Motivation
Cryptocurrency is a digital currency that is created and managed through the use of cryptography, which is an advanced type of encryption technique. Bitcoin was one of the earliest forms of cryptocurrency and since then, many other types of cryptocurrencies have been created. I first heard about Bitcoin back in 2013, when I was a financially struggling college student, and I thought 80 Dollars per coin was a bit too much. Today, one bitcoin is about 100 times that amount.
Below is the index for November 19th 2017 from https://www.coindesk.com
Today’s Open: $8,033.94 Today’s High: $8,049.12 Today’s Low: $8,021.33 Today’s Closed: $8,034.42
*Update: Since the time this proposal was written, Bitcoin has increased dramatically. Below is the index for December 9th 2017 from https://www.coindesk.com
Today’s Open: $16,057.15 Today’s High: $16,291.68 Today’s Low: $15,538.25
Even though Bitcoin seems like a great investment, many investors remain skeptical. The main reason is bitcoin and other cryptocurrencies are virtual currencies are not backed by anything except for a select few. For Bitcoin, there is limited amount of bitcoin (20 million) and since the market purely fluctuates on demand, it is extremely volatile. This gives spectators reason to believe it is a bubble. Currently, many new investors, attracted by its potential, are entering the market, increasing its value significantly, but this can change if the demand drops. Currently there are many theories out there, but for now, I wanted to see how volatile bitcoin is compared to other currencies.
Currencies
Bitcoin: Created in 2009, was the first decentralized cryptocurrency. 20 million coins
Ethereum: Launched in 2015, Ethereum is a decentralized software platform that enables SmartContracts and Distributed Applications (DApps) to be built and run without any downtime, fraud, control or interference from a third party.
Litecoin: Created in 2011 by engineer Charlie Lee to be the silver to bitcoin’s gold. One of the main disparities between the two cryptocurrencies lies in their transaction speeds.
Bitcoin Cash: Hard fork of the cryptocurrency bitcoin, the fork occurred on August 2017. a hard fork is a radical change to the protocol that makes previously invalid blocks/transactions valid (or vice-versa), and as such requires all nodes or users to upgrade to the latest version of the protocol software.
Dash: An open source peer-to-peer cryptocurrency that aims to be the most user-friendly and most on-chain-scalable cryptocurrency in the world. It offers instant transactions and private transactions.
Tether: Backed by US Dollars, offers a way to own and move fiat currency across different cryptocurrencies and exchanges without the need to convert crypto assets into dollars.
Ripple: A privately held, cash flow positive company that aims to create and enable a global network of financial institutions and banks to use the Ripple software to lower the cost of international payments.
Obtain the Data
Bitcoin API from Quandl
Since bitcoin is the most widely known cryptocurrency, there is a lot of data available for it. I was able to find an API for Bitcoin’s historical data.
#Daily Bitcoin exchange rate (BTC vs. USD) on Bitstamp from 09/13/2011 to 12/09/2017
Quandl.api_key('mxcqtzYcf5Co4fbG3WAX')
Bitcoin <- Quandl('BCHARTS/BITSTAMPUSD', start_date='2011-01-01', end_date ='2017-12-07')
head(Bitcoin )
## Date Open High Low Close Volume (BTC)
## 1 2017-12-07 13623.00 16615.62 13085.90 16599.99 25787.677
## 2 2017-12-06 11676.99 13700.00 11659.80 13623.50 19784.873
## 3 2017-12-05 11613.07 11850.00 11384.25 11677.00 11875.034
## 4 2017-12-04 11250.00 11613.07 10850.00 11613.07 13621.482
## 5 2017-12-03 10875.68 11800.01 10513.16 11250.00 14238.526
## 6 2017-12-02 10840.45 11200.00 10637.69 10872.00 9267.161
## Volume (Currency) Weighted Price
## 1 382694044 14840.19
## 2 250560790 12664.26
## 3 138370076 11652.18
## 4 154122918 11314.70
## 5 160176290 11249.50
## 6 101270084 10927.84
tail(Bitcoin)
## Date Open High Low Close Volume (BTC) Volume (Currency)
## 2273 2011-09-18 4.87 4.92 4.81 4.92 119.81280 579.8431
## 2274 2011-09-17 4.87 4.87 4.87 4.87 0.30000 1.4610
## 2275 2011-09-16 4.82 4.87 4.80 4.85 39.91401 193.7631
## 2276 2011-09-15 5.12 5.24 5.00 5.13 80.14080 408.2590
## 2277 2011-09-14 5.58 5.72 5.52 5.53 61.14598 341.8548
## 2278 2011-09-13 5.80 6.00 5.65 5.97 58.37138 346.0974
## Weighted Price
## 2273 4.839576
## 2274 4.870000
## 2275 4.854515
## 2276 5.094272
## 2277 5.590798
## 2278 5.929231
Main Competitors
The second tier of cryptocurrencies in terms of popularity include Ethereum, Litecoin. I was interested in those two, plus a couple more, Dash and Bitcoin cash. I was able to find a csv file on the historical data from the website: https://coinmetrics.io/data-downloads/
I saved the csv files in Github.
Bc_cash <- read.csv("https://raw.githubusercontent.com/mikegankhuyag/607-Projects/master/Final/bch.csv", header = TRUE, stringsAsFactors = FALSE)
Dash <- read.csv("https://raw.githubusercontent.com/mikegankhuyag/607-Projects/master/Final/dash.csv", header = TRUE, stringsAsFactors = FALSE)
Ethereum <- read.csv("https://raw.githubusercontent.com/mikegankhuyag/607-Projects/master/Final/eth.csv", header = TRUE, stringsAsFactors = FALSE)
Litecoin <- read.csv("https://raw.githubusercontent.com/mikegankhuyag/607-Projects/master/Final/ltc.csv", header = TRUE, stringsAsFactors = FALSE)
head(Bc_cash)
## date txVolume.USD. txCount marketcap.USD. price.USD.
## 1 7/31/2017 2406864986 183998 0 294.46
## 2 8/1/2017 906913244 230867 0 380.01
## 3 8/2/2017 603435293 76537 6302360000 452.66
## 4 8/3/2017 83677447 7416 7392030000 364.05
## 5 8/4/2017 218502200 20909 5969720000 233.05
## 6 8/5/2017 263414318 26517 3809330000 213.15
## exchangeVolume.USD. generatedCoins fees
## 1 1075960 1837.5 138.116822
## 2 65988800 1812.5 194.868866
## 3 416207000 1112.5 51.706019
## 4 161518000 87.5 6.288903
## 5 185038000 437.5 20.234731
## 6 144043000 237.5 21.324498
head(Dash)
## date txVolume.USD. txCount marketcap.USD. price.USD.
## 1 2/14/2014 150641.42 3421 702537 0.374024
## 2 2/15/2014 78256.18 3663 1092120 0.314865
## 3 2/16/2014 96549.99 3236 1085280 0.406976
## 4 2/17/2014 367922.18 2766 1360260 1.450000
## 5 2/18/2014 838488.30 2631 3960320 1.040000
## 6 2/19/2014 675559.38 2551 3497850 0.941647
## exchangeVolume.USD. generatedCoins fees
## 1 15422 33849 3.355010
## 2 21119 33235 3.739035
## 3 28017 30944 5.238558
## 4 178618 20713 2.473010
## 5 160779 15940 2.827082
## 6 60551 13548 2.363510
head(Ethereum)
## date txVolume.USD. txCount marketcap.USD. price.USD.
## 1 8/7/2015 0 0 0 2.770000
## 2 8/8/2015 1513209 2016 167911000 0.753325
## 3 8/9/2015 1180418 2807 42637600 0.701897
## 4 8/10/2015 825663 1298 43130000 0.708448
## 5 8/11/2015 1787874 1999 42796500 1.070000
## 6 8/12/2015 1812412 4945 64018400 1.220000
## exchangeVolume.USD. generatedCoins fees
## 1 164329 27075.47 0.00000
## 2 674188 27437.66 37.31841
## 3 532170 27943.44 68.09997
## 4 405283 27178.28 14.09895
## 5 1463100 27817.34 31.16514
## 6 2150620 28027.81 11.31145
head(Litecoin)
## date txVolume.USD. txCount marketcap.USD. price.USD.
## 1 4/28/2013 39038951 8847 73773400 4.35
## 2 4/29/2013 48283929 9408 74952700 4.38
## 3 4/30/2013 38686090 9092 75726800 4.30
## 4 5/1/2013 33849471 9205 73901200 3.80
## 5 5/2/2013 58715299 8927 65242700 3.37
## 6 5/3/2013 13752345 8290 58607400 3.04
## exchangeVolume.USD. generatedCoins fees
## 1 0 32800 511.0816
## 2 0 31500 634.1212
## 3 0 32450 597.0982
## 4 0 31600 755.4951
## 5 0 31450 689.1598
## 6 0 28300 551.3278
More Competitors
I was interested in two coins that are currently valued really low, Ripple and Tether. I was able to find the historical data on https://coinmarketcap.com. Since I couldn’t download it, I decided to scrape the data.
Rip <- read_html("https://coinmarketcap.com/currencies/ripple/historical-data/?start=20130428&end=20171209")
Tet <- read_html("https://coinmarketcap.com/currencies/tether/historical-data/?start=20130428&end=20171209")
Ripp <- html_text(html_nodes(Rip, "td"))
Rippl <- matrix(Ripp, ncol = 7, byrow = TRUE)
Ripple <- data.frame(Rippl[2:1588,], stringsAsFactors = TRUE)
colnames(Ripple) <- c("Date","Open","High","Low","Close","Volume","Market Cap")
head(Ripple)
## Date Open High Low Close Volume
## 1 Dec 08, 2017 0.223636 0.278673 0.222168 0.252125 660,172,000
## 2 Dec 07, 2017 0.232623 0.233760 0.221340 0.222823 275,205,000
## 3 Dec 06, 2017 0.245416 0.245705 0.227742 0.232544 274,526,000
## 4 Dec 05, 2017 0.253598 0.253988 0.245234 0.246101 174,591,000
## 5 Dec 04, 2017 0.252919 0.255362 0.247160 0.253571 104,650,000
## 6 Dec 03, 2017 0.255530 0.263072 0.247391 0.252558 134,710,000
## Market Cap
## 1 8,663,460,000
## 2 9,011,630,000
## 3 9,507,190,000
## 4 9,815,990,000
## 5 9,768,480,000
## 6 9,869,310,000
Teth <- html_text(html_nodes(Tet, "td"))
Tethe <- matrix(Teth, ncol = 7, byrow = TRUE)
Tether <- data.frame(Tethe[2:1013,], stringsAsFactors = TRUE)
colnames(Tether) <- c("Date","Open","High","Low","Close","Volume","Market Cap")
head(Tether)
## Date Open High Low Close Volume Market Cap
## 1 Dec 08, 2017 1.04 1.06 0.986563 1.02 1,993,030,000 843,587,000
## 2 Dec 07, 2017 1.01 1.08 1.00 1.03 1,671,610,000 819,775,000
## 3 Dec 06, 2017 0.999760 1.02 0.995840 1.01 1,281,490,000 813,822,000
## 4 Dec 05, 2017 1.00 1.01 0.996458 1.00 814,146,000 816,872,000
## 5 Dec 04, 2017 1.00 1.01 0.992132 1.00 668,510,000 816,012,000
## 6 Dec 03, 2017 1.00 1.03 0.985320 1.00 946,749,000 814,847,000
Cleaning the data
I want all the to be in the same format as the Bitcoin API. For the CSV files, I first wanted to reorder the data since it starts from the earliest point.
Bc_cash <-Bc_cash[130:1,]
Dash <-Dash[1393:1,]
Ethereum <-Ethereum[854:1,]
Litecoin <-Litecoin[1685:1,]
The Bitcoin API is in a YYYY-MM-DD format. So I needed to change that all the dates to that format.
Bc_cash$date <- as.Date(Bc_cash$date, format = "%m/%d/%Y")
Ethereum$date <- as.Date(Ethereum$date, format = "%m/%d/%Y")
Dash$date <- as.Date(Dash$date, format = "%m/%d/%Y")
Litecoin$date <- as.Date(Litecoin$date, format = "%m/%d/%Y")
Change column name ‘date’ to ‘Date’
colnames(Bc_cash)[1] <- "Date"
colnames(Ethereum)[1] <- "Date"
colnames(Dash)[1] <- "Date"
colnames(Litecoin)[1] <- "Date"
Since the two scraped data is from a html website, all of it needs to be reformatted.
Tether[,2:5] %<>%
mutate_each(funs(if(is.factor(.)) as.character(.) else .)) %<>%
mutate_each(funs(if(is.character(.)) as.numeric(.) else .))
Ripple[,2:5] %<>%
mutate_each(funs(if(is.factor(.)) as.character(.) else .)) %<>%
mutate_each(funs(if(is.character(.)) as.numeric(.) else .))
Also, change the date format to YYYY-MM-DD format.
Tether$Date <- as.Date(Tether$Date, format = "%b %d, %Y")
Ripple$Date <- as.Date(Ripple$Date, format = "%b %d, %Y")
Combining the data
Now that all of our data is in consistent format, lets get all the variables needed for analysis.
I want to average out the highs and lows to get one price for the day.
require(magrittr)
Bitcoin %<>% mutate(bitcoin_price = (High + Low)/2,
bitcoin_gain = Close - Open)
Ripple %<>% mutate(ripple_price = (High + Low)/2,
ripple_gain = Close - Open)
Tether %<>% mutate(tether_price = (High + Low)/2,
tether_gain = Close - Open)
Selecting the data we need.
Bitcoin_data <- select(Bitcoin, Date,bitcoin_price)
Ripple_data <- select(Ripple, Date, ripple_price)
Tether_data <- select(Tether, Date, tether_price)
Ethereum_data <- select(Ethereum, Date, price.USD.)
Dash_data <- select(Dash, Date, price.USD.)
Bc_cash_data <- select(Bc_cash, Date, price.USD.)
Litecoin_data <- select(Litecoin, Date, price.USD.)
Since, all of our data is in the same format, we join them using the ‘Date’ field as the identifier. Bitcoin data has the most rows, so I used it as the left part of left join.
cryptocurrencies <-
left_join(
left_join(
left_join(
left_join(
left_join(
left_join(Bitcoin_data, Ripple_data, "Date"),
Tether_data, "Date"),
Ethereum_data, "Date"),
Dash_data, "Date"),
Bc_cash_data, "Date"),
Litecoin_data, "Date")
colnames(cryptocurrencies) <- c("Date","bitcoin_price","ripple_price","tether_price",
"ethereum_price", "dash_price","bc_cash_price","litecoin_price")
head(cryptocurrencies)
## Date bitcoin_price ripple_price tether_price ethereum_price
## 1 2017-12-07 14850.76 0.2275500 1.040000 434.41
## 2 2017-12-06 12679.90 0.2367235 1.007920 428.59
## 3 2017-12-05 11617.12 0.2496110 1.003229 463.28
## 4 2017-12-04 11231.53 0.2512610 1.001066 470.20
## 5 2017-12-03 11156.58 0.2552315 1.007660 465.85
## 6 2017-12-02 10918.85 0.2549870 1.007805 463.45
## dash_price bc_cash_price litecoin_price
## 1 697.90 1330.93 98.29
## 2 700.07 1430.10 100.35
## 3 756.36 1501.85 102.40
## 4 774.01 1576.92 104.24
## 5 768.88 1559.93 101.26
## 6 778.43 1434.98 100.28
Visualizing the currencies
ggplot(data = cryptocurrencies[which(cryptocurrencies$bitcoin_price != 0),]) +
geom_line(mapping = aes(x= Date, y= bitcoin_price), position = "jitter", color = "Blue") +
ggtitle("Bitcoin Price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$ethereum_price != 0),]) +
geom_line(mapping = aes(x= Date, y= ethereum_price), position = "jitter", color = "Green")+
ggtitle("Ethereum_price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$litecoin_price != 0),]) +
geom_line(mapping = aes(x= Date, y= litecoin_price), position = "jitter", color = "Red")+
ggtitle("Litecoin Price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$dash_price != 0),]) +
geom_line(mapping = aes(x= Date, y= dash_price), position = "jitter", color = "Orange")+
ggtitle("Dash Price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$bc_cash_price != 0),]) +
geom_line(mapping = aes(x= Date, y=bc_cash_price), position = "jitter")+
ggtitle("Bit Coin Cash Price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$ripple_price != 0),]) +
geom_line(mapping = aes(x= Date, y= ripple_price), position = "jitter", color = "Pink")+
ggtitle("Ripple Price")
ggplot(data = cryptocurrencies[which(cryptocurrencies$tether_price != 0),]) +
geom_line(mapping = aes(x= Date, y= tether_price), position = "jitter", color = "Purple")+
ggtitle("Tether Price")
Since the prices of the currencies between each other have huge differences, it would be hard to compare them. I decided to calculate the relative change between them.
I created another table with the previous day’s price, added 1 to the day and did a left join to match the day.
Prior_day <- cryptocurrencies[2:2278,]
Prior_day$Date <- as.Date(Prior_day$Date)+1
colnames(Prior_day) <- c("Date","yes_bitcoin_price","yes_ripple_price","yes_tether_price",
"yes_ethereum_price", "yes_dash_price","yes_bc_cash_price","yes_litecoin_price")
r_cyptocurrencies <- left_join(cryptocurrencies, Prior_day, by ="Date")
r_cyptocurrencies <- select(r_cyptocurrencies, Date, bitcoin_price, yes_bitcoin_price, ripple_price, yes_ripple_price, tether_price, yes_tether_price,
ethereum_price, yes_ethereum_price, dash_price, yes_dash_price, bc_cash_price, yes_bc_cash_price, litecoin_price, yes_litecoin_price)
head(r_cyptocurrencies)
## Date bitcoin_price yes_bitcoin_price ripple_price yes_ripple_price
## 1 2017-12-07 14850.76 12679.90 0.2275500 0.2367235
## 2 2017-12-06 12679.90 11617.12 0.2367235 0.2496110
## 3 2017-12-05 11617.12 11231.53 0.2496110 0.2512610
## 4 2017-12-04 11231.53 11156.58 0.2512610 0.2552315
## 5 2017-12-03 11156.58 10918.85 0.2552315 0.2549870
## 6 2017-12-02 10918.85 10160.00 0.2549870 0.2491330
## tether_price yes_tether_price ethereum_price yes_ethereum_price
## 1 1.040000 1.007920 434.41 428.59
## 2 1.007920 1.003229 428.59 463.28
## 3 1.003229 1.001066 463.28 470.20
## 4 1.001066 1.007660 470.20 465.85
## 5 1.007660 1.007805 465.85 463.45
## 6 1.007805 1.006522 463.45 466.54
## dash_price yes_dash_price bc_cash_price yes_bc_cash_price litecoin_price
## 1 697.90 700.07 1330.93 1430.10 98.29
## 2 700.07 756.36 1430.10 1501.85 100.35
## 3 756.36 774.01 1501.85 1576.92 102.40
## 4 774.01 768.88 1576.92 1559.93 104.24
## 5 768.88 778.43 1559.93 1434.98 101.26
## 6 778.43 797.53 1434.98 1462.68 100.28
## yes_litecoin_price
## 1 100.35
## 2 102.40
## 3 104.24
## 4 101.26
## 5 100.28
## 6 99.00
Calculate relative change
r_cyptocurrencies %<>% mutate(bitcoin_change = ((bitcoin_price - yes_bitcoin_price)/yes_bitcoin_price)*100,
ripple_change = ((ripple_price - yes_ripple_price)/yes_ripple_price)*100,
tether_change = ((tether_price - yes_tether_price)/yes_tether_price)*100,
ethereum_change = ((ethereum_price - yes_ethereum_price)/yes_ethereum_price)*100,
dash_change = ((dash_price - yes_dash_price)/yes_dash_price)*100,
bc_cash_change = ((bc_cash_price - yes_bc_cash_price)/yes_bc_cash_price)*100,
litecoin_change = ((litecoin_price - yes_litecoin_price)/yes_litecoin_price)*100)
Create a new table with just the relative changes.
relative_change <- select(r_cyptocurrencies, "Date", ends_with("change"))
relative_change_ <- relative_change
relative_change_[is.na(relative_change_ <- relative_change)] <- 0
head(relative_change_)
## Date bitcoin_change ripple_change tether_change ethereum_change
## 1 2017-12-07 17.1204820 -3.87519617 3.18279229 1.3579412
## 2 2017-12-06 9.1483478 -5.16303368 0.46759015 -7.4879123
## 3 2017-12-05 3.4331015 -0.65668767 0.21606967 -1.4717142
## 4 2017-12-04 0.6718006 -1.55564654 -0.65438739 0.9337770
## 5 2017-12-03 2.1773365 0.09588724 -0.01443731 0.5178552
## 6 2017-12-02 7.4689469 2.34974893 0.12756806 -0.6623226
## dash_change bc_cash_change litecoin_change
## 1 -0.3099690 -6.934480 -2.0528151
## 2 -7.4422233 -4.777441 -2.0019531
## 3 -2.2803323 -4.760546 -1.7651573
## 4 0.6672042 1.089151 2.9429192
## 5 -1.2268284 8.707438 0.9772637
## 6 -2.3948942 -1.893784 1.2929293
Round the percentages
relative_change_$bitcoin_change <- round(relative_change_$bitcoin_change, 4)
relative_change_$ripple_change <- round(relative_change_$ripple_change ,4)
relative_change_$tether_change <- round(relative_change_$tether_change ,4)
relative_change_$ethereum_change <- round(relative_change_$ethereum_change ,4)
relative_change_$dash_change <- round(relative_change_$dash_change ,4)
relative_change_$bc_cash_change <- round(relative_change_$bc_cash_change ,4)
relative_change_$litecoin_change <- round(relative_change_$litecoin_change ,4)
Analysis
Relative Change
Tidy the data for analysis. The mean relative change shows the average changes.
tidy_change <- (gather(relative_change_, key= Currency, "relative_change", desc("Date")))
tidy_change %>% filter(relative_change !=0, is.finite(relative_change) == TRUE) %>%
group_by(Currency) %>% summarise(mean = mean(relative_change), count = n())
## # A tibble: 7 x 3
## Currency mean count
## <chr> <dbl> <int>
## 1 bc_cash_change 1.97931085 129
## 2 bitcoin_change -0.05745514 2249
## 3 dash_change 0.98085324 1360
## 4 ethereum_change 0.93719399 849
## 5 litecoin_change 0.44710412 1602
## 6 ripple_change 0.44249058 1581
## 7 tether_change 0.02202607 399
From the data, we can see that Bitcoin prices has a negative relative change. This might be due to some skews in our data. Lets limit high percentage changes.
tidy_change <- (gather(relative_change_, key= Currency, "relative_change", desc("Date")))
tidy_change %>% filter(relative_change !=0, is.finite(relative_change) == TRUE, relative_change < 50, relative_change > -50) %>%
group_by(Currency) %>% summarise(mean = mean(relative_change), count = n())
## # A tibble: 7 x 3
## Currency mean count
## <chr> <dbl> <int>
## 1 bc_cash_change 1.16503465 127
## 2 bitcoin_change 0.41088594 2233
## 3 dash_change 0.57381055 1355
## 4 ethereum_change 0.90714835 846
## 5 litecoin_change 0.22708623 1598
## 6 ripple_change 0.29571635 1578
## 7 tether_change 0.02202607 399
We see that when we limit the relative changes between -50 and 50, there is significant changes to the mean, but not the count.
tidy_change <- (gather(relative_change_, key= Currency, "relative_change", desc("Date")))
tidy_change %>% filter(relative_change !=0, is.finite(relative_change) == TRUE, relative_change < 10, relative_change > -10) %>%
group_by(Currency) %>% summarise(mean = mean(relative_change), count = n())
## # A tibble: 7 x 3
## Currency mean count
## <chr> <dbl> <int>
## 1 bc_cash_change -0.71430515 97
## 2 bitcoin_change 0.36466306 2152
## 3 dash_change -0.12132624 1189
## 4 ethereum_change -0.12964241 738
## 5 litecoin_change -0.08903114 1461
## 6 ripple_change -0.14734908 1465
## 7 tether_change 0.02202607 399
When we limit the relative change to between -10 and 10, most of the mean is negative. This means for 5/7 observations, the increases are due to high shifts in the market in the day, but most days, they are losing value. This is opposite for Bitcoin, which is increasing on most days and its losses come from large drops in the market. For Tether, the market looks surprisingly stable. This may be due to it being a small player in the market.
Visualization
ggplot(data =tidy_change[which(tidy_change$relative_change != 0),]) +
geom_line(mapping = aes(x= Date, y= relative_change, color= Currency,fill = Currency), position = "jitter")+
facet_wrap(~ Currency, nrow = 2)+
ylim(-100,100)
## Warning: Ignoring unknown aesthetics: fill
From the visualization, we can see that the spread is very large for most of the currencies with exception to tether. Visualizing the relative changes together.
ggplot(data =tidy_change[which(tidy_change$relative_change != 0),]) +
geom_point(mapping = aes(x= Date, y= relative_change, color= Currency,fill = Currency),alpha = 1/5, position = "jitter") +
ylim(-50,50)+
theme(panel.background = element_rect(fill = 'white'))+
ggtitle("Cryptocurrencies from 2011 to 2017",subtitle = "Relative Changes")
## Warning: Removed 33 rows containing missing values (geom_point).
Gain to Loss Ratio
Let’s dive deeper into how the currencies fluctuate and look at gain to loss ratio. Categorize for each relative change percentage to ‘gain’, ‘loss’ or ‘no change’.
tidy_change$lost_gain <- ifelse(tidy_change$relative_change > 0, "gain",
ifelse(tidy_change$relative_change ==0,"no,change","loss"))
tidy_change[sample(nrow(iris), 10), ]
## Date Currency relative_change lost_gain
## 74 2017-09-25 bitcoin_change 3.0917 gain
## 27 2017-11-11 bitcoin_change -5.3339 loss
## 11 2017-11-27 bitcoin_change 6.0537 gain
## 56 2017-10-13 bitcoin_change 9.6475 gain
## 114 2017-08-16 bitcoin_change 1.2907 gain
## 137 2017-07-24 bitcoin_change 0.1092 gain
## 55 2017-10-14 bitcoin_change 1.3287 gain
## 128 2017-08-02 bitcoin_change -2.4200 loss
## 131 2017-07-30 bitcoin_change -1.3529 loss
## 105 2017-08-25 bitcoin_change 3.4083 gain
Untidy the data for analysis
#All
ratio<- tidy_change %>% group_by(Currency) %>% count(lost_gain)
gain_loss_ratio <- ratio %<>% spread(lost_gain, n) %<>% mutate(ratio = gain/loss)
gain_loss_ratio <- select(gain_loss_ratio,Currency,ratio)
#Limit -100 to 100
ratio100<- tidy_change %>% filter(relative_change < 100, relative_change > -100) %>% group_by(Currency) %>% count(lost_gain)
gain_loss_ratio100 <- ratio100 %<>% spread(lost_gain, n) %<>% mutate(ratio100 = gain/loss)
gain_loss_ratio100 <- select(gain_loss_ratio100,Currency,ratio100)
#Limit -10 to 10
ratio1<- tidy_change %>% filter(relative_change < 10, relative_change > -10) %>% group_by(Currency) %>% count(lost_gain)
gain_loss_ratio1 <- ratio1 %<>% spread(lost_gain, n) %<>% mutate(ratio10 = gain/loss)
gain_loss_ratio1 <- select(gain_loss_ratio1,Currency,ratio10)
All_ratios <- inner_join(gain_loss_ratio,inner_join(gain_loss_ratio100,gain_loss_ratio1,by = "Currency"),by= "Currency")
All_ratios$Currency <- c("Bitcoin Cash", "Bitcoin", "Dash", "Ethereum","Litecoin","Ripple","Tether")
All_ratios
## # A tibble: 7 x 4
## # Groups: Currency [?]
## Currency ratio ratio100 ratio10
## <chr> <dbl> <dbl> <dbl>
## 1 Bitcoin Cash 0.8970588 0.8970588 0.7017544
## 2 Bitcoin 1.2507463 1.2520161 1.2510460
## 3 Dash 0.9209040 0.9180791 0.8405573
## 4 Ethereum 0.9698376 0.9698376 0.8403990
## 5 Litecoin 0.9418182 0.9406061 0.9023438
## 6 Ripple 0.8843862 0.8831943 0.8381430
## 7 Tether 1.1336898 1.1336898 1.1336898
The gain to loss ratio table shows that Bitcoin has the highest ratio with over +25% margin over total losses. 2ND is Tether with a +12% margin over total losses. The rest have more losses than gains in their history.
Lets visualize the results.
ratio_bar <- ggplot(All_ratios, mapping = aes(y= ratio,x= Currency, fill=Currency)) +
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Gain to Loss Ratio")+
labs(y= "Ratio")
ratio_bar + coord_flip()
ratio_bar100 <- ggplot(All_ratios, mapping = aes(y= ratio100,x= Currency, fill=Currency)) +
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Gain to Loss Ratio: Relative Change Limit (-100:100)")+
labs(y= "Ratio")
ratio_bar100 + coord_flip()
ratio_bar1 <- ggplot(All_ratios, mapping = aes(y= ratio10,x= Currency, fill=Currency))+
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Gain to Loss Ratio: Relative Change Limit (-10:10)")+
labs(y= "Ratio")
ratio_bar1 + coord_flip()
When we compare the overall gains to losses, we can see that there is a significant drop for Ripple, Litecoin, Ethereum, Dash and Bitcoin cash.
Probability
Lets look at the probability of currencies increasing on a given day, based on the count of gains and losses.
gain_loss_counts <- tidy_change %>% group_by(Currency, lost_gain) %>% summarise(n=n())
gain_loss_counts <- spread(gain_loss_counts, lost_gain, n)
gain_loss_counts <- mutate(gain_loss_counts, probability = (gain/(gain+loss)))
probability <- select(gain_loss_counts, Currency, probability)
gain_loss_counts100 <- tidy_change %>% filter(relative_change < 100, relative_change > -100) %>%
group_by(Currency, lost_gain) %>% summarise(n=n())
gain_loss_counts100 <- spread(gain_loss_counts100, lost_gain, n)
gain_loss_counts100 <- mutate(gain_loss_counts100, probability100 = (gain/(gain+loss)))
probability100 <- select(gain_loss_counts100, Currency, probability100)
gain_loss_counts10 <- tidy_change %>% filter(relative_change < 10, relative_change > -10) %>%
group_by(Currency, lost_gain) %>% summarise(n=n())
gain_loss_counts10 <- spread(gain_loss_counts10, lost_gain, n)
gain_loss_counts10 <- mutate(gain_loss_counts10, probability10 = (gain/(gain+loss)))
probability10 <- select(gain_loss_counts10, Currency, probability10)
All_probability <- inner_join(probability,inner_join(probability100,probability10,by = "Currency"),by= "Currency")
All_probability$Currency <- c("Bitcoin Cash", "Bitcoin", "Dash", "Ethereum","Litecoin","Ripple","Tether")
All_probability
## # A tibble: 7 x 4
## # Groups: Currency [?]
## Currency probability probability100 probability10
## <chr> <dbl> <dbl> <dbl>
## 1 Bitcoin Cash 0.4728682 0.4728682 0.4123711
## 2 Bitcoin 0.5557029 0.5559534 0.5557621
## 3 Dash 0.4794118 0.4786451 0.4566863
## 4 Ethereum 0.4923439 0.4923439 0.4566396
## 5 Litecoin 0.4850187 0.4846971 0.4743326
## 6 Ripple 0.4693232 0.4689873 0.4559727
## 7 Tether 0.5313283 0.5313283 0.5313283
The probability table shows that Bitcoin has the highest chance of increase on a given day with 55% and that stays consistent when outliers are removed. Second is Tether with 53%. The rest is in the 47%-49% range, but decreases when outliers are removed. out of the remaining 5, Litecoin is most consistent, dropping from 48.5% to 47.4%.
Lets visualize the results.
probability_bar <- ggplot(All_probability, mapping = aes(y=probability,x= Currency, fill=Currency)) +
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Probability of Increase within a day")+
labs(y= "Probability") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
probability_bar
probability_bar100 <- ggplot(All_probability, mapping = aes(y= probability100,x= Currency, fill=Currency)) +
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Probability of Increase within a day: Relative Change Limit (-100:100)")+
labs(y= "Probability") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
probability_bar100
probability_bar1 <- ggplot(All_probability, mapping = aes(y=probability10,x= Currency, fill=Currency))+
geom_bar(stat = "identity") +
ggtitle("Cryptocurrencies from 2011-2017", subtitle = "Probability of Increase within a day: Relative Change Limit (-10:10)")+
labs(y= "Probability")+
theme(axis.text.x = element_text(angle = 45, hjust = 1))
probability_bar1
Conclusion
Based on the day to day relative change of the currencies, we have some insight on how the market fluctuates. We saw that the relative change showed that there was an inverse relationship between Bitcoin and most of the currencies. Bitcoin Cash, Dash, Ethereum, Litecoin, and Ripple’s decrease most days with few large increases making up for the losses. While Bitcoin and Tether increase most days, with large drops evening it out.
When we looked at the gain to loss ratio, Bitcoin and Tether had a surplus in increases over losses, while the other 5 had a ratio of below 1. This supports the previous conclusion that there are more gains than losses and when we look at probabilities of this. We see that Bitcoin and Tether have a chance of increase more than 50% in a given day.
For someone that is interested in investing in Cryptocurrency, based on the analysis above, their best bet would be to invest in Bitcoin since. Since there isn’t much more information on Tether, more in depth research needs to be done for a conclusion. If someone is looking to invest in another currency, the data tells us that Litecoin has the highest upside.
Overall, since the market moves so fast, a day to day analysis might not be enough to come up with an accurate assumption.