What's covered in this lecture?

• Applied AI in Banking and Finance

• Case 1: Financial Index Analysis

  • Review of AIEQ Case

  • Hang Seng Index

• Case 2: Financial Anomoly Detection

  • Outlier Peeling Algorithm

  • Radial Chart Visualization

Applied AI in FinTech and Financial Services

There are multiple Financial Technology (FinTech) and Financial Service areas where Artificial Intelligence (AI) and Machine Learning (ML) can be applied, including

• Credit risk management (Credit scoring, Loan acquisition, Default forecasting, P2P lending)

• Investment analysis (Trading strategy, Portfolio optimization, Money management)

• Cryptocurrency (BitCoin, Crypto index)

• Fraud detection and anti-money laundering

• Customer service (Automated customer support via chatbots, Customer segmentatoin)

Two FinTech Case Studies

In this lecture, we discuss two FinTech case studies with the equipped DataViz skills:

• Case 1: Financial Index Analysis

  • Review of AIEQ Case

  • Hang Seng Index

• Case 2: Finanial Anomoly Detection

  • Outlier Peeling Algorithm

  • Radial Chart Visualization

Investment Stratgies

• Active investing

  • The market is not perfectly efficient

  • Aims to beat the market and take advantage of short-term price fluctuations

• Passive investing

  • The market is efficient in the long run

  • Buy-and-hold philosophy

  • Index tracking strategy, e.g. ETF

Review of AIEQ Case

• Click here for the reviewed case study, including AIEQ Performance, Holdings Analysis and Portfolio Analysis

AIEQ Performance To Date

New Case Study: Hang Seng Index

• Query data from Yahoo!Finance

• Interactive time series display by R:dygraphs (Online Tutorial)

library(quantmod)
library(xts)
library(dygraphs)
library(htmltools)
date0 = "2000-01-01"
dummy = getSymbols("^HSI", src="yahoo", from = date0)
div(dygraph(HSI$HSI.Close, main = "Hang Seng Index since 2000") %>%
      dyRangeSelector(dateWindow = c("2010-01-01", "2018-12-31")) %>%
      dyOptions(colors = c("red")), 
    align="center")

Hang Seng Index Constituents

• HSI historical data can be obtained from Bloomberg Terminal

• We collected daily weights of HSI constituents since 2016-01-01

Hang Seng Index Constituents

• X.Shares <- X.Weight * HSI.Close * HSI.Divisor / X.Close

• Data Preprocessing to convert data into space-time format

DataX = read.csv("DataHSI.csv")
DataX$Date2 = as.factor(DataX$Date)
tmp = DataX[!duplicated(DataX$Ticker, fromLast = T),]
idx = order(tmp$Date, tmp$Weight, decreasing = T)
ListTicker=tmp$Ticker[idx]
n = length(ListTicker) 
ListDate = as.character(unique(DataX$Date))
MatWeight <- MatShares <- matrix(NA, length(ListDate), n)
for (j in 1:n){
  TmpX = DataX[DataX$Ticker==ListTicker[j], c("Date2", "Weight", "Shares")]
  MatWeight[as.numeric(TmpX$Date2),j] = TmpX$Weight
  MatShares[as.numeric(TmpX$Date2),j] = TmpX$Shares
}
colnames(MatWeight) <- colnames(MatShares) <- ListTicker
MatWeight = xts(MatWeight, order.by=as.Date(ListDate, "%Y%m%d"))
MatShares = xts(MatShares, order.by=as.Date(ListDate, "%Y%m%d"))

HSI Constituent Weights and Shares

par(mfrow=c(1,2), mar=rep(2,4))
image(as.matrix(MatWeight), ylim=c(1,0), col = heat.colors(10), xaxt='n', yaxt='n', main="Weights")
image(as.matrix(MatShares), ylim=c(1,0), col = terrain.colors(10), xaxt='n', yaxt='n', main="Shares")

div(dygraph(MatWeight, main = "HSI Constituent Weights",height=450) %>%
      dyRangeSelector(dateWindow = c("2016-01-01", "2018-12-31")) %>%
      dyLegend(show = "never"), align="center")

HSI Constituent Weights

0

2

4

6

8

10

12

Jan 2016

Apr 2016

Jul 2016

Oct 2016

Jan 2017

Apr 2017

Jul 2017

Oct 2017

Jan 2018

Apr 2018

Jul 2018

Oct 2018

TmpX = MatShares[,1:4]; TmpX = TmpX/rep(1, dim(TmpX)[1])%*%t(colMeans(TmpX, na.rm=T))
div(dygraph(TmpX, main = "HSI Constituent Shares", ylab="Normalized shares",height=450) %>%
      dyRangeSelector(dateWindow = c("2016-01-01", "2018-12-31")) %>%
      dyOptions(colors = c("red", "blue", "green", "cyan"), strokeWidth=2), align="center")

HSI Constituent Shares

Normalized shares

5.HK

1299.HK

700.HK

939.HK

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

Jan 2016

Apr 2016

Jul 2016

Oct 2016

Jan 2017

Apr 2017

Jul 2017

Oct 2017

Jan 2018

Apr 2018

Jul 2018

Oct 2018

HSI Constituent Rebalancing

TmpX = MatShares[,"700.HK"]
div(dygraph(TmpX, main = "HSI Constituent Shares: 700.HK", height=360) %>%
      dyRangeSelector(dateWindow = c("2016-01-01", "2018-12-31")) %>%
      dyOptions(colors = "green"), align="center")

HSI Constituent Shares: 700.HK

200000

250000

300000

350000

400000

450000

Jan 2016

Apr 2016

Jul 2016

Oct 2016

Jan 2017

Apr 2017

Jul 2017

Oct 2017

Jan 2018

Apr 2018

Jul 2018

Oct 2018