–Independent Work Report Spring 2015–

Predicting Stock Price Direction using Support Vector

Machines
Saahil Madge
Advisor: Professor Swati Bhatt

Abstract
Support Vector Machine is a machine learning technique used in recent studies to forecast stock
prices. This study uses daily closing prices for 34 technology stocks to calculate price volatility
and momentum for individual stocks and for the overall sector. These are used as parameters to
the SVM model. The model attempts to predict whether a stock price sometime in the future will be
higher or lower than it is on a given day. We find little predictive ability in the short-run but definite
predictive ability in the long-run.
1. Introduction
Stock price prediction is one of the most widely studied and challenging problems, attracting
researchers from many fields including economics, history, finance, mathematics, and computer
science. The volatile nature of the stock market makes it difficult to apply simple time-series or
regression techniques. Financial institutions and traders have created various proprietary models to
try and beat the market for themselves or their clients, but rarely has anyone achieved consistently
higher-than-average returns on investment. Nevertheless, the challenge of stock forecasting is so
appealing because an improvement of just a few percentage points can increase profit by millions of
dollars for these institutions.
Traditionally, many prediction models have focused on linear statistical time series models such
as ARIMA [7]. However, the variance underlying the movement of stocks and other assets makes
linear techniques suboptimal, and non-linear models like ARCH tend to have lower predictive error
[17]. Recently, researchers have turned to techniques in the computer science fields of big data and
machine learning for stock price forecasting. These apply computational power to extend theories in
mathematics and statistics. Machine learning algorithms use given data to “figure out” the solution
to a given problem. Big data and machine learning techniques are also the basis for algorithmic and
high-frequency trading routines used by financial institutions.
In this paper we focus on a specific machine learning technique known as Support Vector
Machines (SVM). Our goal is to use SVM at time t to predict whether a given stock’s price is higher
or lower on day t + m. We look at the technology sector and 34 technology stocks in particular. We
input four parameters to the model - the recent price volatility and momentum of the individual stock
and of the technology sector. These parameters are calculated using daily closing prices for each
stock from the years 2007 through 2014. We analyze whether this historical data can help us predict
price direction. If the Efficient Markets Hypothesis (EMH) holds true, prices should follow a random
walk and be unpredictable based on historical data. We find that in the short-term this holds true, but
in the long-term we are able to reach prediction accuracies between 55% and 60%. We conclude that
our model is able to achieve significant prediction accuracies with some parameters in the long-term,
but that we must look at more granular intra-day trading data to achieve prediction accuracies in the
short-term. The code written can be found at https://github.com/SaahilMadge/Spring-2015-IW.
2. Background Information
2.1. Stock Market Efficiency
Much economic research has been conducted into the Efficient Markets Hypothesis theory, which
posits that stock prices already reflect all available information [18] and are therefore unpredictable.
According to the EMH, stock prices will only respond to new information and so will follow a
random walk. If they only respond to new information, they cannot be predicted. That the stocks
follow a random walk is actually a sign of market efficiency, since predictable movement would
mean that information was not being reflected by the market prices.
There are three variants of this theory – weak, semi-strong, and strong. Most research has
concluded that the semi-strong version holds true. This version claims that stock prices reflect all
publicly available information, but private information can be used to unfairly predict profits. This
is the basis behind strong insider trading laws.
Nevertheless, there are certain market phenomena that actually run contrary to EMH. These are
known as market anomalies. Jegadeesh and Titman discovered that in the short term, stock prices
tend to exhibit momentum[13]. Stocks that have recently been increasing continue to increase,
and recently decreasing stocks continue to decrease. This type of trend implies some amount of
predictability to future stock prices, contradicting the EMH.
The stock market also exhibits seasonal trends. Jacobsen and Zhang studied centuries’ worth
of data and found that trading strategies can exploit trends in high winter returns and low summer
returns to beat the market [2][3].
If the EMH held perfectly true, then the direction of future stock prices could not be predicted
with greater than 50% accuracy. That is, one should not be able to guess whether future prices
will go up or down better than simple random guessing. However, the studies discussed in 2.3 are
all able to predict price direction with greater than 50% accuracy, implying that machine learning
techniques are able to take advantage of momentum and other price trends to forecast price direction.
We are able to replicate these results, as discussed in 4.
2.2. General Machine Learning
There are two general classes of machine learning techniques. The first is supervised learning,
in which the training data is a series of labeled examples, where each example is a collection of
features that is labeled with the correct output corresponding to that feature set [5]. This means
that the algorithm is given features and outputs for a particular dataset (training data), and must
apply what it “learns” from this dataset to predict the outputs (labels) for another dataset (test data).
Unsupervised learning, on the other hand, consists of examples where the feature set is unlabeled.
The algorithms generally try to cluster the data into distinct groups.
Supervised learning can be further broken down into classification and regression problems.
In classification problems there are a set number of outputs that a feature set can be labeled as,
whereas the output can take on continuous values in regression problems. In this paper we treat the
problem of stock price forecasting as a classification problem. The feature set of a stock’s recent
price volatility and momentum, along with the index’s recent volatility and momentum, are used to
predict whether or not the stock’s price m days in the future will be higher (+1) or lower (−1) than
the current day’s price. Specifically, we are solving a binary classification problem.

2
2.3. Previous Research
Most research with machine learning forecasting has focused on Artificial Neural Networks (ANN)
[4]. ANNs have a series of interconnected nodes that simulate individual neurons, and are organized
into different layers based on function (input layer, processing layer, output layer, etc.). The
ANN assigns weights to connections, and the output is calculated based on the inputs and the
weights. As the machine trains, it notices patterns in the training data and reassigns the weights.
Kar demonstrates that ANNs are quite accurate when the data does not have sudden variations
[11]. Patel and Yalamalle agree that ANNs can predict with accuracy slightly greater than 50%,
but caution that since stock market data varies so greatly with time and nonlinearly, prediction is
difficult even with advanced techniques like ANNs [12].
Recent research in the field has used another technique known as Support Vector Machines
in addition to or as an alternative to ANNs. Whereas ANNs are models that try to minimize
classification error within the training data, SVMs may make classification errors within training
data in order to minimize overall error across test data. A major advantage of SVMs is that it
finds a global optimum, whereas neural networks may only find a local optimum. See 2.4 for the
mathematics behind SVMs.
Using the SVM model for prediction, Kim was able to predict test data outputs with up to 57%
accuracy, significantly above the 50% threshold [9]. Shah conducted a survey study on stock
prediction using various machine learning models, and found that the best results were achieved
with SVM[15]. His prediction rate of 60% agrees with Kim’s conclusion. Since most recent research
has incorporated SVMs, this is the technique we use in our analysis.
2.4. Support Vector Machines
Support Vector Machines are one of the best binary classifiers. They create a decision boundary
such that most points in one category fall on one side of the boundary while most points in the
other category fall on the other side of the boundary. Consider an n-dimensional feature vector
x = (X1 , ..., Xn ) [8]. We can define a linear boundary (hyperplane) as
n
β0 + β1 X1 + ... + βn Xn = β0 + ∑ βi Xi = 0
i=1

Then elements in one category will be such that the sum is greater than 0, while elements in the
other category will have the sum be less than 0. With labeled examples, β0 + ∑ni=1 βi Xi = y, where y
is the label. In our classification, y ∈ {−1, 1}.
We can rewrite the hyperplane equation using inner products.
y = β0 + ∑ αi yi x(i) ∗ x
where ∗ represents the inner product operator. Note that the inner product is weighted by its label.
The optimal hyperplane is such that we maximize the distance from the plane to any point. This
is known as the margin. The maximum margin hyperplane (MMH) best splits the data. However,
since it may not be a perfect differentiation, we can add error variables ε1 ...εn and keep their sum
below some budget B. The crucial element is that only the points closest to the boundary matter for
hyperplane selection; all others are irrelevant. These points are known as the support vectors, and
the hyperplane is known as a Support Vector Classifier (SVC) since it places each support vector in
one class or the other.

3
 The concept of the SVC is similar to another popular linear regression model, Ordinary Least
Squares (OLS), but the two optimize different quantities. The OLS finds the residuals, the distance
from each data point to the fit line, and minimizes the sum of squares of these residuals [10]. The
SVC on the other hand looks at only the support vectors, and uses the inner product to maximize
the distance from hyperplane to the support vector. Additionally, the inner products in SVC are
weighted by their labels, whereas in OLS the square of residuals serves as the weighting. Thus SVC
and OLS are two different methods of approaching this problem.
SVCs are limited in that they are only linear boundaries. SVMs fix this by applying non-linear
kernel functions to map the inputs into a higher-dimensional space and linearly classify in that space.
A linear classification in the higher-dimensional space will be non-linear in the original space. The
SVM replaces the inner product with a more general kernel function K which allows the input to be
mapped to higher-dimensions. Thus in an SVM,

y = β0 + ∑ αi yi K(x(i), x)

3. Model Creation and Evaluation Methods

In this paper we focus on using the SVM model with RBF Kernel for price forecasting. We found
that most recent research has used the SVM model and saw an opportunity to apply this to stock
price data through the Great Recession and subsequent recovery period.
3.1. Data Collection and Timeframe
Economic conditions greatly deteriorated in the Great Recession. Unemployment increased drasti-
cally, making the Great Recession the worst “labor market downturn since the Great Depression”[16].
The S&P500 index dropped 38.49% in 2008 and then increased by 23.45% the next year[1]. In every
year except 2011 the index had a double-digit increase in price. We wanted to see how the SVM
model, which has had such success in previous literature, would work in such an abnormally volatile
market. Although Rosillo, et al found that SVM actually has better accuracy in high-volatility
markets than other types of markets, their study used simulated markets, whereas we used historical
data from the Great Recession time period[14].
We focus specifically on the technology sector. Focusing on a sector as opposed to the broad
market allow us to test the model on companies that are similar to each other, making our results
relatively standardized. We use the NASDAQ-100 Technology Sector Index (NDXT ) as the general
technology sector index. The index consists of technology giants like Microsoft and Apple along
with smaller companies like Akamai and NetApp.
We look at 34 of the 39 stocks in the index. For each individual company we look at daily price
data from the start of 2007 through the end of 2014. This allows us to analyze the fall of each
company during the Recession as well as the recovery up to current times. In previous studies on
this topic the machine learning model has typically been trained on 70% of the dataset and tested on
the remaining 30%. We keep similar proportions. We use the time period 2007-2011 (5 years) for
training, and 2012-2014 (3 years) for testing. This corresponds to approximately 62.5% training
and 37.5% testing set.

4
 Stock price data was obtained from the CRSP stock database, and index data from Yahoo Finance.
We use the daily closing prices. We do not analyze are NXP Semiconductors, Facebook, Google
Class C (non-voting stock), Avago Technologies, and Texas Instruments. The first four stocks are
not part of our analysis because they were publicly listed after 2009. We want to analyze a machine
learning model trained through the Great Recession and recovery, but this cannot be done for stocks
which were not listed during the Recession. Texas Instruments is missing some price data, so we do
not analyze it either.
3.2. SVM Model
The specific kernel function we use in this study is the radial kernel. This function is one of the
most popular kernel functions and is defined as
!
1 n
K(xi , xk ) = exp − 2 ∑ (xi j − xk j )2
δ j=1

where δ is known as the bandwidth of the kernel function [9]. The advantage of this function is that
it can handle diverse input sets, as there are “few conditions on the geometry” of the input data [6].
Additionally, it classifies test examples based on the example’s Euclidean distance to the training
points, and weights closer training points more heavily. This means that classification is based
heavily on the most similar training examples and takes advantage of patterns in the data. This is
exactly what is needed for time-series data such as stock prices that display trends, as discussed in
2.1. We use the python library sklearn’s implementation of SVM.
3.3. Feature Selection
In this study we use four features to predict stock price direction – price volatility, price momentum,
sector volatility, and sector momentum. More detail is provided in Table 1, styled in the form used
by Kim[9].
3.4. Method
In Table 1 we describe how each of the four features are calculated by averaging some quantity over
the past n days. We conduct our study by varying this parameter n to see exactly how trends in
volatility and momentum, both of the particular stock and the index, can be used to predict future
changes in that stock.
Let n1 be parameter n for the index, and n2 be for the given stock, where n1 , n2 ∈ {5, 10, 20, 90, 270}.
These represent one week, two weeks, one month, one quarter, and one year. In each iteration we
supply some combination of n1 , n2 , use these parameters to calculate our feature sets, train on the
training data, and then predict on the testing set and check accuracy of results. We run 25 iterations,
one for each combination of n1 , n2 .
In order to calculate the features we look at every trading date from 2007 through 2014 and
calculate the four features on that date. The four features on that particular date are collected as one
vector. Since we average over the past n1 days for index and n2 days for stock, we start calculating
feature vectors on the d = (max(n1 , n2 ) + 1)-th date. For example, if n1 = 5, n2 = 10, then d = 11
and we start from the 11th date. This is because we skip the first date, as volatility and momentum
are both calculated based on the previous day’s data, and there is no data before the first date.

5
 Table 1: Features used in SVM

Feature name Description Formula

σs Stock price volatility. This is an average

t
over the past n days of percent change Ci −Ci−1
∑
in the given stock’s price per day. i=t−n+1 Ci−1
n

Stock Momentum This is an average of the given stock’s

t
momentum over the past n days. Each y
day is labeled 1 if closing price that day
∑
i=t−n+1
is higher than the day before, and −1 if n
the price is lower than the day before.

σi Index volatility. This is an average over

t
the past n days of percent change in the Ii − Ii−1
∑
index’s price per day. i=t−n+1 Ii−1
n

Index Momentum This is an average of the index’s mo-

t
mentum over the past n days. Each day d
is labeled 1 if closing price that day is
∑
i=t−n+1
higher than the day before, and −1 if n
the price is lower than the day before.
We let Ct be the stock’s closing price at time t, where t is the current day, and define It as the index’s closing price
that day. The stock’s directional change on a given day is labeled as y ∈ {−1, 1}, and the index’s directional change
is defined as d ∈ {−1, 1}. We use these features to predict the direction of price change between t and t + m, where
m ∈ {1, 5, 10, 20, 90, 270}.

We are calculating the features to try and predict the price direction m days in the future, where
m ∈ {1, 5, 10, 20, 90, 270}. As a result we skip the last m dates since we do not have the price m
days after them. There are a total of 2014 trading days between 2007 and 2014, so we have a total
of 2014 − d − m days. The total set of feature vectors is called X.
We also have a set of output vectors y. y is calculated by finding the price direction on each
of the 2014 − d − m days. We then split X and Y into the training and testing sets as described
in 3.1, which we call Xtrain , ytrain , Xtest , ytest . We supply the feature vectors Xtrain as well as their
corresponding output vectors ytrain to the SVM model. This is the training phase. We then supply
only the testing feature vectors Xtest and have the model predict their corresponding output vectors.
We then compare this output to ytrain .

6
4. Results
The full table of results can be found in the Appendix.

5. Discussion
One notable result is that the SVM model is only tenths of a percentage point better than simple
random guessing when it comes to predicting price direction one day ahead (m = 1). This has
several important implications. First, it strongly reinforces the Efficient Markets Hypothesis. If a
model that incorporates several types of historical data, including some features like momentum
that economists have demonstrated are present in stock price data, is not able to do better than a
coin flip when it comes to predicting the next day’s price direction, then this is very strong evidence
that prices follow a random walk. Prices that already reflect available information will change only
based on new information, so tomorrow’s price direction will be dependent only on new information
that arrives tomorrow. A model like ours which analyzes only historical data should not be able to
predict the price direction, as all this historical data should already be incorporated into the price.
Therefore, our model’s difficulty in predicting the next day’s stock price supports EMH.
Several trends emerge as we increase the parameter m. The most striking is that the mean and
median returns increase when m = 5, 10, 20 but then decrease slightly when m = 90 and m = 270.
Figure 1 shows the mean prediction accuracy against the parameter m, where mean prediction
accuracy here is the mean of the mean accuracies for all 25 combinations of n1 , n2 with a fixed m,
where each combination itself reports the mean accuracy across the 34 stocks.

Figure 1: The mean prediction accuracy for each parameter m

7
 This number does not tell the whole story, however. Looking at the full results in the Appendix,
we see that when m is small, varying n1 and n2 has little effect. For example, when m = 1 the
mean accuracy is within 49.5% and 50.53% for every combination of n1 , n2 . It is interesting to note
that for prediction when m = 1 very small values or very large values of n1 , n2 work best. Mean
and median accuracy are highest when at least one of the two is 5, or when both are 90 or 270.
This implies that very short-term trends, or long-term trends, are best for predicting the next day’s
price direction. Trends across two weeks or a month, though, perform worse than simple random
guessing.
The parameters n1 and n2 start to become more important as we increase m. When m = 10 the
mean prediction accuracy varies between 53.3% and 56.8%, a much larger range than at m = 1.
This disparity is exaggerated when m = 90, that is, when we try to predict the price direction across
the next quarter. Some combinations of n1 , n2 , such as n1 = 10, n2 = 90 actually result in less than
50% accuracy, which means one would be better off flipping a coin, whereas other combinations
have very high accuracies. For example, n1 = 270, n2 = 5 results in a 61.5% prediction accuracy.
Such discrepancies indicate that as the model tries to forecast farther into the future, the input
parameters and historical data have drastically more influence. This again reinforces the EMH and
the idea of stocks as a random walk. Regardless of the historical data used, short-term changes are
difficult to predict, indicating that trends do not matter much in the short run, but may still offer
some predictive ability. On the other hand, long-term changes are subject to seasonality and the
trends discussed in 2.1. With the right input data, the model can take advantage of these trends. If
the training data reflects similar conditions to those in the test dataset, the model will have high
predictive power. However, the wrong training dataset can skew the model and even lead to worse
than 50% prediction accuracy. Figure 2 shows the prediction accuracies for each combination of
n1 , n2 when m = 90.

Figure 2: The mean prediction accuracy for each combination of n1 , n2 when m = 90

8
 Another noticeable trend is that in general, n1 = 270 results in the highest prediction accuracy for
all m. That is, the overall sector’s historical data over the past year tends to be helpful in predicting
the price direction at any point in the future. It is more helpful as we try to predict further into the
future, but even in the short-term, it appears that the general sector trend is a better predictor than
the short-term sector trend.
Almost the opposite is true for the parameter n2 . Mean prediction accuracy tends to be highest
when n2 = 5, the smallest value, but this is affected by parameter m. When m is small too, choosing
n2 = 5 has most effect; when m is small it still performs better than other values of n2 , but not as
much. This makes intuitive sense; a stock’s recent momentum is better than other indicators since
it reflects most recent information. However, recent momentum is not as helpful when trying to
predict farther into the future.
When m is small, the choice of parameter n2 has more effect than choice of parameter n1 , and
here performance is best when n2 = 5. As m increases, the choice of parameter n1 becomes more
important. Again, this makes intuitive sense. When trying to predict price direction in the short term,
the particular stock’s momentum can provide some indication (but not much). For predicting long-
term price direction, the long-term sector trends are most helpful. Long-term sector trends reflect
the market’s overall macro-conditions, and are not swayed by small fluctuations. The particular
stock’s data, however, reflects only idiosyncratic conditions. This indicates that predicting long-term
price direction for a particular stock might depend more on the overall market’s direction, and larger
macro-conditions, than on the stock’s own trends. Once again, this supports EMH. If long-term
stock direction were predictable based solely on that stock itself, this would mean prices react not
only to new information but also to existing information, violating EMH.
It is also important to note that for short-term periods, simply being able to somewhat accurately
predict tomorrow’s price direction does not necessarily translate into trading profits. Treating the
problem as binary classification allows us to more easily construct models, but it also means that
we cannot predict the magnitude of the price change. Stock transactions have transaction costs,
and the profit gained from small trades might not outweigh the costs. Even though we did not find
predictive ability for the next day, we were able to find some predictive ability within the next week
or month. Nevertheless, translating this small predictive advantage into trading profits is entirely
another issue.
Finally, we note that the long-term predictive ability also may not directly translate to long-term
profits, although not due to trading costs. When m = 90 and m = 270 the overall mean predictive
accuracy is less than in the short term, but with the right choice of parameters n1 , n2 the accuracy
can be higher than 60%. This number is itself an average of the prediction accuracies for each of
the 34. As discussed above and seen in the Appendix, the range of prediction for stocks increases as
we increase m, and becomes very large when m = 90, 270. We see that the model is able to predict
price direction for some stocks with greater than 80% accuracy, but for others cannot predict with
more than 30% accuracy. The problem is that we do not yet know ahead of time which stocks the
model will be able to predict accurately and which it will not, so profiting off the model is still
difficult without more experimentation.

9
6. Future Work
6.1. Adding Granularity
One limitation of this study is that we only looked at daily stock price data. As a result, our
momentum and volatility parameters were calculated over weeks. This is different from actual
hedge funds and quantitative trading institutions, which study price data at a far more granular level,
on the order of minutes or even seconds. We can further this study by looking at intra-day trading
data in addition to closing price data. By observing intra-day trends, we can create more robust
models that capitalize on sudden changes in momentum during intra-day trading. Individual stocks
often have price swings that last only for a couple minutes, which an intra-day model can analyze,
and then use to predict price direction in the next few seconds or minutes. This is how such financial
institutions make trading profits.
6.2. Feature Selection
Feature selection often has the biggest impact on a machine learning model’s accuracy. Another
area for future work is to add to our feature list. Here we have looked at price volatility and
momentum for the particular stock and for the technology sector. Future work would involve adding
features related to the specific company and related to broader macroeconomic factors. Features
related to the specific company include its Price/Earnings ratio, its market cap, working capital ratio,
cost-of-capital, etc. Features related to broader factors include interest rate, inflation, GDP growth
rate, etc.
6.3. Analyzing Other Sectors and Sizes
Another area for future growth is to apply our model to stocks in other sectors. In this study we
have only focused on the technology sector, specifically 34 mature technology companies. We can
apply our methods to other sectors such as healthcare, retail, service, etc. We can also apply our
methods to mid-size and startup companies, to see how the size of a company affects our model’s
predictive ability.

10
Appendix: Results Data
Here we present the full data table of results.

Table 2: Data showing prediction accuracy summary statistics for each combination of m, n1 , n2
m n1 n2 mean median max min
1 5 5 0.503945 0.508632 0.532537 0.471448
5 10 0.501211 0.501992 0.552457 0.468792
5 20 0.504179 0.507968 0.547145 0.463479
5 90 0.505273 0.508632 0.544489 0.463479
5 270 0.503828 0.507304 0.545817 0.459495
10 5 0.499253 0.496658 0.532086 0.467914
10 10 0.494571 0.494024 0.556441 0.456839
10 20 0.494336 0.498008 0.545817 0.455511
10 90 0.495508 0.499336 0.543161 0.438247
10 270 0.498477 0.498008 0.540505 0.456839
20 5 0.502909 0.503388 0.536585 0.471545
20 10 0.498971 0.500000 0.557201 0.464334
20 20 0.502578 0.505976 0.540505 0.443559
20 90 0.499258 0.496016 0.543161 0.455511
20 270 0.504726 0.511952 0.543161 0.456839
90 5 0.500925 0.502246 0.532934 0.453593
90 10 0.500568 0.502229 0.540862 0.453195
90 20 0.499871 0.500732 0.543192 0.452416
90 90 0.498984 0.495352 0.543161 0.456839
90 270 0.503711 0.504648 0.543161 0.456839
270 5 0.502712 0.509221 0.561475 0.461066
270 10 0.495168 0.498986 0.553753 0.430020
270 20 0.501696 0.501988 0.554672 0.425447
270 90 0.500616 0.496510 0.541012 0.441536
270 270 0.503437 0.506640 0.543161 0.456839
5 5 5 0.523757 0.530708 0.618158 0.411215
5 10 0.525053 0.530040 0.588785 0.452603
5 20 0.528548 0.534713 0.579439 0.403204
5 90 0.526310 0.529372 0.592790 0.401869
5 270 0.530001 0.538051 0.598131 0.401869
10 5 0.524984 0.524194 0.598118 0.419355
10 10 0.521872 0.528037 0.588785 0.427236
10 20 0.521165 0.528037 0.590120 0.403204
10 90 0.515668 0.527370 0.596796 0.404539
10 270 0.525563 0.534045 0.599466 0.404539
20 5 0.531215 0.532698 0.588556 0.435967
20 10 0.527700 0.535183 0.587280 0.441137
20 20 0.533103 0.535381 0.610147 0.419226
20 90 0.531061 0.536716 0.612817 0.380507
20 270 0.540642 0.547397 0.612817 0.385848
90 5 0.528792 0.526355 0.602410 0.403614
90 10 0.525763 0.528401 0.596413 0.445441
90 20 0.532617 0.536819 0.625920 0.387334
90 90 0.529765 0.536048 0.598131 0.401869
90 270 0.532828 0.546061 0.588785 0.401869
270 5 0.548311 0.549587 0.607438 0.479339
270 10 0.548178 0.550102 0.605317 0.449898

11
 270 20 0.549275 0.552104 0.607214 0.446894
270 90 0.534736 0.549209 0.604569 0.363796
270 270 0.540603 0.548064 0.615487 0.401869
10 5 5 0.537200 0.540323 0.596774 0.452957
5 10 0.539295 0.545027 0.602151 0.440860
5 20 0.540283 0.545027 0.618280 0.375000
5 90 0.540006 0.542339 0.610215 0.361559
5 270 0.543683 0.548387 0.607527 0.376344
10 5 0.536058 0.540595 0.595399 0.441137
10 10 0.534472 0.549059 0.607527 0.438172
10 20 0.535421 0.545699 0.622312 0.372312
10 90 0.533286 0.538978 0.623656 0.380376
10 270 0.544711 0.547715 0.603495 0.412634
20 5 0.539377 0.544582 0.620027 0.441701
20 10 0.540030 0.542234 0.632153 0.420981
20 20 0.540560 0.545699 0.650538 0.381720
20 90 0.541469 0.550403 0.618280 0.369624
20 270 0.550166 0.555108 0.662634 0.373656
90 5 0.539097 0.545524 0.603945 0.435508
90 10 0.535746 0.537651 0.620482 0.405120
90 20 0.535477 0.550445 0.637982 0.341246
90 90 0.540639 0.545699 0.622312 0.380376
90 270 0.539097 0.548387 0.615591 0.357527
270 5 0.559069 0.557411 0.653445 0.427975
270 10 0.556089 0.557851 0.652893 0.407025
270 20 0.568052 0.571862 0.663968 0.374494
270 90 0.549645 0.559397 0.643617 0.347518
270 270 0.543169 0.552419 0.615591 0.357527
20 5 5 0.553775 0.559946 0.658038 0.336512
5 10 0.555658 0.566757 0.668937 0.331063
5 20 0.550809 0.555177 0.686649 0.325613
5 90 0.546642 0.564033 0.663488 0.325613
5 270 0.568040 0.580381 0.653951 0.335150
10 5 0.549827 0.551440 0.663923 0.336077
10 10 0.557020 0.566757 0.688011 0.325613
10 20 0.545560 0.555858 0.688011 0.324251
10 90 0.545560 0.560627 0.664850 0.324251
10 270 0.575373 0.581063 0.713896 0.322888
20 5 0.554365 0.556328 0.671766 0.321280
20 10 0.554802 0.558011 0.680939 0.325967
20 20 0.548325 0.556540 0.697548 0.324251
20 90 0.550128 0.574932 0.678474 0.324251
20 270 0.567879 0.580381 0.664850 0.299728
90 5 0.548491 0.565485 0.677966 0.302003
90 10 0.552303 0.566514 0.689602 0.307339
90 20 0.547263 0.564006 0.697289 0.311747
90 90 0.546281 0.568801 0.667575 0.324251
90 270 0.555698 0.577657 0.653951 0.324251
270 5 0.572871 0.575693 0.705757 0.353945
270 10 0.578369 0.580169 0.708861 0.373418
270 20 0.573955 0.586777 0.714876 0.342975
270 90 0.561319 0.596570 0.709386 0.296029
270 270 0.570724 0.580381 0.679837 0.324251

12
90 5 5 0.555679 0.570030 0.849398 0.251506
5 10 0.554571 0.545934 0.849398 0.256024
5 20 0.550053 0.551205 0.849398 0.222892
5 90 0.492470 0.491717 0.849398 0.203313
5 270 0.572821 0.602410 0.814759 0.161145
10 5 0.551415 0.562974 0.848255 0.239757
10 10 0.565556 0.574548 0.849398 0.236446
10 20 0.567328 0.588102 0.849398 0.201807
10 90 0.497387 0.520331 0.849398 0.194277
10 270 0.571536 0.597139 0.774096 0.165663
20 5 0.549760 0.566256 0.852080 0.221880
20 10 0.556845 0.563456 0.847095 0.244648
20 20 0.570030 0.571536 0.849398 0.225904
20 90 0.501329 0.516566 0.849398 0.167169
20 270 0.577826 0.617470 0.796687 0.118976
90 5 0.498324 0.524180 0.911917 0.138169
90 10 0.504684 0.517123 0.912671 0.160959
90 20 0.512280 0.520202 0.910774 0.173401
90 90 0.495748 0.496988 0.849398 0.240964
90 270 0.555679 0.576807 0.774096 0.259036
270 5 0.614993 0.585213 0.939850 0.258145
270 10 0.605926 0.602723 0.893564 0.118812
270 20 0.604149 0.608696 0.915459 0.200483
270 90 0.598991 0.596074 0.911157 0.123967
270 270 0.571580 0.602410 0.774096 0.207831
270 5 5 0.392076 0.331612 0.983471 0.000000
5 10 0.390010 0.321281 0.985537 0.000000
5 20 0.403804 0.376033 0.987603 0.000000
5 90 0.389645 0.297521 0.971074 0.000000
5 270 0.517076 0.482438 0.987603 0.010331
10 5 0.397765 0.354906 0.987474 0.000000
10 10 0.398213 0.354339 0.987603 0.000000
10 20 0.408483 0.372934 0.987603 0.000000
10 90 0.399003 0.317149 0.971074 0.000000
10 270 0.514463 0.488636 0.987603 0.010331
20 5 0.406434 0.373134 0.982942 0.006397
20 10 0.415488 0.360759 0.985232 0.000000
20 20 0.421123 0.387397 0.987603 0.000000
20 90 0.413345 0.359504 0.971074 0.000000
20 270 0.521269 0.491736 0.987603 0.022727
90 5 0.444788 0.408521 1.000000 0.007519
90 10 0.449840 0.474010 1.000000 0.000000
90 20 0.461921 0.457729 0.997585 0.002415
90 90 0.432547 0.358471 0.971074 0.004132
90 270 0.526434 0.534091 0.987603 0.002066
270 5 0.691244 0.705479 1.000000 0.027397
270 10 0.687763 0.732143 1.000000 0.035714
270 20 0.675214 0.713675 1.000000 0.047009
270 90 0.674632 0.731908 1.000000 0.111842
270 270 0.568000 0.621901 0.987603 0.008264

13
References
[1] “S&p dow jones indices,” PDF, March 31 2015.
[2] C. Y. Z. Ben Jacobsen, “Are monthly seasonals real? a three century perspective,” 2012.
[3] ——, “The halloween indicator, ’sell in may and go away’: An even bigger puzzle,” October 2012.
[4] G. F. Bjoern Krollner, Bruce Vanstone, “Financial time series forecasting with machine learning techniques: A
survey,” in European Symposium on Artificial Neural Networks: Computational and Machine Learning, Bruges,
Belgium, April 2010.
[5] J. Brownlee. A tour of machine learning algorithms.
[6] M. Buhmann, “Radial basis function.” Mathematisches Institut, Justus-Liebig-Universität Giessen, 2010.
[7] Y.-A. L. B. Gianluca Bontempi, Souhaib Ben Taieb, Machine Learning Strategies for Time Series Forecasting,
Machine Learning Group, Computer Science Department, Univerite Libre de Bruxelles.
[8] M. Halls-Moore, “Support vector machines: A guide for beginners,” September 2014. [Online]. Available:
http://www.quantstart.com/articles/Support-Vector-Machines-A-Guide-for-Beginners
[9] K. jae Kim, “Financial time series forecasting using support vector machines,” Neurocomputing, vol. 55, 2003.
[10] M. W. W. James H. Stock, Introduction to Econometrics. Addison-Wesley, 2015.
[11] A. Kar, “Stock prediction using artificial neural networks,” Dept. of Computer Science and Engineering, IIT
Kanpur.
[12] S. R. Y. Mayankkumar B Patel, “Stock prediction using artificial neural network,” International Journal of
Innovative Research in Science, Engineering, and Technology, 2014.
[13] S. T. Narasimhan Jegadeesh, “Returns to buying winners and selling losers: Implications for stock market
efficiency,” The Journal of Finance, vol. 48, no. 1, pp. 65–91, March 1993.
[14] D. d. l. F. Rafael Rosillo, Javier Giner, “Stock market simulation using support vector machines,” Journal of
Forecasting, vol. 33, pp. 488–500, July 2014.
[15] V. H. Shah, “Machine learning techniques for stock price prediction,” pp. 1–18, 2007.
[16] Long-Term Unemployment in the Great Recession. U.S. Congress Joint Economic Committee, 2010.
[17] J. Zhang, “Applying time series analysis builds stock price forecast model,” Modern Applied Science, vol. 3, no. 5,
2009.
[18] A. M. Zvi Bodie, Alex Kane, Investments. McGraw-Hill, 2014.

14