Stocks are paths to quick riches without putting in a lot of effort. Yet, it could doom one without effort also. It seems that some people consider stock investments/speculation as a vital part of their life. Good stocks could transfer to high wealth - which will lead to early retirement and hence a higher standard of living. This is the core reason that I decided to do this topic - especially in Thailand, where speculation is very evident due to the low brokerage transaction costs.
There are several types of investment analysis. It could be split into two main camps: technical analysis and fundamental analysis. This report focuses on the latter. However, it is helpful to give an overview of the former.
According to Wikipedia, Technical analysisÂ is aÂ security analysisÂ discipline for forecasting the future direction of prices through the study of past market data, primarily price and volume, and aÂ quantitative analystÂ is a person who works in finance using numerical or quantitative techniques. These technical analyses try to find some kind of past price patterns and exploit them for their benefits. Examples of the schools include candlestick charting,Â Dow Theory, andÂ Elliott wave theory. Technical analyses differ in their methods - some advocate a strict mathematical rule while others succumb to subjective measurements. Despite the difference in their methods, the three main principles that the technical analyses share are 1) market action discounts everything; 2) prices move in trends; 3) history tends to repeat itself.
Get your grade
or your money back
using our Essay Writing Service!
Fundamental analysis of a business, on the other hand, involves analyzing itsÂ financial statementsÂ and health, its management and competitive advantages, and itsÂ competitorsÂ andÂ markets. There are two basic approaches one can use when analyzing stock: bottom up analysis and top down analysis. The bottom up investor starts analyzing a specific business; in contrast, a top down investor begins with global economics, and zooms down to a particular sector. The goal of fundamental analysis is to conduct a company stock valuation and forecast its probable price evolution, to make a projection on the business performance, to evaluate its management and make internal business decisions, and to calculate its credit risk. Fundamental analysis states that markets are liable to misprice a security in the short run but that the correct price will be reached. Hence, it is profitable to take advantage of the mispriced security.
II. Theoretical Framework
The main point of this research paper is to prove that econometrics is not able to forecast the results of the stock market to the degree that it doesn't allow the short term speculator to earn significant profits. Therefore, it is useless to use econometrics, which is what some chartists do, to forecast stock markets. I have included lags from 1 to 3 in the stock market and use the SET 100, SET 50, and the MAI to forecast the SET INDEX next month. As you can see, there are no significant results at all. The reason that I chose only 3 lags for my project is that the forecast results got worse if I put many lags in, as could be seen in the Akaike Criterion.
Thus, my project paper's main objective revolves upon analyzing the company's fundamentals before deciding to put money into the market.
III. Literature Review
Heteroscedasticity, or unequal variance, has become the fascination of financial forecasters. Stock prices are known to have periods of extreme fluctuation, or volatility, followed by periods of relative quiet. In portfolio management, balancing risk and reward calls for forecasting of the relative variability of stocks and other financial instruments that could be part of the portfolio.
With time-series data, heteroscedasticity can be viewed as dynamic and can be modeled as a function of time. Financial analysts have devoted much study to changes in variance, or volatility, in stock prices and in series, such as the daily closing values of the S&P 500 share index. Rather than search for variables that might explain these changes in volatility, the typical approach is to model dynamic volatility with one of the family of autoregressive conditional heteroscedasticity (ARCH) models first proposed by Engle (1982). A test for dynamic heteroscedasticity is therefore a test to see whether an ARCH model should be developed. An ARCH model is another example of a varying-parameter approach. Like the varying-parameter approaches discussed earlier, ARCH modeling adds a layer of complexity and sometimes leads to improved point and probability forecasts.
Always on Time
Marked to Standard
Should the forecaster ignore heteroscedasticity? Or just ignore the test for dynamic heteroscedasticity and model it anyway? If so, which of the growing number of variants should he or she model? Heteroscedasticity probably should not be ignored, but empirical evidence is scarce on all these issues. While the literature on dynamic heteroscedasticity is quite large, forecast comparisons are both and recent.
In comparisons of whether the improved efficiency of ARCH-type specifications improved point forecasts, three studies ran in favor (Barrett 1997, Bera & Higgins 1997, Christou, Swamy & Tavlas 1996) and one against (Alexander 1995). The first two studies used GARCH models, the last two ARCH models. (GARCH, or Generalized ARCH, allows variables that explain variance to be included.) Christou, Swamy and Tavlas (1996), found that an ARCH model was more accurate than random walk forecasts for four of five weekly series (returns on financial assets denominated in five major currencies). They also found that a random coefficient (RC) model was even better, and an extended-ARCH model was the most accurate of all. (Extended-ARCH permits some of the parameter variation in the RC model.) In contrast, Alexander (1995) found that ARIMA forecasts were more accurate than ARCH forecasts for the quarterly earnings of about 300 companies.
Turning to volatility forecasts, five studies favored ARCH or GARCH models (Akgiray 1989, Brailsford & Faff 1996, McCurdy & Stengos 1991, Myers & Hanson 1993, Noh, Engel & Kane 1993) while four favored other methods (Batchelor & Dua 1993, Campa & Chang 1995, Figlewski 1994, Frennberg & Hansson 1995) and one found no difference except at very short (one week) horizons (West & Cho 1995). In all but two of the studies (Batchelor & Dua 1993, Brailsford & Faff 1996) the authors used GARCH models. Competing methods were of several kinds, including historical and implied volatilities, nonparametric estimation, univariate autoregressive models, and regression. A study=s conclusion appears to be unrelated to the authors= choice of competing method. With such limited and conflicting evidence, we are unable to make useful recommendations about when heteroscedasticity corrections should be made or even whether they are worth making at all.
IV. Data Description & Summary
In order to do a time series analysis, all variables and the error term must be stationery. The original data and its error term is not stationery with a Durbin-Watson value of 0.266570. After saving the residuals and computing the Augmented Dickey Fuller Test with one lag, the P-Value is 0.6393 for the stochastic trend and 0.765 for the deterministic trend (with unit root null hypothesis). Since the null hypothesis cannot be rejected, the data is not stationery. The first and the log differences and the error term of the SET yield stationery results in the time-series graph. The results are shown below with standard OLS. There is likely to be multicollinearity too since the SET 50 is included in the SET 100 (i.e. the top 50 companies are in the top 100). However, I've already took care of that by transformation of the variables (i.e. first differences, log differences, etc.).
V. Econometric Model
The reason that I decided to use regression analysis instead of correlation analysis is because that the regression analysis is used to forecast the future. It would be nice to see the effect of the top 50 and top 100 SET companies and the MAI stock index affecting the SET index fund in the future month. Graham also said that you could outperform a lot of funds just by only buying index fund (SET in this case). If that's true, it would be nice to be able to forecast the index fund by the SET 100, SET 50, and the MAI index.
I essentially created three models using OLS for each of the first differences and the log differences. The main model is SET = Beta1 + Beta2SET100(-t) + Beta3SET50(-t) + Beta4MAI(-t). The SET Index is not lagged. All the models are derived from this equation. (-1) means lag 1 month. "t" is the number of the model. For example, if t =2, the lag of the independent variables all equal two while the dependent variable (the SET Index) has no lags.
VI. Estimation & Results
It could be seen in all of the models that none of the independent variable with lags is good forecasters of the regression models. In Model 2 (first differences with lags two), the SET 100 and 50 have significant results - more likely of a fluke, since there's no reason to back the significant results up. All of the models yield very low adjusted R-Squares. The best of which is the Model 1 (1st log dif.) with an adjusted R-squared of 0.021767, and a root mean squared error for forecasting of 0.063206.
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
The summary of the econometric models is that the SET 100, SET 50, and the MAI indexes cannot and shouldn't be used for forecasting the stock market. You could now be convinced not to rely on econometric chartists, which will save you a lot of time and money. Normally, a huge SET Index rise is due to foreign investors, who will mostly buy stocks from the banking and the energy sectors; undoubtedly, these companies are in the tops of the market.
As always, do not rely on the forecasts since there are limitations involved. If you could forecast the market by using regression analysis, the econometric academics would've been rich already. A forecast is merely helpful only to the degree of probabilities. No chartists could actually forecast everything precisely and correctly. The difference between chartists and value investors is that the latter allowed for a margin for error while the former don't.
If everyone thinks in the same way and rationally, all the inefficiencies in all the markets will be rid off. Warren Buffett says that by spilling out his techniques, he loses the chance of making a fortune. However, he still educates everyone who cared to listen to him - he is one of the world's greatest philanthropist, having pledged to give away 85% of his fortune to the Gates Foundation.
If everyone follows my guide that I assembled from the leading professional stock players' point of view, I see no reason why anyone would suffer a great loss in the stock market.
Finally, for anyone who wished to invest in stocks, I suggest that they read the following books cited in the references. I am very thankful for these books since some of my ideas in the paper came from these people.