bitarisk. BITA Vision a product from corfinancial. london boston new york BETTER INTELLIGENCE THROUGH ANALYSIS better intelligence through analysis

Transcription

1 bitarisk. BETTER INTELLIGENCE THROUGH ANALYSIS better intelligence through analysis BITA Vision a product from corfinancial. london boston new york

2 Expertise and experience deliver efficiency and value BITA Risk is a global provider of portfolio construction, analysis, risk and reporting tools to investment managers, hedge and quantitative funds in offices around the world; meeting the needs of a wide variety of investment approaches. For more than 20 years, BITA Risk has combined deep industry and academic expertise in portfolio risk analysis, construction and optimisation to bring added value and insight to its institutional and quant clients businesses. Simple delivery of complex analysis making numbers work for you BITA Vision works for both quantitative and fundamental investment approaches: it lets you design, test, implement and report on a quantitative strategy, or test, analyse and question your fundamental assumptions. reports Full range of portfolio optimisation using BITA s mean-variance, robust and gain-loss optimisers N-tile portfolio building, factor and alpha signal research Risk model building Stress testing Support for a blended risk model between stressed and normal economic regimes Implied return and reverse optimisation analysis Efficient frontier, surface construction, analysis in multiple dimensions Very easy and comprehensive backtesting routines over a user defined date range, based on N-tile and filtering and or optimisation and simulation strategies Report pack generation Full access to functionality through a graphical user interface Full programming control with and without Matlab for additional functionality Joining the dots for a rapid and flexible portfolio build and analysis processes. In both cases BITA Vision provides significant depth of functionality and flexibility working both with ex-ante and ex-post data, through its powerful user interface. And for those who want it, direct access to MATLAB scripting is possible, enabling even greater flexibility. BITA Vision is a fast application, packed with functionality, available as modules, that deliver: Analysis and reporting based upon user defined date range and asset universe Ex-ante portfolio analysis reports Ex-post portfolio analysis reports Risk attribution on an absolute and relative basis Performance return attribution on an absolute and relative basis Portfolio simulation including boot-strapping and Monte Carlo analysis as well as associated analysis 1/6.

3

4 1 Alpha Research provides the tools needed to build and test your own return forecast/ signal models, e.g. optimally combine style factors or build a model based on broker estimates. The model can then be used in portfolio construction. 2 Strategy Research supports the development and testing of investment strategies at both asset allocation and holding level. These include simulation, optimisation, filtering, regime, market and fundamentally based approaches. In each case, research can then be switched into production. 3 Portfolio Construction gives complete flexibility with 6 optimisation methods, Monte Carlo, n-tile, and VaR based. Optimisation includes extensive range of constraints, cost and market impact handling. Construction methods can be directly used in back tests and directly output portfolio weights for use in other applications. 4 Portfolio Tests to validate a portfolio strategy or construction, covering back testing and stress testing, rolling statistics analysis, VaR methodology and a complete range of tests to investigate portfolio, asset class and asset return characteristics through the analysis above, normality tests and GARCH modelling. 5 Portfolio analysis breaks down your portfolio to give understanding of the portfolio characteristics and comparison of portfolios through risk and return contributions, a full range of risk measures and ratios, drilling down from portfolio through asset allocation to holding level. 6 Attribution Ex-ante Risk Attribution through single and multi-period attribution using an exogenous factor model or a system generated dense model. Performance Attribution on an ex-post basis supporting return based attribution as well as factor based attribution. Both can be run on the back test generated portfolios as well as real portfolio returns. 7 Reports cover all the analytics, providing rapid output in a customisable format. Report packs enable efficient production of sets of reports for an account, or across multiple accounts. All report data can be output for use in other programs. 8 Risk Models use your preferred model or models, or use Vision to generate models on the fly to support back-testing. Even create blended models with variable blends through time. Supports use of factor models as well as dense models (asset level covariance matrix). 3/6.

5 Better Intelligence Through Analysis.... BITA Vision is an analysis tool that lets you look at strategies through back testing, portfolios through expost data and markets and stocks. Designed to be flexible and modular, it is agnostic to any given approach, meaning you can compare and contrast different measures giving different perspectives. It allows you to test qualitative judgements and perform quantitative analysis on qualitative portfolios as much as researching complete quantitative strategies. Alpha Research Risk models 1 Strategy Research 8 2 A better solution Reports 7 that flexes to 3 Construction meet your problem 6 4 Attribution 5 Portfolio Test Portfolio Analysis

6 Analysis, testing and understanding are the critical focus producing the data should be efficient and easy. BITA Vision A-Z: The following sections set the headline functionality available in BITA Vision. All the functionality is designed to join up, so portfolios created in one place are always available for analysis and reporting in other parts of the application. As part of BITA R&D, we continually add to the Vision functionality, embracing custom requests from clients. Alpha Research BITA Vision provides tools to build and test your own return forecast/signal models. The models can then be used in portfolio construction or ex-ante reporting. Typical approaches are: N-tile Analysis: Factor analysis for testing relationship between alpha signal and forward returns Alpha Signal Generation: Combine alpha signals into a single signal by using rule based or optimisation based routines Black-Litterman reverse optimisation with alpha calibration to derive a set of return forecasts from an existing portfolio Universe screening based on user defined custom logic Alpha hit rate reports and moving correlation, t-statistic, r-squared with actual returns All approaches support qualitative and quantitative research, portfolio management, stock screening and can easily be used in the Vision back testing routines. Back Tests Easy and fast back tests in Vision enable you to run your investment strategy for real using historic data, alpha and risk models, whatever approach your strategy takes. This enables validation, refinement and full understanding of your chosen model before applying in production. An essential tool for the development and on-going monitoring of a successful investment strategy. BITA Vision has extremely flexible problem definition capabilities it adapts to your problem and not the other way around, enabling you to develop a strategy proven to work in the past using contemporary data. Key back testing features include: Rule based portfolio construction Optimisation based construction N-tile based signal research driven Any regular or irregular calendar frequency including user defined Trigger based rebalancing User defined custom triggers, e.g. Rebalance when the ex-post tracking error breaches a threshold value for 5 consecutive days Regime switching: e.g. Switch to a short-term risk model when market volatility is high Time-varying constraints Create dynamic investment strategies by batching back tests to explore parameter space in a training period Simulating production level portfolio construction, including costs and market impact Daily performance calculation of portfolio/ benchmark Batching back test runs A user can: Use fixed weights with the portfolio rebalanced at the end of each period Use drifted weights moving with the market Specify a set of optimisation parameters and a 5/12.

7 historical date range Specify n-tile filtering rules User defined regime switching BITA Vision generates a sequence of optimum portfolios, using the specified parameters. At each time step it feeds the specified benchmark, forecasts and the previous optimal portfolio into the optimiser and generates a new optimal portfolio, which is then saved. If the user has provided an historic series of portfolios and forecasts, then those are used at the appropriate step of the back test. However, if they have provided a single value set, then the same forecast is used at each step and the portfolio rebalanced to maintain the specified asset weights or optionally drifted with the market. It stores each backtest portfolio as a new portfolio, labelled with the parameters used in the optimisation and generates a report page showing a sequence of plots showing: Cumulative returns of the portfolio and benchmark Portfolio downside Benchmark downside Active return (the difference between the portfolio return and the benchmark return) Portfolio return Benchmark return A wide range of analytics can be run on the back test series, e.g. time-series of risk attribution or factor exposures. Frontiers BITA Vision generates a range of efficient frontiers and surfaces including total risk, active risk and expected return axes. It has the following options: 3D surface presentation 2D active risk plotted against expected return 2D total risk plotted against expected return Based upon user defined date range and inputs Automatically outputs key frontier portfolios such as the best Sharpe ratio and minimum tracking error portfolios Optimisation & Rebalancing BITA Risk has 20 years experience in building optimisers, resulting in feature rich, powerful and efficient solutions. These can be used in BITA Vision to generate a single portfolio or create a back test series of portfolios for analysis and reporting and export. The data used for optimisation is user defined and any date range can be specified within the available data. In addition to a series of utility functions, there is an extensive constraint set. There is full support for long/short positions. The main methodologies supported are: Mean variance: Not a pre-canned process but one that defines risk and suitability in the context of your firm s investment offering. Robust: allows the user to apply multiple non-linear constraints while maximising return. For example: applying both active risk and total risk constraints in the same optimisation, illustrating the trade-off between tracking error, total risk and return. Glo+: combines mean variance and gain-loss optimisers, maximising gain, subject to a loss aversion. By combining the two optimisers in one utility function, the portfolio is optimised for variance and loss aversion simultaneously. Risk parity: generates a portfolio where each holding or group of holdings contributes equally to the portfolio total risk. CVaR: Optimises the expected tail loss. Reverse: as used in Black-Litterman and other applications, reverse optimisation in BITA Vision is used to determine the implied expected returns given the weights and risk of portfolio positions. Maximum entropy: minimising cross-entropy

8 giving optimal portfolio diversification. Custom: BITA welcomes firm specific requests for optimisation solutions and can develop these for exclusive use or as part of the generic product. Constraints: Wide-range of real-life constraints built up in response to user demands over 20 years Minimum trade and holding thresholds Maximum number of positions Linear and non-linear constraints Multiple non-linear constraints to multiple risk models Integer Constraints Flexible handling of cash flow Universe specific constraints Group constraints entire group, sub-group tiers, asset level Extensive cost/market impact model support including linear, piece-wise and non-linear model. Transaction cost Fixed cost, piece-wise linear, direct estimation Special constraints such as UCITS III Flexible constraints, e.g. if a problem is infeasible incrementally relax particular constraints in specific order or change the constraint setup entirely Performance Attribution BITA Vision allows you to quickly understand why your portfolio out-performed or under-performed a benchmark, or in absolute terms, using multi-period return attribution. This can be applied to historic portfolio data or the output of the backtesting module. Performance Attribution - Single period and multiperiod for factor and Brinson approaches Attribute ex-post total and active return to asset allocation and stock selection decisions Attribute ex-post return to different risk model factors Use time varying weights to model real portfolios Define portfolios in terms of trades and distinguish between capital events such as investments and withdrawals, and performance related events such as dividend payments and transaction fees Attribute performance with and without the interaction term between asset allocation and stock selection Portfolio Construction BITA Vision supports a wide range of ways in which to build a portfolio ranging from optimisation with rich constraints to rule-based approaches. The portfolios can then be analysed using the extensive reporting capability, saved and exported for use in other applications. All of the construction approaches can be used in the back testing module. Build portfolios with utmost flexibility Set proprietary rules and constraints Generate trade lists at a click Rule based or optimisation based Flexible transaction cost modelling Rule-based including universe screening that includes programmable complex logic on any data attribute Parameters From Date define start date of historic returns used for the risk model To Date define end date of historic returns used for the risk model Benchmark benchmark portfolio Forecasts return forecasts / models 7/12.

9 Risk risk model selection / definition Method: Portfolio optimisation See optimisation section above for five principal optimisation methods Return scatter Return auto correlation QQ-plot one sample QQ-plot two sample Method: N-Tile User defined filtering and rules to select portfolio constituents from the asset universe Method: Regime Switching Custom strategies defined by alpha and risk model switching and blending dependant on market volatility or other user defined parameters Method: Simulated Returns User can select re-sampling to be used in optimisation. Here optimisations are repeated with a different set of sampled returns as input and the mean of the optimal weights is the optimal portfolio Method: Sampling Historic Re-sampling (boot-strapping) Monte Carlo (zero mean) Monte Carlo (historic mean) Reporting: Use BITA Vision to rapidly produce high quality analysis reports on your portfolios, that are customisable in terms of layout, content and calculated fields. Reports can also be exported as CSV and other formats for use in presentations and further analysis. The user can specify any reporting periods within BITA Vision. On the basis of the period chosen, the risk and return models will be built or based upon that period as appropriate. Reporting: Return Characteristics Portfolio Return characteristics Multi-Portfolio return characteristics Portfolio rolling characteristics Reporting: Simulated returns Simulated returns distributions Simulated returns portfolio statistics Reporting: Risk Risk attribution (summary) Risk attribution (detailed) Implied return Implied return sorted Factor attribution Projected returns GARCH Reporting: Value At Risk (VaR) VaR Summary VaR 5 model backtest CVaR Backtest Reporting: Return Attribution Return attribution (summary) Return attribution (detailed) Return attribution (summary) no Interaction effect Return attribution (detailed) no Interaction effect Reporting: Historic Single-portfolio Performance Distributions Weights 6/6.

10 Portfolio statistics Rolling statistics Reporting: Historic Multi-portfolio Multi-portfolio performance Multi-portfolio statistics Multi-portfolio ex-post risk-return spider Multi-portfolio ex-ante risk-return spider Multi-portfolio best and worst return bar Reporting: Report Packs There are a set of reporting packs combining individual reports and others can be easily generated. Re-Sampling BITA Vision provides functionality to help reduce sample bias, the effect of data noise and anomalies in portfolio construction and reporting for a more robust portfolio. It is possible to test different resampling models quickly and easily in back-testing and compare with construction without resampling. Build re-sampled covariance and return models Historic re-sampling with insertion or Monte Carlo Use resampled risk and return models in portfolio construction for a more robust portfolio Return Analysis A comprehensive set of analytics enables understanding of the characteristics of an investible instrument s return series. This can direct to more appropriate construction and risk models e.g. if instruments are found to be non-normally distributed, using the BITA gain-loss optimiser may be more appropriate, and a Cornish-Fisher VaR model rather than parametric-normal. Analysis of historic portfolio or asset returns Demonstration of return characteristics, e.g. normality testing, skewness, kurtosis, Q-Q plots Identify auto-correlation Report economic measures including Sortino Ratio, Sharpe Ratio, Information Ratio, Jensen s Alpha View measures on a rolling basis, e.g. view the rolling volatility Risk Attribution BITA Vision provides a rich set of risk attribution analytics to help understand portfolio risk, where it is coming from and how best to change it. Attribute portfolio ex-ante risk and tracking error to assets, sectors, countries, regions, or other custom classifications Break down risk contributions into underlying components of correlation, weight and asset risk to help understand risk contribution values Quickly identify trades that have the biggest impact on reducing tracking error and the required turnover Using a factor risk model, attribute factor risk to different factors and understand the styles and factor tilts of the portfolio View implied returns for each asset in a portfolio to understand where risk bets are being taken and whether they are consistent with expectations View risk attribution through time Risk Models BITA Vision supports a wide range of risk models: Vision constructs full-covariance models on the fly based on user parameters and date range, for analysis and backtesting. Historic full-covariance risk models with and without exponential time-weighting Blend stressed-normal components of a risk 9/12.

11 model to model risk in different market conditions, with the user defining a stressed period and the degree of blend Factor risk models including historic models for backtesting A full range of 3rd party models can be integrated R-Squared custom risk models including factor models to support portfolio factor analysis BITA Global risk model covering individual assets and asset class series Stress Tests BITA Vision supports a wide range of stress tests that can be performed on an asset allocation or holding level portfolio, subject to the data being loaded. These tests can include: Back test through given market and economic crisis with full analysis of portfolio performance Test portfolio against a stressed market risk model, allowing the risk and risk attribution to be tested against a variety of levels of market stress Analyse portfolio return sensitivity and correlation to a series of economic, macro and statistical factors to understand the impact of a shift in one or a combination of these. For example, for a given set of factors (e.g. oil, indices) and a chosen date range, the output effect on portfolio return for a given shock in factor return (e.g. oil increases by 30%) both in case where factor return changes in isolation, and when other related factors move based on their correlations Factor impact effect on portfolio volatility/var/ CVaR for shifts in factor volatility, e.g. what happens to my portfolio volatility/var/cvar if the volatility of oil where to increase by 50% Diversification analysis: understand the risk contribution and covariance of the portfolio constituents and their individual factor characteristics Drawdown analysis, looking at the drawdown (loss from previous peak value) distribution over given periods By running risk statistics through time (rolling analysis) how these change through different market periods and events can be analysed and conclusions drawn on the portfolio s responsive characteristics VaR analysis, an analysis of the potential VaR using a range of different models. Monte Carlo simulations can be applied to a number of the approaches above to generate synthetic return distributions for combinations of events. For example: Using Monte Carlo analysis to carry out re-sampling of factor returns from different distributions. This would in turn give a resampled portfolio return distribution from which volatility/var/cvar can be calculated. The above could be output in a stress testing report pack which analyses the different stress tests. VaR CVaR GARCH Value at risk and conditional value at risk have become essential measures for managing the risk of portfolios. There are different ways in which to calculate VaR and CVaR and the choice of model depends very much on the characteristics of the portfolio and underlying assets. BITA Vision provides tools to calculate VaR and CVaR using different models and the ability to test and backtest how effective the models are for a particular portfolio, through best fit analysis. Model value-at-risk and conditional-value-at-risk (expected tail loss) for a portfolio using different methodologies including Historic, Age-weighted Historic, Parametric-normal and Cornish-Fisher Back test different VaR/CVaR models for a portfolio by comparing forecasts with actual returns Determine the VaR model best suited to a portfolio using tests such as Lopez and Blanco- Ihle Estimate and forecast return and volatility using ARMA/GARCH models, capturing volatility clustering 6/6.

12 FIVE examples of BITA Vision use: 1 Strategy & Alpha Research: Build, test and analyse your strategy, whether optimisation, n-tile, filtering or simulation based. Vision is a sophisticated and intuitive platform for alpha research, enabling the build of factor/asset rotation models and research on stock level alpha signals. 2 Portfolio Construction: Having researched the strategy, test it, challenge it, then run it live with BITA Vision. Vision s portfolio construction layer supports a wide range of approaches: optimisation, UCITS, simulation, filtering, triggers, all with deep investment functionality developed with managers. All approaches can easily be back tested fast and run both at holding and asset allocation levels. 3 Dynamic Reporting: Vision comes ready with a complete set of print ready reports and report packs, that can be rapidly run for single and multiple portfolios, as well as for analysing individual holdings. The user can customise the layout and fields, including calculated fields. The underlying portfolio and report data can be output to Excel, text files, or to a relational database as well as PDF. 4 Risk Management & Portfolio Analysis: Run ex-post and ex-ante analysis on your portfolio, calculating a wide range of risk measures, understanding risk contributions, what exposures the portfolio has and the risk characteristics and contributions through asset class down to holding level. All industry standard risk reporting approaches are covered in through the comprehensive report set. 5 Simulation and Stress Testing: Take your portfolio and test the implied returns on holdings, back test and stress test the holding / asset allocation / construction strategy, analyse the risk contributions and run risk and return attribution. Vision supports Bootstrapping and Monte Carlo simulation as well as stress testing. User defined triggers can incorporated to model event driven and regime switching strategies. BITA Vision is written in MATLAB, is installed on a user machine and takes data feeds from Excel / CSV files giving maximum flexibility to the user in working with their portfolios and data. The intuitive user interface is backed up with a detailed integrated help function. BITA Risk has a policy of continuous development for BITA Vision and incorporates client requirements into the product as part of this. As such, the breadth and depth of functionality continues to grow rapidly and clients are actively encouraged to provide input of their requirements to this process. 11/12.

13 Risk Analysis Return Attribution Ex-post & Ex-ante VaR CVaR GARCH Reporting Simulation & Optimisation One system joining the dots for risk management, strategy development, portfolio construction, analysis and reporting. Back & Stress Tests Alpha & Risk Models

14 bitarisk. BETTER INTELLIGENCE THROUGH ANALYSIS call visit bitarisk.com solutions from corfinancial. london boston new york. salerio. Automates the flow of securities and treasury trades from matching through to settlement. bitarisk. BETTER INTELLIGENCE THROUGH ANALYSIS Suite of applications addressing needs of private wealth managers, investment advisors, asset managers, quant teams. paragon. Fixed-income accounting hub delivers front-to-back office portfolio accounting and processing solutions. sanctionsmonitor. A sophisticated, easy-to-implement and easy-to-use sanctions monitoring, auditing and reporting tool. abraxsys. Comprehensive integrated banking platform delivering an industryleading banking service. costars. Retail fund/transfer agency solution providing end-to-end administration for collective investments. almeter. Control financial and business risk by assessing the impact of varying interest rate scenarios and hedging activities. kycmonitor. Case managment solution to provide clear evidence to the relevant authorities that effective and sufficiently robust AML controls are in place. Cor Financial is a trading name of COR Financial Solutions Ltd. Salerio, Paragon, BITARisk, SanctionsMonitor, Abraxsys, Co-Stars, Almeter and Cor Financial are all Trademarks of COR Financial Solutions Limited or its subsidiaries. COR Financial Solutions Limited All rights reserved