Using the methodology developed by Caginalp and Laurent (1998), this study tests the predictive ability of eight three-day reversal candlestick patterns on a sample of sixty listed shares on the Johannesburg Stock Exchange (JSE) from 1 January 2010 to 31 December 2019. The study further investigates the predictability of the patterns on three sub-sectors of the JSE; this included the resource, financial and industrial sectors. As the final step, the study tests the use of the candlestick pattern as a trading strategy before and after costs. To the authors knowledge, it is the first study of candlestick patterns on South Africa's JSE. However, similar studies have been conducted in other emerging markets such as Thailand, Taiwan and Brazil. The study finds that only one of the candlestick patterns, the Three Outside Down pattern, has predictive ability on the JSE. However, when looking at the JSE sub-sectors, the results are more divergent. Candlestick patterns have no predictive ability on the resources and financial sector but on the industrial sector three patterns (Three Inside Up, Three Black Crows and Morning Star) were significant at the ten percent level of significance; and the Three Outside Up and Three Outside Down were at the five percent level of significance. The study finds that after fees the use of candlestick patterns is unable to outperform passive benchmarks. The evidence suggests some violations of the Random Walk Theory, but the study finds that the JSE and its sub-indices are weak-form efficient in terms of the Efficient Market Hypothesis. The Thai and Brazilian markets were also found to be weak-form efficient.

This dissertation highlights the performance comparison between two popular contemporary consumer loan credit scoring techniques, namely logistic regression and classification trees. Literature has shown logistic regression to perform better than classification trees in terms of predictiveness and robustness when forecasting consumer loan default events over standard twelve-month outcome periods. One of the major shortcomings with classification trees is its tendency to overfit data eroding its robustness, making it vulnerable to underlying population characteristic shifts. Classification trees remains a popular technique due to its ease of application (algorithm machine learning basis) and model interpretation. Past research has found classification trees to perform marginally better than logistic regression with respect to predictiveness and robustness when modelling short term consumer credit default outcomes related to previously unseen new customer credit loan applications. This dissertation independently tested this finding on reloan consumer loan data, repeat customers who renewed loan facilities at a significant South African micro lender. This dissertation tests the finding if the classification tree technique would outperform logistic regression when modelling this new type of loan data. Credit scoring models were built and tested for each respective technique across identical data sets with the intent to eliminate bias. Robustness tests were constructed via careful iterative data splits. Performance tests measuring predictiveness and robustness were conducted via the weighted sums of squared error evaluation approach. Results reveal logistic regression to outperform classification trees on predictiveness and robustness across the designed uniform iterative data splits, which suggests that logistic regression remains the superior technique when modelling short term credit default outcomes on reloan consumer loan data.

This study examines the relationship between foreign exchange reserve, Pula exchange rate and inflation in Botswana over the period 1995-2020. The period covered contains recent data on level of foreign exchange reserves through global events like Covid-19 pandemic in which significant drawdowns in foreign reserves were experienced and not covered in prior studies. Secondary time series data was sourced from Bank of Botswana financial statistics bulletin and a linear regression was run for two models using R studio statistical software. Unit root and correlation tests were run on the data to ensure the variables were stationery and error terms correlation was eliminated in the time series. Regression equation showed that the relationship between foreign exchange reserves and inflation was negative. Similarly, the second regression revealed that the correlation between foreign exchange reserves and foreign exchange rates was negative. Empirical results further revealed that foreign exchange reserves have no statistically significant impact on inflation and exchange rates in Botswana. The statistically insignificant relationship results between the variables implies that foreign exchange reserves have not significantly influenced exchange rate and inflation which may be due to sterilization by the central bank, therefore, other factors may be responsible for changes in exchange rate and inflation locally. This may indicate that the monetary policy framework which requires foreign exchange reserves have served the country well as it maintained a stable exchange rate and did not stir inflation. However, monetary authorities should note that the current framework may not be sustainable given the tremendous pressure that foreign reserved faced in recent years and a move to a floating exchange regime should be considered as a long-term policy goal. The negative correlation of foreign reserves with inflation and exchange rate further suggest to the monetary authorities that Botswana's economy may be influenced by endogenous monetary policies rather than external variables. The relationship between exchange rates and foreign reserves is consistent with elasticity approach and economic theory of modern mercantilism which predict a negative relationship between the two variables. This may provide monetary authorities and policy makers with a framework that explains the link of Botswana's foreign reserves and exchange rates.