Search

Artificial neural networks are being widely used for time series forecasting. In recent years much effort has been made for the development of particle swarm algorithm for the optimization of neural networks. In this paper, the performance of two variants of particle swarm optimization algorithm (Trelea I and Trelea II) for training neural network has been examined with a real data for financial time series forecasting. Results clearly indicated the superiority of swarm based algorithms over the standard backpropagation training algorithm with respect to common performance measures across three forecasting horizons. In particular, with the Trelea II trained model, we obtained 92.48 %, 56.64 %, and 44.66 % decrease in terms of MSE over the standard back-propagation trained neural network for 10 days, 30 days and 60 days ahead forecasts respectively.

In this paper, we address a novel application of financial option pricing theory to the management of distributed computing resources. To achieve the set objective, first, we highlight the importance of finance models for the given problem and explain how option theory fits well to price the distributed grid compute resources. Second, we design and develop a pricing model and generate pricing results based on the trace data drawn from two real grids: one commercial grid Auvergrid and one experimental platform grid LCG. We evaluate our proposed model using various grid compute resources (such as memory, storage, software, and compute cycles) as individual commodities. By carrying out several experiments, a justification of the pricing model is obtained by comparing real behavior to a simulated system based on the spot price for the resources. We further enhanced our model to achieve a desirable balance between Quality of Service (QoS) and profitability from the perspectives of the users and resource operators respectively.

Option pricing is one of the challenging problems in finance. Finding the best time to exercise an option is a even more challenging problem, especially since the price of the underlying assets change rapidly. In this work, we study complex path dependent options by exploiting and extending a novel idea that we proposed earlier using a nature inspired meta-heuristic algorithm. Ant Colony Optimization (ACO). ACO has been used extensively in combinatorial optimization problems and recently in dynamic applications such as mobile ad-hoc networks where the objective is find a shortest path. However, in finance, especially in option pricing, we look for best time to exercise an option. Specifically, we use ants to decide on the best time to exercise so that the holder of the option contract will get the maximum benefit from his/her investment. Our algorithm and implementation suggests a better way to price options than traditional techniques such as Monte Carlo simulation or binomial lattice algorithm. Our pricing results compare very well with other techniques and at the same time the computational cost is reduced to a large extent.

Option pricing is one of the fundamental problems in finance. This chapter proposes a novel idea for pricing options using a nature inspired meta-heuristic algorithm called Ant Colony Optimization (ACO). ACO has been used in many NP-hard combinatorial optimization problems and most recently in self-organized environments in dynamic networks such as ad hoc and sensor networks. The dynamic changes in financial asset prices poses greater challenges to exercise the option at the right time. The dynamic nature of the option pricing problem lends itself very easily in using the ACO technique to the solution of computing option prices. ACO is as intuitive as other techniques such as binomial lattice approach. ACO searches the computational space eliminating areas that may not provide a profitable solution. The computational cost, therefore, tends to decrease during the execution of the algorithm. There has been no study reported in the literature on the use of ACO for pricing financial derivatives. We first study the suitability of ACO in finance and confirm that ACO could be applied to financial derivatives. We propose two ACO based algorithms to apply to derivative pricing problems in computational finance. The first algorithm, named Sub-optimal Path Generation is an exploitation technique. The second algorithm named the Dynamic Iterative Algorithm captures market conditions by using an exploration and exploitation technique. We analyze the advantages and disadvantages of both the algorithms. With both the algorithms we are able to compute the option values and we find that the sub-optimal path generation algorithm outperforms the binomial lattice method. The dynamic iterative algorithm can be used on any random graph and the uncertainties in the market can be captured easily but it is slower when compared to the sub-optimal path generation algorithm.

In this paper, we design and develop a financial options-based model for pricing grid resources. The objective is to strike and maintain an equilibrium between service satisfaction of grid users and profitability of service providers. We explain how option theory fits well to price the grid resources. We price various grid resources such as memory, storage, software, and compute cycles as individual commodities. We carried out several experiments and provide a mapping of our research results based on the spot prices to the expected cost of utilizing the resources from three real grids that reflects their usage pattern. We further enhance our model to achieve the objective of equilibrium between Quality of Service (QoS) and profitability from the perspectives of the users and grid operators respectively.

Summary form only given. Grid computing has developed extensively in recent years for executing computationally resource-intensive applications. Pricing grid resources have become an important problem and few economic models suggest tools for pricing the resources. In this talk, we review some of the economic models and introduce concepts from finance and the similarities between grid resources market and one of the financial markets - options market. We first highlight the importance of finance models over economic models for the given problem and explain how option theory fits well to price the grid resources. We design and develop a quality of service (QoS)-profit equilibrium model for pricing grid resources that is based on finance concepts. With the objective of striking and maintaining equilibrium between service satisfaction and profitability for using grid resources, we evaluate our proposed model using the data from the above real grids. Various grid resources such as memory, storage, software, and compute cycles are seen as individual commodities and pricing of the resources is done in isolation and in combination of various resources. With large number of experiments carried out, a justification of the pricing model is provided by comparing real behavior to a simulated system based on the spot price for the grid resources.

Communication and synchronization are two main latency issues in computing FFT on parallel architectures. Both latencies have to be either hidden or tolerated to achieve high performance. One approach to achieve this is by multithreading. Another approach to tolerate latency is to map data efficiently onto the processors' local memory and exploiting data locality. Indirect swap networks, an idea proposed in VLSI circuits can be efficiently used to compute the butterfly computations in FFT. Data mapping in the swap network topology reduces the communication overhead by half at each iteration. Cell broadband engine (Cell/B.E.)processor is a heterogeneous multicoreprocessor for stream data applications and high performance computing. Its eight SIMD processing elements, synergistic processor elements (SPEs), provide multi-folded parallelism. In this paper, we investigate the improved Cooley-Tukey FFT algorithm based on indirect swap network, and design the parallel algorithm taking into consideration all the features of the Cell/B.E. architecture. The performance results show that the new algorithm on Cell/B.E. is 3.7 faster than the cluster for 4K input data size and 6.4 faster than the cluster for 16K input data size at the processor level.

A computational grid ensures the on-demand delivery of computing resources, in a security-aware, shared, scalable, and standardsbased computing environment. A major concern is how to evolve a general and an encompassing framework that guarantees users' satisfaction for using grid computing resources measured as Quality of Services (QoS). To obtain a higher QoS, effective QoS perceived by subscribers (users) must conform to specified QoS agreements in the Service Level Agreements (SLAs) document - a legal contract between the Grid Services Provider (GSP) and users. Sometimes, the effective user QoS does not conform to the specifications in the SLA because of the vagueness in linguistic definitions in the SLA. Existing approaches overcommitted grid resources to satisfy QoS requirement in SLA. In this paper, we propose a fuzzy logic framework for calibrating a user QoS that addresses the vagueness in linguistic definitions of the SLA document without overcommitting grid resources.

Mobile ad hoc networks (MANETS) are infrastructureless network consisting of mobile nodes, with constantly changing topologies, that communicate via a wireless medium. Therefore, routing is a challenging issue in MANETs. Recently, nature inspired algorithms have been explored as means of finding an efficient solution to this routing problem. In this paper, we develop an improved routing algorithm for MANETs based on ant colony optimization (ACO) inspired by real ants. The performance of the routing algorithm is evaluated through simulation and is compared to an existing well known MANET routing protocol, ad hoc on-demand distance vector (AODV). Several performance metrics are considered in different scenarios with varying mobility levels and traffic load.

Computing option prices is a challenging problem. Finding the best time to exercise an option is a even more challenging problem. One has to be watchful for the price changes in the market place and act at the right time. That is, prices need to be policed. This paper proposes a novel idea for pricing options using a nature inspired meta-heuristic algorithm, Ant Colony Optimization (ACO). ACO has been used extensively in combinatorial optimization problems and recently in dynamic applications such as mobile ad-hoc networks. Specifically, we adapt the general ACO algorithm to apply to a totally different application, computational finance, in the current study. We police the prices using ants to decide on the best time to exercise so that the holder of the option contract will get the maximum benefit out of his/her investment. Our algorithm and implementation suggests a better way to price options than traditional numerical techniques such as binomial lattice algorithm. From our results we conclude that reactive ants may be best suited for long-dated options whose performance can still be improved.

Instance based learning is a class of data mining learning paradigms that applies specific cases or experiences to new situations by matching known cases and experiences with new cases. This paper presents an application of the instance-based learning algorithm for predicting daily stock index price changes of the S&P 500 stock index between October 1995 and September 2000, given the daily changes in the exchange rate of the Canadian Dollar, the Pound Sterling, the French Franc, the Deutsche Mark and the Yen, the monthly changes in the consumer price index, GDP, and the changes in the monthly rates of certificates of deposit. The algorithm is used to predict an increase, decrease or no change in the S&P 500 stock index between a business day and the previous business day. The predictions are carried out using the IB3 variant of the IBL algorithms. The objective is to determine the feasibility of stock price prediction using the IB3 variant of the IBL algorithms. Various testing proportions and normalization methods are experimented with to obtain good predictions.

In this study, we model pricing of grid/distributed computing resources as a problem of real option pricing. Grid resources are non-storable compute commodities (eg., CPU cycles, memory, etc). The non-storable characteristic feature of the grid resources hinders it from fitting into a risk-adjusted spot price model for pricing financial options. Grid resources users pay upfront to acquire and use grid compute cycles in the future, for example, six months. The user expects a high and acceptable degree of satisfaction expressed as the Quality of Service (QoS) assurance. This requirement further imposes service constraints on the grid because it must provide a user-acceptable QoS guarantee to compensate for the upfront value. This study integrates three threads of our research; pricing the grid compute cycles as a problem of real option pricing, modeling grid resources spot price using a discrete time approach, and addressing uncertainty constraints in the provision of QoS using fuzzy logic. We have proved the feasibility of this model through experiments and we have presented some of our pricing results and discussed them.

Monte Carlo (MC) simulation is one of the popular approaches for approximating the value of options and other derivative securities due to the absence of straightforward closed form solutions for many financial models. However, the slow convergence rate, O(N/sup - 1/2/) for N number of samples of the MC method has motivated research in quasi Monte-Carlo (QMC) techniques. QMC methods use low discrepancy (LD) sequences that provide faster, more accurate results than MC methods. In this paper, we focus on the parallelization of the QMC method on a heterogeneous network of workstations (HNOWs) for option pricing. HNOWs are machines with different processing capabilities and have distinct execution time for the same task. It is therefore important to allocate and schedule the tasks depending on the performance and resources of these machines. We present an adaptive, distributed QMC algorithm for option pricing, taking into account the performances of both processors and communications. The algorithm distributes data and computations based on the architectural features of the available processors at run time. We implement the algorithm using mpC, an extension of ANSI C language for parallel computation on heterogeneous networks. We compare and analyze the performance results with different parallel implementations. The results of our algorithm demonstrate a good performance on heterogenous parallel platforms.

In this paper, we develop an efficient system to detect outliers from real-life financial time series comprising of security prices. Our system consists of a data mining algorithm and a statistical algorithm. When applying each of these two algorithms individually, we observed its strengths and weaknesses. To overcome the weaknesses of the two algorithms, we combine the algorithms together. By so doing, we efficiently detect outliers from the financial time series. Moreover, the resulting (processed) datasets can then be used as input for some financial models used in forecasting future security prices or in predicting future market behaviour. This shows an alternative role of our outlier detection system—serving as a pre-processing step for other financial models.

The problem of growing computational complexity in finance industry demands manageable, high-speed and real-time solutions in solving complex mathematical problems such as option pricing. In option trading scenario, determining a fair price for options “any time” and “any-where” has become vital yet difficult computational problem. In this study, we have designed, implemented, and deployed architecture for pricing options on-line using a hand-held device that is J2ME-based Mobile computing-enabled and is assisted by web mining tools. In our architecture, the client is a MIDP user interface, and the back end servlet runs on a standalone server bound to a known port address. In addition, the server uses table-mining techniques to mine real-time data from reliable web sources upon the mobile trader's directive. The server performs all computations required for pricing options since mobile devices have limited battery power, low bandwidth, and low memory. To the best of our knowledge, this is one of the first studies that facilitate the mobile-enabled-trader to compute the price of an option in ubiquitous fashion. This architecture aims at providing the trader with various computational techniques to avail (to provide results from approximate to accurate results) while on-the-go and to make important and effective trading decisions using the results that will ensure higher returns on investments in option trading.

Computational requirements for solving models of financial derivatives, for example, the option pricing problems, are huge and demand efficient algorithms and high performance computing capabilities. This demand has been rekindled by the recent developments in the mobile technology making wireless trading a possibility. In this paper, we focus on the development of a Monte-Carlo algorithm on a modern multi-core chip architecture, Cyclops-64 (C64) under development at IBM as the experimental platform for our study in pricing options. The timing results on C64 show that various sets of simulations could be done in a real-time fashion while yielding high performance/price improvement over traditional microprocessors for finance applications.

The sine and cosine functions used as the bases in Fourier analysis are very smooth (infinitely differentiable) and very broad (nonzero almost everywhere on the real line), and hence they are not effective for representing functions that change abruptly (jumps) or have highly localized support (diffusive). In response to this shortcoming, there has been intense interest in recent years in a new type of basis functions called wavelets. A given wavelet basis is generated from a single function, called a mother wavelet or scaling function, by dilation and translation. By replicating the mother wavelet at many different scales, it is possible to mimic the behavior of any function; this property of wavelets is called multiresolution. Wavelet is a powerful integral transform technique for studying many problems including financial derivatives such as options. Moreover, the approximation error is much smaller than that of the truncated Fourier expansion. Therefore, one can get better approximation of a function at jump discontinuity with the use of wavelet expansion rather than Fourier expansion. In the current study, we employ wavelet analysis to option pricing problem manifested as partial differential equation (PDE) with jump characteristics. We have used wavelets to develop an optimum finite differencing of the differential equations manifested by complex financial models. In particular, we apply wavelet optimized finite-difference (WOFD) technique on the partial differential equation. We describe how Lagrangian polynomial is used to approximate the partial derivatives on an irregular grid. We then describe how to determine sparse and dense grid with wavelets. Further work on implementation is going on.

In this paper, we develop parallel algorithms for pricing American-style Asian options employing binomial tree method. We describe the algorithm, explain the complexities, and study the performance. We have extended our algorithm to handle Asian options with up to 10 underlying assets and shown that the multi-asset Asian options offer a better problem for parallel computation.

Efficient data distribution is important to overcome latencies in distributed memory multiprocessors. In this paper we have studied the distributed implementation of the FFT algorithm using the ISN topology to improve data locality. The algorithm is implemented on the Beowulf clusters using MPI. We obtain 20% better performance compared to the butterfly network.

A mobile ad hoc network (MANET) consists of mobile wireless nodes that communicate in a distributed fashion without any centralized administration. The nodes instantaneously and dynamically form a network on the fly when it is needed. We define an irregular application as one that changes the network dynamically during runtime, exhibits chaotic load balancing among the processors and unpredictable communication behavior among the nodes during runtime. An ad hoc network has all these characteristics and hence could be considered as an irregular application from the parallel computing perspective. In this paper, we design an on-demand routing algorithm called source update for MANET using a meta-heuristic based on the ant colony optimization (ACO) search technique. We develop a mechanism to detect cycles, parallelize this algorithm on a distributed memory machine using MPI, and study the performance of the parallel algorithm. On a distributed network of workstations, we obtain a relative speedup of 7 with 10 processors.

We focus on the neural network training problem that could be used for price forecasting or other purposes in finance. We design and develop four different parallel and multithreaded backpropagation neural network algorithms: neuron and training set parallelism on a distributed memory architecture using MPI; loop-level (fine-grain) and coarse-grained parallelism in shared memory architecture using OpenMP. We have conducted various experiments to study the performance of these algorithms and compared our results with a traditional autoregression model to establish accuracy of our results. The comparison between our MPI and OpenMP results suggest that the training set parallelism performs better than all the other types of parallelism considered in the study.

Fast Fourier Transform (FFT) has been used in many scientific and engineering applications. In the current study, we have tried to improve a recently proposed model of FFT for pricing financial derivatives so as to help designing an efficient parallel algorithm. We have then developed a new parallel algorithm to compute the FFT using a swapping technique that exploits data locality, and hence showed higher efficiency of this algorithm. We have tested our algorithm on 20 node SunFire 6800 high performance computing system and compared the new algorithm with the traditional Cooley-Tukey algorithm. As an example, we have also plotted the calculated option values for various strike prices with a proper selection of log strike-price spacing to ensure fine-grid integration for FFT computation as well as to maximize the number of strikes lying in the desired region of the asset price.

A mobile ad hoc network (MANET) consists of mobile wireless nodes that communicate in a distributed fashion without any centralized administration. The nodes instantaneously and dynamically form a network on the fly when it is needed. We define an irregular application as one that changes the network dynamically during runtime, exhibits chaotic load balancing among the processors and unpredictable communication behavior among the nodes during runtime. An ad hoc network has all these characteristics and hence could be considered as an irregular application from the parallel computing perspective.In this chapter, we describe the design of an on-demand routing algorithm called source update for MANETs using a metaheuristic based on the ant colony optimization (ACO) search technique. We develop a mechanism to detect cycles, parallelize this algorithm on a distributed memory machine using MPI, and study the performance of the parallel algorithm. We report the performance of this algorithm on a distributed network of workstations.

In this paper, we develop a multithreaded algorithm for pricing simple options and implement it on a 8 node SMP machine using MIT's supercomputer programming language Cilk. The algorithm dynamically creates lots of threads to exploit parallelism and relies on the Cilk runtime system to distribute the computation load. We present both analytical and experimental results and our results explain how Cilk could be used effectively to exploit parallelism in the given problem. The analytical results show that our algorithm has a very high average parallelism and hence Cilk is the target paradigm to implement the algorithm. We conclude from our implementation results that the size of the threads, the number of threads created, the load balancer the cost of spawning a thread are parameters that must be considered while designing the algorithm on the Cilk platform.

We develop parallel algorithms for pricing a class of multidimensional financial derivatives employing a binomial lattice approach. We describe the algorithms, explain their complexities, and study their performance. The limitations posed by the problem size on the recursive algorithm and the solution to overcome this problem by an iterative algorithm are explained through experimental results using MPI. We first present analytical results for the number of computations and communications for both the algorithms. Since the number of nodes in a recombining lattice grows linearly with the problem size, it is feasible to price long-dated options using a recombining lattice. We have extended our algorithm to handle derivatives with many underlying assets and shown that the multi-asset derivatives offer a better problem for parallel computation. This is very important for the finance industry since long-dated derivatives with many underlying assets are common in practice.

In this paper, we study multithreaded algorithms for pricing American Style options. We describe the algorithms, explain their relative complexities, and study their performance. The binomial lattice problem has been formulated in two distinct ways. In the first approach, the recursive algorithm, we establish a parent-child relationship between threads while fully exploiting the inherent parallelism. The second approach, the iterative algorithm, follows a data-flow model based on the producer-consumer style of programming. We implement the algorithms on the EARTH platform. The limitations posed by the problem size on the recursive algorithm and the solution to overcome this problem by the iterative algorithm are explained through the performance results. We have then extended these algorithms to study complicated options with dividend paying underlying assets and reported the performance results.

Pricing of derivatives is one of the central problems in Computational Finance. Since the theory of derivative pricing is highly mathematical, numerical techniques such as lattice approach, finite-difference and finite-element techniques among others have been resorted in the past. Recently Fast Fourier Transform (FFT) have been used for such applications as derivative pricing. In the current work, we develop a parallel algorithm for FFT and implement it to price options. Our main aim is to study the performance of this algorithm. For a data size of N and P processors, a blocked data distribution for the algorithm in general produces log(N) - log(P) iterations of local communications and log(P) iterations of remote communications. Therefore, the algorithm is divided into two parts: local and remote. In the local algorithm, the processors perform the computations on their locally partitioned data elements without any communications. In the case of remote algorithm, the processors perform the computation on the local data elements with remote communications. In this paper we focus on the remote communication and computation aspect of the algorithm. We discuss the performance of our algorithm and the results (in general terms) from FFT algorithm and binomial tree algorithm developed and implemented for the same/similar problem. We make some general observation on these two algorithms.

This paper reports a study of sparse matrix vector multiplication on a parallel distributed memory machine called EARTH, which supports a fine-grain multithreaded program execution model on off-the-shelf processors. Such sparse computations, when parallelized without graph partitioning, have a high communication to computation ratio, and are well known to have limited scalability on traditional distributed memory machines. EARTH offers a number of features which should make it a promising architecture for this class of applications, including local synchronizations, low communication overheads, ability to overlap communication and computation, and low context-switching costs. On the NAS CG benchmark Class A inputs, we achieve linear speedups on the 20-node MANNA platform, and an absolute speedup of 79 on 120 nodes on a simulated extension. The speedup improves to 90 on 120 nodes for Class B. This is achieved without inspector/executor, graph partitioning, or any communication minimization phase, which means that similar results can be expected for adaptive problems.

Finite automata, in which each transition is augmented with an output label in addition to the familiar input label, are considered finite-state transducers. Transducers have been used to analyze some fundamental issues in bioinformatics. Weighted finite-state transducers have been proposed to pairwise alignments of DNA and protein sequences; as well as to develop kernels for computational biology. Machine learning algorithms for conditional transducers have been implemented and used for DNA sequence analysis. Transducer learning algorithms are based on conditional probability computation. It is calculated by using techniques, such as pair-database creation, normalization (with Maximum-Likelihood normalization) and parameters optimization (with Expectation-Maximization - EM). These techniques are intrinsically costly for computation, even worse when are applied to bioinformatics, because the databases sizes are large. In this work, we describe a parallel implementation of an algorithm to learn conditional transducers using these techniques. The algorithm is oriented to bioinformatics applications, such as alignments, phylogenetic trees, and other genome evolution studies. Indeed, several experiences were developed using the parallel and sequential algorithm on Westgrid (specifically, on the Breeze cluster). As results, we obtain that our parallel algorithm is scalable, because execution times are reduced considerably when the data size parameter is increased. Another experience is developed by changing precision parameter. In this case, we obtain smaller execution times using the parallel algorithm. Finally, number of threads used to execute the parallel algorithm on the Breezy cluster is changed. In this last experience, we obtain as result that speedup is considerably increased when more threads are used; however there is a convergence for number of threads equal to or greater than 16.

In this study, we model pricing of grid/distributed computing resources as a problem of real option pricing. Grid resources are non-storable compute commodities (e.g., CPU cycles, memory, etc.). The non-storable characteristic feature of the grid resources hinders it from fitting into a risk-adjusted spot price model for pricing financial options. Grid resources users pay upfront to acquire and use grid compute cycles in the future, for example, six months. The user expects a high and acceptable degree of satisfaction expressed as the quality of service (QoS) assurance. This requirement further imposes service constraints on the grid because it must provide a user-acceptable QoS guarantee to compensate for the upfront value. This study integrates three threads of our research; pricing the grid compute cycles as a problem of real option pricing, modelling grid resources spot price using a discrete time approach, and addressing uncertainty constraints in the provision of QoS using fuzzy logic. We have proved the feasibility of this model through experiments and we have presented some of our pricing results and discussed them.

Analysis of gas-particle nozzle flows is carried out with attention to the effect of dust particles on the vibrational relaxation phenomena and consequent effect on the lasing performance of a gas mixture. The relaxation of the particBles during the nozzle expansion process is taken into account simultaneously. The governing equations of the two-phase nozzle flow have been transformed into similar form and general correlating parameters have been obtained. It is shown from the present analysis that the size and the loading ratio of particles in the mixture are the two important controlling parameters in optimizing the lasing performance. The effect of the particles on the optimum performance of a two-phase flow gasdynamic laser is insignificant for smaller loading ratios and for particles of large size. However, the effect is observed to be adverse when the particles of smaller size are involved or when the loading ratio is high. 24 refs.

In this paper, we study the option pricing problem, one of the prominent and challenging problems in computational finance. Using the Pade approximation, we have developed a second order L0 stable discrete parallel algorithm for experimentation on advanced architectures. We have implemented the sequential version of this algorithm and evaluated the European Options. Numerical results are compared with those obtained using other commonly used numerical methods and shown that the new algorithm is robust and efficient than the traditional schemes. Also, using explicit Forward Time Centered Space (FTCS) on the reduced Black-Scholes partial differerential equation, we report pricing of European options. We have done our experiments on a shared memory multiprocessor machine using OpenMP and report a maximum speedup of 3.43 with 16 threads.