Appears in CIFEr '97, March 23-25, New York City HIGH FREQUENCY TIME SERIES ANALYSIS AND PREDICTION USING MARKOV MODELS. Constantine P.

Size: px
Start display at page:

Download "Appears in CIFEr '97, March 23-25, New York City HIGH FREQUENCY TIME SERIES ANALYSIS AND PREDICTION USING MARKOV MODELS. Constantine P."

Transcription

1 Appears in CIFEr '97, March 23-25, New York City HIGH FREQUENCY TIME SERIES ANALYSIS AND PREDICTION USING MARKOV MODELS Constantine P. Papageorgiou Center for Biological and Computational Learning and Articial Intelligence Laboratory Massachusetts Institute of Technology Cambridge, MA 2139 USA Abstract There has been a surge in interest in the analysis and prediction of high frequency time series in recent years. We consider the problem of predicting the direction of change in tick data of the Swiss franc/u.s. dollar exchange rate. To accomplish this, we show that a Markov model can nd regularities in certain local regions of the data and can be used to predict the direction of the next tick. Predictability seems to decrease in more recent years. With transaction costs, the model is unlikely to be protable. 1 Introduction The analysis and prediction of time series has received much interest in recent years. With respect to nancial time series, understanding the dynamics of these processes is a lucrative challenge; the ability to predict the direction and magnitude of change in these processes could lead to signicant prot. One of the basic premises of our work is that there is indeed some amount of regularity and predictability in nancial time series. The prediction problem that we tackle is formulated as one where we are trying to derive future price patterns in currency exchange trading from the past. Most models have focused on analysis at the level of daily, weekly, or monthly data; the model we present nds patterns at the resolution of ticks. We do not expect to nd a perfect predictor. Rather, we are interested in developing a system that can recognize certain regularities in the data and, when these patterns start emerging in the data, predict what follows at greater than 5% accuracy. Such a system, while possibly losing in the short term, should win in the long term. This approach uses a Markov model, a probabilistic process, to model the increases and decreases in the exchange rate at the tick level. The probabilities in the model are analyzed and used to develop simple prediction rules that are correct more then 5% of the time. The system is trained using a window of data extending N trading days prior to the current trading day. It is from these N days that we are trying to extract information about what might happen in the (N + 1) th day. We show that this simple model uncovers regularities that exist in the tick-by-tick exchange rate data. Using this model, we are able to make predictions in situations where the model has found a pattern; these predictions are correct more than 5% of the time in several dierent periods of data. 2 Foreign Exchange Market The foreign exchange market is a decentralized market where the money exchanged totals approximately one billion dollars each day. Banks and other market makers send their exchange rates to certain companies that disseminate this information to subscribers. Exchange rates actually have two components, the bid price and the ask price. The bid price is the highest price at which the market maker is willing to buy the currency, while the ask price is the lowest price at which the market maker is willing to sell the currency. To trade a currency, a dealer will call a market maker, request the current quote, and decide whether or not to place a trade. 1

2 Period 1 Period 2 Period /12/ /2/1985 4/25/1987 7/3/1989 x 1 5 Figure 1: The bid price time series for the Swiss franc/u.s. dollar exchange rate. The data that we use is the Swiss franc/u.s. dollar data for a single bank from May 2, 1985 to April 12, This series is at the resolution of ticks; a new tick (data point) occurs whenever the bank changes either its bid or ask price. The time between ticks ranges from seconds to minutes. This is an expanded version of the time series used in the 1991 Santa Fe Time Series Competition [7]. 3 Markov Models Markov models are stochastic models that capture variability in a process through time. Markov chains and, in particular, their generalizations, hidden Markov models, have been used heavily in signal processing elds like speech recognition with great success. This section presents a brief introduction to Markov chains; for an excellent introduction to hidden Markov models see Rabiner and Juang [6]. The Markov chain is an integer time process composed of a set of N states and N 2 transition probabilities. We denote the states by Q = fq i g N i=1 and let Q(t) denote the state of the system at time t. Let A(t) = fa ij (t)g N i;j=1 be the matrix of transition probabilities. More formally, a ij (t) = P (Q(t) = jjq(t? 1) = i) (1) is the probability that, given that the process is currently in state i at time t? 1, the process makes a transition to state j at time t. It is important to note that we are not assuming stationarity, so A(t) is time-dependent. The processes that we will be looking at will always have a well dened next state, so: NX a ij (t) = 1 (2) j=1 1 This data was obtained from the University of Colorado at Boulder Time Series Repository ( 2

3 C1 C2 C3 C4 C5 C6 C7 C8 C9 14 number of occurances log difference in bid price x 1 3 Figure 2: Histogram of the log dierences in prices for Period 1; dashed lines denote cluster boundaries. Such a model is characterized by the Markov independence assumption, that the state at time t is dependent only on the state at time t? 1. We can relax this assumption of local dependence to allow the state at time t to depend on the past n states. In this way, our transition probabilities become A(t) = fa in::i1j(t)g N i 1;:::;in;j=1 where, a in::i1j(t) = P (Q(t) = jjq(t? 1) = i1; Q(t? 2) = i2; :::; Q(t? n) = i n ) (3) With the above construction, if we know the current thru (n? 1) th states and the transition probabilities, this n th order Markov model is a complete probability distribution describing the next state. 4 System Architecture We divide the data into three dierent periods { Period 1 (5/2/1985-4/25/1987), Period 2 (4/26/1987-7/3/1989), and Period 3 (7/4/1989-4/12/1991) { shown in Figure (1). This paper analyzes only the bid price time series. Also, as is typically done with nancial time series, the data is preprocessed so that our data points, r t, are the log ratios between consecutive bid prices, p t?1 and p t. r t = ln( p t p t?1 ) (4) We analyze the histogram of the log ratio series for Period 1 and discretize the data into nine states. The nine states are shown in Figure (2) for Period 1. Points at zero signify no change from one tick to the next and constitute a signicant portion of the data set; these are included as a distinct state. To model the dynamics of the system we impose a second order assumption on the tick-by-tick price changes; this means that the state at time t is dependent on the states at time t? 1 and t? 2 and nothing else. Since we are using a second order Markov model, the Markov chain is indexed by the three variables i, j, and k, corresponding to times t? 2, t? 1, and t, respectively. While this model is rather simple, LeBaron[4] shows, for a one year portion of this data set, that there are 3

4 indeed signicant rst-order negative autocorrelations. Our system should be able to model these, and other more complex, patterns. Since we do not assume stationarity, we estimate the transition probabilities for the Markov chain by counting the occurrences of the dierent patterns in a xed window of data. The counts are kept track of in a matrix C. If the current training point is a price dierence from state i to state j to state k, we increment C ijk by one. This process is iterated over each data point in the day and the process is repeated for each day in the training window. We do not model changes between the closing price of one day and the opening price of the next. a ijk = C P ijk 9 x=1 C ijx (5) Hence, given that we know the past two price ratios, the next price ratio is a random variable whose value (actually, the state in which the value falls) is a ijx (t) of the transition matrix. The Markov model then generates trading rules by seeing if the probability values of the probabilities on either side of zero sum to a value greater than some threshold H greater than 5%; in our experiments, we varied this from 5% to 7%. We hope that by modeling the system using the smaller price changes, the predictions will be more reliable; the causes of large price movements have little chance of being modeled by such a simple system. For the purposes of this paper, we use the two states on either side of zero; these four states contain approximately 8% of the data points. If the probability values for a ij6(t) and a ij7(t) sum to some value greater than a threshold H, then we create the following rule to predict the upward direction of change of the next tick: if Q(t? 1) = i and Q(t) = j then Q(t + 1) will be an increase Similarly for decreases, if the sum of the probability values for a ij3 and a ij4 are greater than H, the rule will be: if Q(t? 1) = i and Q(t) = j then Q(t + 1) will be a decrease The state boundaries are dynamically adjusted for each window of training data and maintain the discretization of the price changes into nine states. 5 Experimental Results This process may be non-stationary so, to generate predictions for the current day, the Markov model's probabilities are estimated using a window of data that extends N days in the past. The experiments were run using values of N from 1 to 3. As stated above, the percentage threshold above which the system creates a rule is varied from 5% to 7%. As a benchmark, the Markov-based model is compared with one that uses a simple reversal strategy as in LeBaron[4]. For the next tick, t, this strategy gives a buy signal when r t?1 < and a sell signal when r t?1 >. For a closer comparison of these two strategies, the reversal strategy makes a prediction only when the Markov system makes a prediction. Graphs of a) the number of correct predictions as a percentage of the prediction made using the Markov model, b) the number of correct predictions as a percentage of the prediction made using the reversal strategy, and c) the number of predictions as a percentage of the total number of ticks, each plotted against the training window size and the training threshold size for the current day are plotted for each of the three periods in Figure (3). From the graphs, we see the empirical eects of varying the threshold and training window size. As we would expect, prediction accuracy increases with higher rule thresholds, but the number of predictions that the model makes decreases as the threshold increases. In Period 1, the Markov model seems to be generating robust rules for predicting the next tick; 98.4% (599/9) of the tested congurations are correct over 5% of the time, most are actually correct between and 7% of the time. Just as importantly, the patterns apply to up to 63% of the ticks that are seen. This is important for a high frequency trading system that is based on generating a small prot on each trade { in total, these prots will be substantial. Compared to 4

5 5 Markov model accuracy for Period 1 reversal strategy accuracy for Period 1 number of predictions as a percentage of total ticks for Period number of predictions out of total ticks (%) Markov model accuracy for Period 2 reversal strategy accuracy for Period 2 number of predictions as a percentage of total ticks for Period number of predictions out of total ticks (%) Markov model accuracy for Period 3 reversal strategy accuracy for Period 3 number of predictions as a percentage of total ticks for Period number of predictions out of total ticks (%) Figure 3: Correct predictions as a percentage of predictions made using the Markov model and the reversal strategy, and number of predictions as a percentage of total ticks for, by row, Periods 1, 2, and 3. the reversal strategy, the Markov model is much better at predicting the direction of change during this period; out of the 9 test runs, the system achieves better than 5% accuracy only two times. A possible reason for this could be that the reversal strategy is failing due to the sharp downward trend in this part of the series; the rst-order negative autocorrelations may not hold here. Things are less clear when we look at the Markov model's performance on Period 2. Here, the algorithm makes predictions on fewer ticks (at most 25%), but is still correct at least 5% of the time for 84.2% (513/9) of congurations and, for most congurations, between % and % of the time. Disregarding the higher threshold values where the number of predictions that are made is extremely small, the graph shows that the predictions are more accurate when the smaller sized windows of more recent training data is used to generate the model. This may indicate that the shorter term price patterns are more strongly correlated. The reversal strategy does not do as well as the Markov model; only 46.5% (383/9) of the tests yield performance of better than 5%. In Period 3, the model achieves 5% accuracy in 83.9% (511/9) of the tests with accuracy of between % and % for most of the congurations. The number of ticks that the system makes predictions on drops even more, to a maximum of 18%. The accuracy of the reversal strategy is nearly equivalent to the Markov model over all congurations; the Markov model seems to be uncovering only rst-order negative autocorrelations. To determine statistical signicance, we use the central limit theorem for Bernoulli random variables with = :5 as the level of signicance and measure the signicance of predictions over all tests that yielded accuracy of better than 5%. Signicance is measured separately for buy and 5

6 sell predictions. The tests take into account the actual distribution of increases and decreases for each prediction period; baseline guessing accuracy may dier from 5%. This test is related to the Henriksson-Merton test [3], but this connection has not been fully explored yet. Table (1) shows the results; overall the Markov-based system has a greater number of results where the prediction accuracy is above 5% and a higher percentage of these tests are statistically signicant. percent of tests where predictions are statistically signicant ( = :5) Increases Decreases Period 1 Markov 89.8% (491/547) 87.2% (524/1) reversal. (/2) NA Period 2 Markov 66.7 (316/474) 68.9 (346/52) reversal 33.5 (93/278) 39.3 (17/272) Period 3 Markov 44.2 (242/547) 83.1 (2/37) reversal 44.4 (243/547) 83.4 (256/37) Table 1: Number of statistically signicant tests of increase and decrease predictions for the Markov model and reversal strategy for each of the three periods, measured as a percentage of tests where the prediction accuracy is above 5%. 6 Discussion We present a simple Markov model trained to generate rules to predict the direction of the next tick in Swiss franc/u.s. dollar currency exchange trading. The system nds situations in which it can make predictions at reasonable accuracy and only makes predictions in these situations. For most congurations of the system, the prediction rules it generates are accurate over 5% of the time and often over -% of the time. The rules that the model creates apply to up to 63% of the ticks in Period 1, up to 25% of the ticks in Period 2, and up to 18% of the ticks in Period 3. Prediction accuracy decreases from Period 1 to Period 3. This may indicate that the short-term dynamics of this particular time series are becoming more complex over time, and thereby less predictable using this model. The fundamental aspect of this model that makes it particularly attractive is its simplicity. By discretizing the input space into a set of nine states, all the computations that the system does become trivial. This directly translates into the high speed at which the model is trained and predictions are made; training the model over a window size of 3 days and a threshold of 5% takes less than two seconds on a Sun SPARC 2. The shortcomings of this model are discussed here: A small simulation we performed shows that, when we factor in transaction costs, the model is not protable due to the fact that it trades on very small price dierences. We may want to predict changes in price over a longer time frame of hours or days to allow for larger dierences in prices. The determination of the state space is ad hoc. More experiments should be done to determine the optimal set of states. The model is too simple; assuming that the direction of the next tick can be predicted knowing only the past three tick values is overly restrictive. This system was run over the data from a single bank. This is only a small part of the picture as there are many other market makers and their interactions play a role in determining prices. Only the bid price series was used and this series does not reect prices at which actual trades were executed. 6

7 A direct comparison of this model to other techniques is not straightforward since most other systems predict actual price changes and do not attempt to make \next tick" predictions. We nevertheless present a brief overview of several related systems. Of particular relevance to this paper are those systems that attempt predictions at the level of ticks. Zhang [9] uses a multi-layer perceptron to forecast Swiss franc/u.s. dollar exchange rates based on several inputs from up to 18 minutes prior to the prediction time. These inputs are used to predict the exchange rate 3 minutes in the future. Weigend, Huberman, and Rumelhart [8] predict the Tuesday U.S. dollar/deutsch mark rate using a feedforward neural network over daily data. This system takes as input past measurements for this exchange rate, volatility and trend measures derived exclusively from the exchange rate series, and the current and past exchange rates for other currencies. Mozer [5] uses a recurrent neural network that takes tick data and predicts the exchange rate 1, 15, and minutes in the future. It is also important to note that several researchers have used a similar model to our Markov chain model, but at a much coarser resolution. In particular, Engel and Hamilton [2] and Engel [1] use a Markov switching model to model exchange rates for quarterly data. Engel concludes that the Markov switching model is more eective at predicting the direction of change, rather than the actual value, of exchange rates. This work demonstrates that Markov models may be extremely useful in uncovering and analyzing the hidden dynamics of nancial markets. This paper shows the eectiveness of using a nine state Markov model to create rules that predict the direction of the next tick in Swiss franc/u.s. dollar trading and lays the foundation for future work. References [1] C. Engel. Can the markov switching model forecast exchange rates? Journal of International Economics, 36:151{, [2] C. Engel and J.D. Hamilton. Long swings in the dollar: Are they in the data and do markets know it? The American Economic Review, 8(4):689{713, September 199. [3] R.D. Henriksson and R.C. Merton. On market timing and investment performance. ii. statistical procedures for evaluating forecasting skills. Journal of Business, 54(4):513{533, [4] B. LeBaron. Nonlinear diagnostics and simple trading rules for high-frequency foreign exchange rates. In A.S. Weigend and N.A. Gershenfeld, editors, Time Series Prediction: Forecasting the Future and Understanding the Past, pages 457{74, [5] M.C. Mozer. Neural net architectures for temporal sequence processing. In A.S. Weigend and N.A. Gershenfeld, editors, Time Series Prediction:Forecasting the Future and Understanding the Past, pages 243{64, [6] L.R. Rabiner and B.H. Juang. An introduction to hidden markov models. IEEE Acoustics, Speech, and Signal Processing Magazine, 3(1):4{16, January [7] A.S. Weigend and N.A. Gershenfeld, editors. Time Series Prediction: Forecasting the Future and Understanding the Past. SFI Studies in the Sciences of Complexity. Addison Wesley, [8] A.S. Weigend, B.A. Huberman, and D.E. Rumelhart. Predicting sunspots and exchange rates. In M. Casdagli and S. Eubank, editors, Nonlinear Modeling and Forecasting, pages 395{432, [9] Xiru Zhang. Non-linear predictive models for intra-day foreign exchange trading. Intelligent Systems in Accounting, Finance and Management, 3:293 { 32,

Optimization of technical trading strategies and the profitability in security markets

Optimization of technical trading strategies and the profitability in security markets Economics Letters 59 (1998) 249 254 Optimization of technical trading strategies and the profitability in security markets Ramazan Gençay 1, * University of Windsor, Department of Economics, 401 Sunset,

More information

Markov Chains for the RISK Board Game Revisited. Introduction. The Markov Chain. Jason A. Osborne North Carolina State University Raleigh, NC 27695

Markov Chains for the RISK Board Game Revisited. Introduction. The Markov Chain. Jason A. Osborne North Carolina State University Raleigh, NC 27695 Markov Chains for the RISK Board Game Revisited Jason A. Osborne North Carolina State University Raleigh, NC 27695 Introduction Probabilistic reasoning goes a long way in many popular board games. Abbott

More information

NEURAL networks [5] are universal approximators [6]. It

NEURAL networks [5] are universal approximators [6]. It Proceedings of the 2013 Federated Conference on Computer Science and Information Systems pp. 183 190 An Investment Strategy for the Stock Exchange Using Neural Networks Antoni Wysocki and Maciej Ławryńczuk

More information

Using News Articles to Predict Stock Price Movements

Using News Articles to Predict Stock Price Movements Using News Articles to Predict Stock Price Movements Győző Gidófalvi Department of Computer Science and Engineering University of California, San Diego La Jolla, CA 9237 gyozo@cs.ucsd.edu 21, June 15,

More information

A Simple Model for Intra-day Trading

A Simple Model for Intra-day Trading A Simple Model for Intra-day Trading Anton Golub 1 1 Marie Curie Fellow, Manchester Business School April 15, 2011 Abstract Since currency market is an OTC market, there is no information about orders,

More information

Forecasting Trade Direction and Size of Future Contracts Using Deep Belief Network

Forecasting Trade Direction and Size of Future Contracts Using Deep Belief Network Forecasting Trade Direction and Size of Future Contracts Using Deep Belief Network Anthony Lai (aslai), MK Li (lilemon), Foon Wang Pong (ppong) Abstract Algorithmic trading, high frequency trading (HFT)

More information

Tracking Moving Objects In Video Sequences Yiwei Wang, Robert E. Van Dyck, and John F. Doherty Department of Electrical Engineering The Pennsylvania State University University Park, PA16802 Abstract{Object

More information

Bilateral Exposures and Systemic Solvency Risk

Bilateral Exposures and Systemic Solvency Risk Bilateral Exposures and Systemic Solvency Risk C., GOURIEROUX (1), J.C., HEAM (2), and A., MONFORT (3) (1) CREST, and University of Toronto (2) CREST, and Autorité de Contrôle Prudentiel et de Résolution

More information

COMBINED NEURAL NETWORKS FOR TIME SERIES ANALYSIS

COMBINED NEURAL NETWORKS FOR TIME SERIES ANALYSIS COMBINED NEURAL NETWORKS FOR TIME SERIES ANALYSIS Iris Ginzburg and David Horn School of Physics and Astronomy Raymond and Beverly Sackler Faculty of Exact Science Tel-Aviv University Tel-A viv 96678,

More information

Channel Allocation in Cellular Telephone. Systems. Lab. for Info. and Decision Sciences. Cambridge, MA 02139. bertsekas@lids.mit.edu.

Channel Allocation in Cellular Telephone. Systems. Lab. for Info. and Decision Sciences. Cambridge, MA 02139. bertsekas@lids.mit.edu. Reinforcement Learning for Dynamic Channel Allocation in Cellular Telephone Systems Satinder Singh Department of Computer Science University of Colorado Boulder, CO 80309-0430 baveja@cs.colorado.edu Dimitri

More information

Algorithmic Trading Session 1 Introduction. Oliver Steinki, CFA, FRM

Algorithmic Trading Session 1 Introduction. Oliver Steinki, CFA, FRM Algorithmic Trading Session 1 Introduction Oliver Steinki, CFA, FRM Outline An Introduction to Algorithmic Trading Definition, Research Areas, Relevance and Applications General Trading Overview Goals

More information

A Neural Network and Web-Based Decision Support System for Forex Forecasting and Trading

A Neural Network and Web-Based Decision Support System for Forex Forecasting and Trading A Neural Network and Web-Based Decision Support System for Forex Forecasting and Trading K.K. Lai 1, Lean Yu 2,3, and Shouyang Wang 2,4 1 Department of Management Sciences, City University of Hong Kong,

More information

On closed-form solutions of a resource allocation problem in parallel funding of R&D projects

On closed-form solutions of a resource allocation problem in parallel funding of R&D projects Operations Research Letters 27 (2000) 229 234 www.elsevier.com/locate/dsw On closed-form solutions of a resource allocation problem in parallel funding of R&D proects Ulku Gurler, Mustafa. C. Pnar, Mohamed

More information

An Environment Model for N onstationary Reinforcement Learning

An Environment Model for N onstationary Reinforcement Learning An Environment Model for N onstationary Reinforcement Learning Samuel P. M. Choi Dit-Yan Yeung Nevin L. Zhang pmchoi~cs.ust.hk dyyeung~cs.ust.hk lzhang~cs.ust.hk Department of Computer Science, Hong Kong

More information

IMPLICIT COLLUSION IN DEALER MARKETS WITH DIFFERENT COSTS OF MARKET MAKING ANDREAS KRAUSE Abstract. This paper introduces dierent costs into the Dutta-Madhavan model of implicit collusion between market

More information

Introduction to Machine Learning and Data Mining. Prof. Dr. Igor Trajkovski trajkovski@nyus.edu.mk

Introduction to Machine Learning and Data Mining. Prof. Dr. Igor Trajkovski trajkovski@nyus.edu.mk Introduction to Machine Learning and Data Mining Prof. Dr. Igor Trakovski trakovski@nyus.edu.mk Neural Networks 2 Neural Networks Analogy to biological neural systems, the most robust learning systems

More information

Ulrich A. Muller UAM.1994-01-31. June 28, 1995

Ulrich A. Muller UAM.1994-01-31. June 28, 1995 Hedging Currency Risks { Dynamic Hedging Strategies Based on O & A Trading Models Ulrich A. Muller UAM.1994-01-31 June 28, 1995 A consulting document by the O&A Research Group This document is the property

More information

Alerts & Filters in Power E*TRADE Pro Strategy Scanner

Alerts & Filters in Power E*TRADE Pro Strategy Scanner Alerts & Filters in Power E*TRADE Pro Strategy Scanner Power E*TRADE Pro Strategy Scanner provides real-time technical screening and backtesting based on predefined and custom strategies. With custom strategies,

More information

Multiple Kernel Learning on the Limit Order Book

Multiple Kernel Learning on the Limit Order Book JMLR: Workshop and Conference Proceedings 11 (2010) 167 174 Workshop on Applications of Pattern Analysis Multiple Kernel Learning on the Limit Order Book Tristan Fletcher Zakria Hussain John Shawe-Taylor

More information

Multiagent Reputation Management to Achieve Robust Software Using Redundancy

Multiagent Reputation Management to Achieve Robust Software Using Redundancy Multiagent Reputation Management to Achieve Robust Software Using Redundancy Rajesh Turlapati and Michael N. Huhns Center for Information Technology, University of South Carolina Columbia, SC 29208 {turlapat,huhns}@engr.sc.edu

More information

LECTURE 4. Last time: Lecture outline

LECTURE 4. Last time: Lecture outline LECTURE 4 Last time: Types of convergence Weak Law of Large Numbers Strong Law of Large Numbers Asymptotic Equipartition Property Lecture outline Stochastic processes Markov chains Entropy rate Random

More information

Linear Programming. March 14, 2014

Linear Programming. March 14, 2014 Linear Programming March 1, 01 Parts of this introduction to linear programming were adapted from Chapter 9 of Introduction to Algorithms, Second Edition, by Cormen, Leiserson, Rivest and Stein [1]. 1

More information

Prediction of Closing Stock Prices

Prediction of Closing Stock Prices 1 Prediction of Closing Stock Prices Garth Garner Portland State University department of Electrical and Computer Engineering Email: garth.garner@biotronik.com This work was completed as part of a course

More information

Face Locating and Tracking for Human{Computer Interaction. Carnegie Mellon University. Pittsburgh, PA 15213

Face Locating and Tracking for Human{Computer Interaction. Carnegie Mellon University. Pittsburgh, PA 15213 Face Locating and Tracking for Human{Computer Interaction Martin Hunke Alex Waibel School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 Abstract Eective Human-to-Human communication

More information

Review of Fundamental Mathematics

Review of Fundamental Mathematics Review of Fundamental Mathematics As explained in the Preface and in Chapter 1 of your textbook, managerial economics applies microeconomic theory to business decision making. The decision-making tools

More information

1. (First passage/hitting times/gambler s ruin problem:) Suppose that X has a discrete state space and let i be a fixed state. Let

1. (First passage/hitting times/gambler s ruin problem:) Suppose that X has a discrete state space and let i be a fixed state. Let Copyright c 2009 by Karl Sigman 1 Stopping Times 1.1 Stopping Times: Definition Given a stochastic process X = {X n : n 0}, a random time τ is a discrete random variable on the same probability space as

More information

atching Currency Moves with The Schaff Trend Cycle Indicator

atching Currency Moves with The Schaff Trend Cycle Indicator JULY/AUGUST 2002. VOLUME 4 CRunning a trend indicator through a cycle oscillator creates an effective entry technique into today s strongly trending currency markets, says Doug Schaff, a 20-year veteran

More information

Stock Market Forecasting Using Machine Learning Algorithms

Stock Market Forecasting Using Machine Learning Algorithms Stock Market Forecasting Using Machine Learning Algorithms Shunrong Shen, Haomiao Jiang Department of Electrical Engineering Stanford University {conank,hjiang36}@stanford.edu Tongda Zhang Department of

More information

A Simple Feature Extraction Technique of a Pattern By Hopfield Network

A Simple Feature Extraction Technique of a Pattern By Hopfield Network A Simple Feature Extraction Technique of a Pattern By Hopfield Network A.Nag!, S. Biswas *, D. Sarkar *, P.P. Sarkar *, B. Gupta **! Academy of Technology, Hoogly - 722 *USIC, University of Kalyani, Kalyani

More information

6.2.8 Neural networks for data mining

6.2.8 Neural networks for data mining 6.2.8 Neural networks for data mining Walter Kosters 1 In many application areas neural networks are known to be valuable tools. This also holds for data mining. In this chapter we discuss the use of neural

More information

Ericsson T18s Voice Dialing Simulator

Ericsson T18s Voice Dialing Simulator Ericsson T18s Voice Dialing Simulator Mauricio Aracena Kovacevic, Anna Dehlbom, Jakob Ekeberg, Guillaume Gariazzo, Eric Lästh and Vanessa Troncoso Dept. of Signals Sensors and Systems Royal Institute of

More information

Application of Markov chain analysis to trend prediction of stock indices Milan Svoboda 1, Ladislav Lukáš 2

Application of Markov chain analysis to trend prediction of stock indices Milan Svoboda 1, Ladislav Lukáš 2 Proceedings of 3th International Conference Mathematical Methods in Economics 1 Introduction Application of Markov chain analysis to trend prediction of stock indices Milan Svoboda 1, Ladislav Lukáš 2

More information

Investment Statistics: Definitions & Formulas

Investment Statistics: Definitions & Formulas Investment Statistics: Definitions & Formulas The following are brief descriptions and formulas for the various statistics and calculations available within the ease Analytics system. Unless stated otherwise,

More information

by Maria Heiden, Berenberg Bank

by Maria Heiden, Berenberg Bank Dynamic hedging of equity price risk with an equity protect overlay: reduce losses and exploit opportunities by Maria Heiden, Berenberg Bank As part of the distortions on the international stock markets

More information

Design call center management system of e-commerce based on BP neural network and multifractal

Design call center management system of e-commerce based on BP neural network and multifractal Available online www.jocpr.com Journal of Chemical and Pharmaceutical Research, 2014, 6(6):951-956 Research Article ISSN : 0975-7384 CODEN(USA) : JCPRC5 Design call center management system of e-commerce

More information

Technical Trading Rules as a Prior Knowledge to a Neural Networks Prediction System for the S&P 500 Index

Technical Trading Rules as a Prior Knowledge to a Neural Networks Prediction System for the S&P 500 Index Technical Trading Rules as a Prior Knowledge to a Neural Networks Prediction System for the S&P 5 ndex Tim Cheno~ethl-~t~ Zoran ObradoviC Steven Lee4 School of Electrical Engineering and Computer Science

More information

4. Simple regression. QBUS6840 Predictive Analytics. https://www.otexts.org/fpp/4

4. Simple regression. QBUS6840 Predictive Analytics. https://www.otexts.org/fpp/4 4. Simple regression QBUS6840 Predictive Analytics https://www.otexts.org/fpp/4 Outline The simple linear model Least squares estimation Forecasting with regression Non-linear functional forms Regression

More information

A Robustness Simulation Method of Project Schedule based on the Monte Carlo Method

A Robustness Simulation Method of Project Schedule based on the Monte Carlo Method Send Orders for Reprints to reprints@benthamscience.ae 254 The Open Cybernetics & Systemics Journal, 2014, 8, 254-258 Open Access A Robustness Simulation Method of Project Schedule based on the Monte Carlo

More information

Is the Forward Exchange Rate a Useful Indicator of the Future Exchange Rate?

Is the Forward Exchange Rate a Useful Indicator of the Future Exchange Rate? Is the Forward Exchange Rate a Useful Indicator of the Future Exchange Rate? Emily Polito, Trinity College In the past two decades, there have been many empirical studies both in support of and opposing

More information

Financial Risk Forecasting Chapter 8 Backtesting and stresstesting

Financial Risk Forecasting Chapter 8 Backtesting and stresstesting Financial Risk Forecasting Chapter 8 Backtesting and stresstesting Jon Danielsson London School of Economics 2015 To accompany Financial Risk Forecasting http://www.financialriskforecasting.com/ Published

More information

Error Log Processing for Accurate Failure Prediction. Humboldt-Universität zu Berlin

Error Log Processing for Accurate Failure Prediction. Humboldt-Universität zu Berlin Error Log Processing for Accurate Failure Prediction Felix Salfner ICSI Berkeley Steffen Tschirpke Humboldt-Universität zu Berlin Introduction Context of work: Error-based online failure prediction: error

More information

Caput Derivatives: October 30, 2003

Caput Derivatives: October 30, 2003 Caput Derivatives: October 30, 2003 Exam + Answers Total time: 2 hours and 30 minutes. Note 1: You are allowed to use books, course notes, and a calculator. Question 1. [20 points] Consider an investor

More information

4. Continuous Random Variables, the Pareto and Normal Distributions

4. Continuous Random Variables, the Pareto and Normal Distributions 4. Continuous Random Variables, the Pareto and Normal Distributions A continuous random variable X can take any value in a given range (e.g. height, weight, age). The distribution of a continuous random

More information

Linear Programming. April 12, 2005

Linear Programming. April 12, 2005 Linear Programming April 1, 005 Parts of this were adapted from Chapter 9 of i Introduction to Algorithms (Second Edition) /i by Cormen, Leiserson, Rivest and Stein. 1 What is linear programming? The first

More information

Algorithmic Trading: A Quantitative Approach to Developing an Automated Trading System

Algorithmic Trading: A Quantitative Approach to Developing an Automated Trading System Algorithmic Trading: A Quantitative Approach to Developing an Automated Trading System Luqman-nul Hakim B M Lukman, Justin Yeo Shui Ming NUS Investment Society Research (Quantitative Finance) Abstract:

More information

Particle Filter-Based On-Line Estimation of Spot (Cross-)Volatility with Nonlinear Market Microstructure Noise Models. Jan C.

Particle Filter-Based On-Line Estimation of Spot (Cross-)Volatility with Nonlinear Market Microstructure Noise Models. Jan C. Particle Filter-Based On-Line Estimation of Spot (Cross-)Volatility with Nonlinear Market Microstructure Noise Models Jan C. Neddermeyer (joint work with Rainer Dahlhaus, University of Heidelberg) DZ BANK

More information

Empirical Analysis of an Online Algorithm for Multiple Trading Problems

Empirical Analysis of an Online Algorithm for Multiple Trading Problems Empirical Analysis of an Online Algorithm for Multiple Trading Problems Esther Mohr 1) Günter Schmidt 1+2) 1) Saarland University 2) University of Liechtenstein em@itm.uni-sb.de gs@itm.uni-sb.de Abstract:

More information

CAPM, Arbitrage, and Linear Factor Models

CAPM, Arbitrage, and Linear Factor Models CAPM, Arbitrage, and Linear Factor Models CAPM, Arbitrage, Linear Factor Models 1/ 41 Introduction We now assume all investors actually choose mean-variance e cient portfolios. By equating these investors

More information

CHAPTER NINE WORLD S SIMPLEST 5 PIP SCALPING METHOD

CHAPTER NINE WORLD S SIMPLEST 5 PIP SCALPING METHOD CHAPTER NINE WORLD S SIMPLEST 5 PIP SCALPING METHOD For a book that was intended to help you build your confidence, it s spent a long time instead discussing various trading methods and systems. But there

More information

Chapter 2.4. Trading using multiple time-frames

Chapter 2.4. Trading using multiple time-frames 1 Chapter 2.4 Trading using multiple time-frames 0 Contents TRADING USING MULTIPLE TIME- FRAMES Stock markets worldwide function because, at any given time, some traders want to buy whilst others want

More information

An analysis of price impact function in order-driven markets

An analysis of price impact function in order-driven markets Available online at www.sciencedirect.com Physica A 324 (2003) 146 151 www.elsevier.com/locate/physa An analysis of price impact function in order-driven markets G. Iori a;, M.G. Daniels b, J.D. Farmer

More information

Prediction Model for Crude Oil Price Using Artificial Neural Networks

Prediction Model for Crude Oil Price Using Artificial Neural Networks Applied Mathematical Sciences, Vol. 8, 2014, no. 80, 3953-3965 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/10.12988/ams.2014.43193 Prediction Model for Crude Oil Price Using Artificial Neural Networks

More information

Contents. Abstract. Acknowledgments. 1 Introduction 1 1.1 Project Description... 2

Contents. Abstract. Acknowledgments. 1 Introduction 1 1.1 Project Description... 2 Abstract The purpose of this IQP is to use the basics of trading and investment and apply it scientically to create a protable system made up of several systems. The project consists of the creation of

More information

1. Classification problems

1. Classification problems Neural and Evolutionary Computing. Lab 1: Classification problems Machine Learning test data repository Weka data mining platform Introduction Scilab 1. Classification problems The main aim of a classification

More information

Overlapping ETF: Pair trading between two gold stocks

Overlapping ETF: Pair trading between two gold stocks MPRA Munich Personal RePEc Archive Overlapping ETF: Pair trading between two gold stocks Peter N Bell and Brian Lui and Alex Brekke University of Victoria 1. April 2012 Online at http://mpra.ub.uni-muenchen.de/39534/

More information

SELECTING NEURAL NETWORK ARCHITECTURE FOR INVESTMENT PROFITABILITY PREDICTIONS

SELECTING NEURAL NETWORK ARCHITECTURE FOR INVESTMENT PROFITABILITY PREDICTIONS UDC: 004.8 Original scientific paper SELECTING NEURAL NETWORK ARCHITECTURE FOR INVESTMENT PROFITABILITY PREDICTIONS Tonimir Kišasondi, Alen Lovren i University of Zagreb, Faculty of Organization and Informatics,

More information

Texas Christian University. Department of Economics. Working Paper Series. Modeling Interest Rate Parity: A System Dynamics Approach

Texas Christian University. Department of Economics. Working Paper Series. Modeling Interest Rate Parity: A System Dynamics Approach Texas Christian University Department of Economics Working Paper Series Modeling Interest Rate Parity: A System Dynamics Approach John T. Harvey Department of Economics Texas Christian University Working

More information

A Spectral Clustering Approach to Validating Sensors via Their Peers in Distributed Sensor Networks

A Spectral Clustering Approach to Validating Sensors via Their Peers in Distributed Sensor Networks A Spectral Clustering Approach to Validating Sensors via Their Peers in Distributed Sensor Networks H. T. Kung Dario Vlah {htk, dario}@eecs.harvard.edu Harvard School of Engineering and Applied Sciences

More information

Machine Learning in FX Carry Basket Prediction

Machine Learning in FX Carry Basket Prediction Machine Learning in FX Carry Basket Prediction Tristan Fletcher, Fabian Redpath and Joe D Alessandro Abstract Artificial Neural Networks ANN), Support Vector Machines SVM) and Relevance Vector Machines

More information

More informed trading Technical Analysis: Trading Using Multiple Time-frames

More informed trading Technical Analysis: Trading Using Multiple Time-frames Technical Analysis: Trading Using Multiple Time-frames Intermediate Level Introduction 1 Stock markets worldwide function because, at any given time, some traders want to buy whilst others want to sell.

More information

Understanding Currency

Understanding Currency Understanding Currency Overlay July 2010 PREPARED BY Gregory J. Leonberger, FSA Director of Research Abstract As portfolios have expanded to include international investments, investors must be aware of

More information

Modeling a Foreign Exchange Rate

Modeling a Foreign Exchange Rate Modeling a foreign exchange rate using moving average of Yen-Dollar market data Takayuki Mizuno 1, Misako Takayasu 1, Hideki Takayasu 2 1 Department of Computational Intelligence and Systems Science, Interdisciplinary

More information

Forecasting of Indian Rupee (INR) / US Dollar (USD) Currency Exchange Rate Using Artificial Neural Network

Forecasting of Indian Rupee (INR) / US Dollar (USD) Currency Exchange Rate Using Artificial Neural Network Forecasting of Indian Rupee (INR) / US Dollar (USD) Currency Exchange Rate Using Artificial Neural Network Yusuf Perwej 1 and Asif Perwej 2 1 M.Tech, MCA, Department of Computer Science & Information System,

More information

INCORPORATION OF LIQUIDITY RISKS INTO EQUITY PORTFOLIO RISK ESTIMATES. Dan dibartolomeo September 2010

INCORPORATION OF LIQUIDITY RISKS INTO EQUITY PORTFOLIO RISK ESTIMATES. Dan dibartolomeo September 2010 INCORPORATION OF LIQUIDITY RISKS INTO EQUITY PORTFOLIO RISK ESTIMATES Dan dibartolomeo September 2010 GOALS FOR THIS TALK Assert that liquidity of a stock is properly measured as the expected price change,

More information

It is remarkable that a science, which began with the consideration of games of chance, should be elevated to the rank of the most important

It is remarkable that a science, which began with the consideration of games of chance, should be elevated to the rank of the most important PROBABILLITY 271 PROBABILITY CHAPTER 15 It is remarkable that a science, which began with the consideration of games of chance, should be elevated to the rank of the most important subject of human knowledge.

More information

Predict the Popularity of YouTube Videos Using Early View Data

Predict the Popularity of YouTube Videos Using Early View Data 000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050

More information

Introduction to Algorithmic Trading Strategies Lecture 2

Introduction to Algorithmic Trading Strategies Lecture 2 Introduction to Algorithmic Trading Strategies Lecture 2 Hidden Markov Trading Model Haksun Li haksun.li@numericalmethod.com www.numericalmethod.com Outline Carry trade Momentum Valuation CAPM Markov chain

More information

Chapter 10. Key Ideas Correlation, Correlation Coefficient (r),

Chapter 10. Key Ideas Correlation, Correlation Coefficient (r), Chapter 0 Key Ideas Correlation, Correlation Coefficient (r), Section 0-: Overview We have already explored the basics of describing single variable data sets. However, when two quantitative variables

More information

TRADING SECRET NO. 3 How to combine two simple tools to capture big trending moves, especially in currencies

TRADING SECRET NO. 3 How to combine two simple tools to capture big trending moves, especially in currencies TRADING SECRET NO. 3 How to combine two simple tools to capture big trending moves, especially in currencies 25 Currencies have a reputation for being markets that trend well. And, every business day over

More information

THREE DIMENSIONAL REPRESENTATION OF AMINO ACID CHARAC- TERISTICS

THREE DIMENSIONAL REPRESENTATION OF AMINO ACID CHARAC- TERISTICS THREE DIMENSIONAL REPRESENTATION OF AMINO ACID CHARAC- TERISTICS O.U. Sezerman 1, R. Islamaj 2, E. Alpaydin 2 1 Laborotory of Computational Biology, Sabancı University, Istanbul, Turkey. 2 Computer Engineering

More information

STock prices on electronic exchanges are determined at each tick by a matching algorithm which matches buyers with

STock prices on electronic exchanges are determined at each tick by a matching algorithm which matches buyers with JOURNAL OF STELLAR MS&E 444 REPORTS 1 Adaptive Strategies for High Frequency Trading Erik Anderson and Paul Merolla and Alexis Pribula STock prices on electronic exchanges are determined at each tick by

More information

Chapter 3 RANDOM VARIATE GENERATION

Chapter 3 RANDOM VARIATE GENERATION Chapter 3 RANDOM VARIATE GENERATION In order to do a Monte Carlo simulation either by hand or by computer, techniques must be developed for generating values of random variables having known distributions.

More information

Artificial Neural Network for Speech Recognition

Artificial Neural Network for Speech Recognition Artificial Neural Network for Speech Recognition Austin Marshall March 3, 2005 2nd Annual Student Research Showcase Overview Presenting an Artificial Neural Network to recognize and classify speech Spoken

More information

Foreign exchange trading models and market behavior

Foreign exchange trading models and market behavior Available online at www.sciencedirect.com Journal of Economic Dynamics & Control 27 (2003) 909 935 www.elsevier.com/locate/econbase Foreign exchange trading models and market behavior Ramazan Gencay a;,

More information

A framework for parallel data mining using neural networks R. Owen Rogers rogers@qucis.queensu.ca November 1997 External Technical Report ISSN-0836-0227- 97-413 Department of Computing and Information

More information

3 An Illustrative Example

3 An Illustrative Example Objectives An Illustrative Example Objectives - Theory and Examples -2 Problem Statement -2 Perceptron - Two-Input Case -4 Pattern Recognition Example -5 Hamming Network -8 Feedforward Layer -8 Recurrent

More information

MANAGEMENT OPTIONS AND VALUE PER SHARE

MANAGEMENT OPTIONS AND VALUE PER SHARE 1 MANAGEMENT OPTIONS AND VALUE PER SHARE Once you have valued the equity in a firm, it may appear to be a relatively simple exercise to estimate the value per share. All it seems you need to do is divide

More information

CONTENT 1. 2. 5-8 9 5. 6. 7.

CONTENT 1. 2. 5-8 9 5. 6. 7. User Manual TM CONTENT 1. 2. 3. 4. 5. 6. 7. 8. 9. Introduction The Autochartist Interface Analysis Toolbar (A) Pattern Display (B) Search Pane (C) Results Pane (Completed Patterns) (D) Results Pane (Emerging

More information

The Moving Average. 2004 W. R. Booker II. All rights reserved forever and ever. And ever.

The Moving Average. 2004 W. R. Booker II. All rights reserved forever and ever. And ever. The Moving Average By Rob Booker 2004 W. R. Booker II. All rights reserved forever and ever. And ever. The information contained in this ebook is designed to teach you methods of watching forex quotes

More information

Simple Investment Strategy During a Global Crisis

Simple Investment Strategy During a Global Crisis 2012 Simple Investment Strategy During a Global Crisis This document describes a simple investment strategy designed for people who do not have much experience in investments, but would like to optimize

More information

Hedging a Foreign Equity Position

Hedging a Foreign Equity Position Hedging a Foreign Equity Position Hedging a US Equity Position in a Portfolio Based in Canadian Dollars : An Example Comparing Strategies SUMMARY We compare three different strategies for hedging a long

More information

Section A. Index. Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1. Page 1 of 11. EduPristine CMA - Part I

Section A. Index. Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1. Page 1 of 11. EduPristine CMA - Part I Index Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1 EduPristine CMA - Part I Page 1 of 11 Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting

More information

Lecture 6. Artificial Neural Networks

Lecture 6. Artificial Neural Networks Lecture 6 Artificial Neural Networks 1 1 Artificial Neural Networks In this note we provide an overview of the key concepts that have led to the emergence of Artificial Neural Networks as a major paradigm

More information

OPENING RANGE BREAKOUT (ORB- Advanced)

OPENING RANGE BREAKOUT (ORB- Advanced) OPENING RANGE BREAKOUT (ORB- Advanced) Weekly volatility breakout strategy. On Monday morning check the London opening price, 08.00 GMT. The difference between this weekly ORB Basic and this ORB Advanced

More information

OPTIMAL DESIGN OF A MULTITIER REWARD SCHEME. Amir Gandomi *, Saeed Zolfaghari **

OPTIMAL DESIGN OF A MULTITIER REWARD SCHEME. Amir Gandomi *, Saeed Zolfaghari ** OPTIMAL DESIGN OF A MULTITIER REWARD SCHEME Amir Gandomi *, Saeed Zolfaghari ** Department of Mechanical and Industrial Engineering, Ryerson University, Toronto, Ontario * Tel.: + 46 979 5000x7702, Email:

More information

MUSICAL INSTRUMENT FAMILY CLASSIFICATION

MUSICAL INSTRUMENT FAMILY CLASSIFICATION MUSICAL INSTRUMENT FAMILY CLASSIFICATION Ricardo A. Garcia Media Lab, Massachusetts Institute of Technology 0 Ames Street Room E5-40, Cambridge, MA 039 USA PH: 67-53-0 FAX: 67-58-664 e-mail: rago @ media.

More information

Warrants, Certificates and other products

Warrants, Certificates and other products Interconnection Trading System Warrants, Certificates and other products MARKET MODEL DESCRIPTION January 2015 TABLE OF CONTENTS 1. INTRODUCTION 4 1.1. Background 4 1.2. Institutional market configuration

More information

Chapter 1 Introduction to Correlation

Chapter 1 Introduction to Correlation Chapter 1 Introduction to Correlation Suppose that you woke up one morning and discovered that you had been given the gift of being able to predict the future. Suddenly, you found yourself able to predict,

More information

PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION

PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION Introduction In the previous chapter, we explored a class of regression models having particularly simple analytical

More information

Prediction of Stock Performance Using Analytical Techniques

Prediction of Stock Performance Using Analytical Techniques 136 JOURNAL OF EMERGING TECHNOLOGIES IN WEB INTELLIGENCE, VOL. 5, NO. 2, MAY 2013 Prediction of Stock Performance Using Analytical Techniques Carol Hargreaves Institute of Systems Science National University

More information

High Frequency Trading + Stochastic Latency and Regulation 2.0. Andrei Kirilenko MIT Sloan

High Frequency Trading + Stochastic Latency and Regulation 2.0. Andrei Kirilenko MIT Sloan High Frequency Trading + Stochastic Latency and Regulation 2.0 Andrei Kirilenko MIT Sloan High Frequency Trading: Good or Evil? Good Bryan Durkin, Chief Operating Officer, CME Group: "There is considerable

More information

Financial Mathematics and Simulation MATH 6740 1 Spring 2011 Homework 2

Financial Mathematics and Simulation MATH 6740 1 Spring 2011 Homework 2 Financial Mathematics and Simulation MATH 6740 1 Spring 2011 Homework 2 Due Date: Friday, March 11 at 5:00 PM This homework has 170 points plus 20 bonus points available but, as always, homeworks are graded

More information

MTH6120 Further Topics in Mathematical Finance Lesson 2

MTH6120 Further Topics in Mathematical Finance Lesson 2 MTH6120 Further Topics in Mathematical Finance Lesson 2 Contents 1.2.3 Non-constant interest rates....................... 15 1.3 Arbitrage and Black-Scholes Theory....................... 16 1.3.1 Informal

More information

Simple model of a limit order-driven market

Simple model of a limit order-driven market Physica A 278 (2000) 571 578 www.elsevier.com/locate/physa Simple model of a limit order-driven market Sergei Maslov Department of Physics, Brookhaven National Laboratory, Upton, NY 11973, USA Received

More information

Online Investments. Our Fund Range and Investments

Online Investments. Our Fund Range and Investments Online Investments Our Fund Range and Investments Why is it important to read this document? This document explains the funds available for you to invest in through our Investment ISA, which is a Stocks

More information

Analyzing Portfolio Expected Loss

Analyzing Portfolio Expected Loss Analyzing Portfolio Expected Loss In this white paper we discuss the methodologies that Visible Equity employs in the calculation of portfolio expected loss. Portfolio expected loss calculations combine

More information

INFERRING TRADING STRATEGIES FROM PROBABILITY DISTRIBUTION FUNCTIONS

INFERRING TRADING STRATEGIES FROM PROBABILITY DISTRIBUTION FUNCTIONS INFERRING TRADING STRATEGIES FROM PROBABILITY DISTRIBUTION FUNCTIONS INFERRING TRADING STRATEGIES FROM PROBABILITY DISTRIBUTION FUNCTIONS BACKGROUND The primary purpose of technical analysis is to observe

More information

Forward exchange rates

Forward exchange rates Forward exchange rates The forex market consists of two distinct markets - the spot foreign exchange market (in which currencies are bought and sold for delivery within two working days) and the forward

More information

Time series analysis as a framework for the characterization of waterborne disease outbreaks

Time series analysis as a framework for the characterization of waterborne disease outbreaks Interdisciplinary Perspectives on Drinking Water Risk Assessment and Management (Proceedings of the Santiago (Chile) Symposium, September 1998). IAHS Publ. no. 260, 2000. 127 Time series analysis as a

More information

How To Beat The Currency Market Without (Much) Skill

How To Beat The Currency Market Without (Much) Skill How To Beat The Currency Market Without (Much) Skill January 2010 FQ Perspective by Dori Levanoni and Juliana Bambaci We ve written about rebalancing strategies for nearly two decades. 1 We ve done so

More information