Skip to main content

Table 3 Proportion p(1) of class 1 per stock as function of the return threshold d, p(0) = 1 − p(1)

From: Forecasting relative returns for S&P 500 stocks using machine learning

No

Stocks

p(1)

d = 0.3%

p(1)

d = 0.5%

p(1)

d = 1.8%

p(1)

d = 2%

p(1)

d = 2.2%

p(1)

d = 9.5%

p(1)

d = 10%

1

AMAT

0.82

0.80

0.62

0.61

0.59

0.12

0.11

2

BSX

0.79

0.75

0.50

0.48

0.44

0.02

0.02

3

CDNS

0.84

0.82

0.61

0.59

0.55

0.05

0.05

4

PHM

0.82

0.79

0.63

0.60

0.56

0.09

0.08

493

SIVB

0.80

0.77

0.62

0.60

0.57

0.14

0.13

494

ZTS

0.83

0.80

0.55

0.51

0.47

0.02

0.01

 

Average

0.77

0.74

0.52

0.49

0.47

0.06

0.06

  1. We use the average value of the observed proportion of class 1 across all stocks and all trading days as an approximation to the probability p(1) of class 1