If the system works, here the results of the weekly (Friday-Thursday) quantitative analysis and neural network prediction of the selected NYSE stocks.

The quantitative analysis is based on day by day data.

Based on:

- risk free rate (based on current three-month U.S. treasury bill) of last days( Please evaluate the values of the "Estimated_Gain" with utmost caution!

- past volatility of the stock

- last closed price of the stock

- (this is a daytrading system, the dividend does not play a role here)

(Click the table header to sort data according to that column.)

Reference value: close price with EMA-12

ARIMA(p,d,q): p = best choice out of 1..3; d = 1; q = best choice out of 1..3

EGARCH(p,q): p = best choice out of 1..3; q = best choice out of 1..3

exo. input: S&P 500 Index

More configuration information about ARIMA-EGARCH model on demand.

Reference value: log. returns

More configuration information about NN model on demand.

Sym = symbol at the stock exchange

Nr = number in my stock symbol list

Date = last data date of the stock

Price = last price at the end of the day in USD

A_MSE_Tr = ARIMA-EGARCH mean squared error of autoregressive integrated moving average model on train data (decimal)

A_MAPE_Tr = ARIMA-EGARCH mean absolute percentage error of autoregressive integrated moving average model on train data (decimal)

A_HR_Tr = ARIMA-EGARCH hit rate of autoregressive integrated moving average model on train data (decimal)

A_MSE_Te = ARIMA-EGARCH mean squared error of autoregressive integrated moving average model on test data (decimal)

A_MAPE_Te = ARIMA-EGARCH mean absolute percentage error of autoregressive integrated moving average model on test data (decimal)

A_HR_Te = ARIMA-EGARCH hit rate of autoregressive integrated moving average model on test data (decimal)

N_MSE_Tr = NEURAL NETWORK mean squared error of train data (neural network) (decimal)

N_MAPE_Tr = NEURAL NETWORK mean absolute percentage error of train data (neural network) (decimal)

N_HR_Tr = NEURAL NETWORK hit rate of train data (neural network) (decimal)

N_MSE_Te = NEURAL NETWORK mean squared error of test data (neural network) (decimal)

N_MAPE_Te = NEURAL NETWORK mean absolute percentage error of test data (neural network) (decimal)

N_HR_Te = NEURAL NETWORK hit rate of test data (neural network) (decimal)

N_MSP = NEURAL NETWORK multistep performance (next days: 5)

N_DEV = NEURAL NETWORK mean deviation of predicted vs. reached log. returns in % (last days: 5)

VaR = Value at Risk

Min = minimal step

Max = maximal step

Mean = mean step at observation time frame (see Obs, 1 Obs = 1 passed tradeday)

Std = standard deviation

Skew = skewness

Kurt = kurtosis

Obs = observation

CCI (momentum) = Commodity Channel Index (days: 20, const.: 0.015)

RSI (momentum) = Relative Strength Index (days: 14)

WILLR (momentum) = Williams %R (days: 5)

EMA (trend) = Exponential Moving Average (days: 10)

MACD (trend) = Moving Average Convergence Divergence (short: 12 days, long: 26 days, signal: 9 days)

ADX (trend) = Wilder's DMI (days: 14)

OBV (volume) = On Balance Volume

CMF (volume) = Chaikin Money Flow (days: 10)

MFI (volume) = Money Flow Index (days: 10)

ATR (volatility) = Average True Range (days: 20)

VR (volatility) = Volatility Ratio (days: 14)

HHLL (volatility) = Highest High, Lowest Low (days: 20)

LONG_Sig (signal) = long signal (1 = potential to buy)

SHORT_Sig (signal) = shot signal (1 = potential to buy)

QA_Score (score) = my own score based on choiced indicators (on demand)

MA_Choice (score) = markowitz choice (efficient frontier), decimal percentage of possible portfolio, e.g. 0.23 = 23% of whole portfolio

(Click the table header to sort data according to that column.)

© 2017 quantanal.net - *"the decision is only a bit"*- all rights reserved.