy(data_trades, states, actions, control)
#>           TradeState Policy
#> tradeloss  tradeloss     ON
#> tradewin    tradewin    OFF

Example - generating passwords for trading platforms login

Multiple trading accounts require passwords, package contains function that may easily generate random passwords:

library(lazytrade)
library(stringr)
library(magrittr)
library(openssl)
library(readr)

#generate 8digit password for trading platform
util_generate_password(salt = 'random text')
#>          .
#> 1 ac5cE049

Example - generate initialization files for MT4 platform

Facilitate generation of initialisation files:

library(lazytrade)

dir <- normalizePath(tempdir(),winslash = "/")

# test file to launch MT4 terminal with parameters
write_ini_file(mt4_Profile = "Default",
               mt4_Login = "12345678",
               mt4_Password = "password",
               mt4_Server = "BrokerServerName",
               dss_inifilepath = dir,
               dss_inifilename = "prod_T1.ini",
               dss_mode = "prod")

Notes to remind myself how to create R package

This readme file

What is special about using README.Rmd instead of just README.md? You can include R chunks like so:

summary(cars)
#>      speed           dist       
#>  Min.   : 4.0   Min.   :  2.00  
#>  1st Qu.:12.0   1st Qu.: 26.00  
#>  Median :15.0   Median : 36.00  
#>  Mean   :15.4   Mean   : 42.98  
#>  3rd Qu.:19.0   3rd Qu.: 56.00  
#>  Max.   :25.0   Max.   :120.00

You’ll still need to render README.Rmd regularly, to keep README.md up-to-date.

taken from https://r-pkgs.org/

Communicate about lifecycle changes

taken from https://lifecycle.r-lib.org/articles/communicate.html

Run Once:

usethis::use_lifecycle()

To insert badge:

Add badges in documentation topics by inserting one of:

#’ #’ #’

Generating Documentation

Title of the package

Create right title case for the title of the package By running this command… tools::toTitleCase("Learn computer and data science using algorithmic trading") the Title will become: “Learn Computer and Data Science using Algorithmic Trading”

Re-generating documentation

Run this code to re-generate documentation devtools::document()

Fixing License

Run this code to fix license: usethis::use_mit_license(name = "Vladimir Zhbanko")

Adding data to the package for internal tests

Run this code to add data to the folder data/x <- sample(1000)usethis::use_data(x)

To update this data: x <- sample(2000)usethis::use_data(x, overwrite = T)

To convert character into time: mutate(across('X1', ~ as.POSIXct(.x, format = "%Y.%m.%d %H:%M:%S")))

Note: use option ’LazyLoad` to make data available only when user wants it always include LazyData: true in your DESCRIPTION. Note: to document dataset see https://stackoverflow.com/questions/2310409/how-can-i-document-data-sets-with-roxygen

Document dataset using the R script R/datasets.R

Use data in the function with data(x)

Adding files to the package

Place data like small files to the folder: inst/extdata

Adding examples to test package function

Tests setup first time

Run this command to setup tests ‘usethis::use_testthat()’

This will create a folder with the name tests

Inside this folder there will be another folder testthat.

Examples in Roxygen code

@examples …

code to execute during package checks

@examples

/donttest{

code to NOT execute during package checks

}

Testing a package

Create a test script

Run this command to create a new script with the test skeleton:

usethis::use_test("profit_factor.R")

Enrich the test script

Details:

add libraries used for test
add function context("profit_factor")
add function test_that(“test description”, {test process})
load data using function data(named_data_object)

Example:

library(testthat)
#> 
#> Attaching package: 'testthat'
#> The following object is masked from 'package:dplyr':
#> 
#>     matches
#> The following objects are masked from 'package:readr':
#> 
#>     edition_get, local_edition
#> The following objects are masked from 'package:magrittr':
#> 
#>     equals, is_less_than, not
library(dplyr)
library(magrittr)

context("profit_factor")

test_that("test value of the calculation", {

  data(profit_factor_data)

  DF_Stats <- profit_factor_data %>%
    group_by(X1) %>%
    summarise(PnL = sum(X5),
              NumTrades = n(),
              PrFact = util_profit_factor(X5)) %>%
    select(PrFact) %>%
    head(1) %>%
    pull(PrFact) %>%
    round(3)

  expect_equal(DF_Stats, 0.68)

})
#> Test passed 🎊

Test of the coverage for the script

Test coverage shows you what you’ve tested devtools::test_coverage_file()

devtools::test_coverage_file()

Automated checks

This will add automatic test coverage badge to the readme file on github usethis::use_coverage()

Checking package

Step 1. devtools::document() Step 2. devtools::run_examples() Step 3. Menu ‘Build’ Clean and Rebuild Step 4. ‘Check’ devtools::check()

Locally checking package with –run-donttest enabled

This is now a default option

Whenever examples construct is used author of the package must insure that those examples are running. Such examples are those that would require longer test execution. To perform this test package needs to be checked with the following command:

devtools::check(run_dont_test = TRUE)

whenever a quick check is required:

devtools::check(run_dont_test = FALSE) ???

Handling functions that write files

In case functions are writing files there are few considerations to take into account:

examples section must contain working example of code that writes files
example code must write to the temporary directory defined by tempdir() function
after package check performed with devtools::check() there should nothing remain in the ‘tmp/’ directory

Considerations

File names defined by function tempdir() would look like this:

# > tempdir()
# [1] "/tmp/RtmpkaFStZ"

File names defined by function tempfile() would look like this:

# > tempfile()
# [1] "/tmp/RtmpkaFStZ/file7a33be992b4"

This is example of how function write_csv example works:

tmp <- tempfile()
write_csv(mtcars, tmp)

results of this code are correctly stored to the temporary file

however this example from readr package function write_csv is showing that file will be written to the ‘/tmp/’ directory

dir <- tempdir()
write_tsv(mtcars, file.path(dir, "mtcars.tsv.gz"))

Deleting files after running examples:

We use function unlink() to do this:

unlink("/tmp/*.csv", recursive = TRUE, force = TRUE)

and we check that there is nothing more remained:

dir("/tmp/*.csv")

Delete deprecate functions

To remove function from the package we can use:

Delete function from the folder (careful not to break other functions)
Build -> More -> Document
Build -> More -> Clean and Rebuild
Build -> Check

CRAN Submission Tips and Tricks

Many notes while using global variables:

see see <https://github.com/HughParsonage/grattan/blob/master/R/zzz.R

Unfortunate note on specific flavors

After first submission there are some notes on specific R flavors

This question was addressed here but yet it’s not answered: https://stackoverflow.com/questions/48487541/r-cmd-check-note-namespace-in-imports-field-not-imported

To search for specific function in the scripts one can do the following:


list_of_functions <- c(
  "drop_na",
  "fill",
  "extract",
  "gather",
  "nest",
  "separate"
)

for (FN in list_of_functions) {

if(!exists("res")){
 res <- BurStMisc::scriptSearch(FN)  
} else {
  res2 <- BurStMisc::scriptSearch(FN)  
  res <- mapply(c, res, res2, SIMPLIFY=FALSE)}
}

Define min R version

https://stackoverflow.com/questions/38686427/determine-minimum-r-version-for-all-package-dependencies

When functions are writing to the file

It’s important to avoid that function write to the directory other then tempdir() Construct file name must be done using file.name() function as follow:

# use plane temp directory
dir_name <- normalizePath(tempdir(),winslash = "/")
file_name <- paste0('my_file', 1, '.csv')
# this needs to be used in the function
full_path <- file.path(dir_name, file_name)

# when using sub-directory
sub_dir <- file.path(dir_name, "_SUB")
if(!dir.exists(sub_dir)){dir.create(sub_dir)}

Versioning of the package

https://r-pkgs.org/description.html#version

Test Environments

Clone package from GitHub and test check it in Docker Container

started docker container vladdsm/docker-r-studio
new project package
clone from vzhomeexperiments/lazytrade.git
use check button to pass the test

Build package

devtools::build()

Adding Readme Rmd

usethis::use_readme_rmd()

Automatic check with GitHub Actions

usethis::use_github_action()

To be elaborated

Upload package to CRAN

Setup the new version of the package:

usethis::use_release_issue()

Follow checklist before upload to CRAN:

devtools::release_checks()

then:

devtools::release()

before release checks

spelling devtools::spell_check()

checking on R hub

See ?rhubv2

checking with release

devtools::check_win_release()

checking win devel

devtools::check_win_devel()

checking win old devel

devtools::check_win_oldrelease()

check with rocker R in container

use docker image with R Studio,
clone repo, build, check package…

Update news.md file

Explain the changes

uploading the package archive to CRAN

https://cran.r-project.org/submit.html.