Browsing by Author "Akyildirim, E."

Now showing 1 - 5 of 5

Open Access
Big data analytics, order imbalance and the predictability of stock returns
(Elsevier, 2021-09-24) Akyildirim, E.; Şensoy, Ahmet; Gulay, G.; Corbet, S.; Salari, Hajar Novin
Financial institutions have adopted big data to a considerable extent to provide better investment decisions. Consequently, high-frequency algorithmic traders use a vast amount of historical data with various statistical models to maximize their trading profits. Until recently, high-frequency algorithmic trading was the domain of institutional traders with access to supercomputers. Nowadays, any investor can potentially make high-frequency trades because of easy access to big data and software to analyze and execute trades. With that in mind, Borsa Istanbul introduced real time big data analytics as a product to its customers. These analytics are derived in real time from order book and trade data and aim to level the playing field between investment firms and retail traders. Using classical benchmark models in the literature, we show that Borsa Istanbul’s order imbalance-based data analytics are useful in predicting both time-series and cross-sectional intraday excess future returns, proving that this product is extremely beneficial to market participants, particularly day traders.
Open Access
Extending the merton model with applications to credit value adjustment
(Springer Link, 2023-07) Akyildirim, E.; Hekimoglu, A. A.; Şensoy, Ahmet; Fabozzi, F. J.
Following the global financial crisis, the measurement of counterparty credit risk has become an essential part of the Basel III accord with credit value adjustment being one of the most prominent components of this concept. In this study, we extend the Merton structural credit risk model for counterparty credit risk calculation in the context of calculating the credit value adjustment mainly by estimating the probability of default. We improve the Merton model in a variance-convoluted-gamma environment to include default dependence between counterparties through a linear factor decomposition framework. This allows one to tackle dependence through a systematic common component. Our set-up allows for easier, faster and more accurate fitting for the credit spread. Results confirm that use of the variance-gamma-convolution clearly solves the vanishing credit spread problem for short time-to-maturity or low leverage cases compared to a Brownian motion environment and its modifications.
Open Access
Forecasting high‐frequency excess stock returns via data analytics and machine learning
(John Wiley and Sons Inc., 2023-01) Akyildirim, E.; Nguyen, D.K.; Şensoy, Ahmet; Šikić, M.
Borsa Istanbul introduced data analytics to present additional information about its market conditions. We examine whether this product can be utilized via var ious machine learning methods to predict intraday excess returns. Accordingly, these analytics provide significant prediction ratios above 50% with ideal profit ratios that can reach up to 33%. Among all the methods considered, XGBoost (logistic regression) performs better in predicting excess returns in the long‐term analysis (short‐term analysis). Results provide evidence for the benefits of both the analytics and the machine learning methods and raise further discussion on the semistrong market efficiency.
Open Access
Prediction of cryptocurrency returns using machine learning
(Springer, 2021-02) Akyildirim, E.; Goncu, A.; Sensoy, Ahmet
In this study, the predictability of the most liquid twelve cryptocurrencies are analyzed at the daily and minute level frequencies using the machine learning classification algorithms including the support vector machines, logistic regression, artificial neural networks, and random forests with the past price information and technical indicators as model features. The average classification accuracy of four algorithms are consistently all above the 50% threshold for all cryptocurrencies and for all the timescales showing that there exists predictability of trends in prices to a certain degree in the cryptocurrency markets. Machine learning classification algorithms reach about 55–65% predictive accuracy on average at the daily or minute level frequencies, while the support vector machines demonstrate the best and consistent results in terms of predictive accuracy compared to the logistic regression, artificial neural networks and random forest classification algorithms.
Open Access
Statistical arbitrage in jump-diffusion models with compound poisson processes
(Springer Nature, 2021-02-26) Akyildirim, E.; Fabozzi, J.F.; Goncu, A.; Sensoy, Ahmet
We prove the existence of statistical arbitrage opportunities for jump-diffusion models of stock prices when the jump-size distribution is assumed to have finite moments. We show that to obtain statistical arbitrage, the risky asset holding must go to zero in time. Existence of statistical arbitrage is demonstrated via ‘buy-and-hold until barrier’ and ‘short until barrier’ strategies with both single and double barrier. In order to exploit statistical arbitrage opportunities, the investor needs to have a good approximation of the physical probability measure and the drift of the stochastic process for a given asset.