tech.nmfin.signal.infantino2010

Class Infantino2010PCA

java.lang.Object

tech.nmfin.signal.infantino2010.Infantino2010PCA

public class Infantino2010PCA
extends Object

The objective is to predict the next H-period accumulated returns from the past H-period dimensionally reduced returns. The dimensionally reduced returns are computed using PCA to remove idiosyncratic noises from the raw, realized accumulated returns.

See Also:

• Infantino, Leandro Rafael; Itzhaki, Savion "Developing High-Frequency Equities Trading Models," Massachusetts Institute of Technology, 2010
• Wikipedia: Principal component analysis
• Wikipedia: Linear regression

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Infantino2010PCA.Signal

Field Summary

Fields

Modifier and Type	Field and Description
protected int	H
protected int	k
protected int	N
protected int	T

Constructor Summary

Constructors

Constructor and Description
Infantino2010PCA(int T, int N, int H, int k, boolean parallelized)

Method Summary

Modifier and Type	Method and Description
Infantino2010PCA.Signal	getSignal(Matrix X)
static Vector	sumUpLastRows(Matrix D, int lastIndex, int nRows) Sums up, for each column, the last nRows rows.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

T

protected final int T

N

protected final int N

H

protected final int H

k

protected final int k

Constructor Detail

Infantino2010PCA

public Infantino2010PCA(int T,
                        int N,
                        int H,
                        int k,
                        boolean parallelized)

Method Detail

getSignal

public Infantino2010PCA.Signal getSignal(Matrix X)
                                  throws MultipleExecutionException

Throws:

MultipleExecutionException

sumUpLastRows

public static Vector sumUpLastRows(Matrix D,
                                   int lastIndex,
                                   int nRows)

Sums up, for each column, the last nRows rows.

Parameters:

D - a matrix

lastIndex - the ending index of the rows to be summed up

nRows - the number of rows including and above the ending index to be summed up

Returns:

the (running) column sum