Strategy analyzer

Strategy Analyzer analyzes a given portfolio and its sub-portfolios by comparing the portfolio’s strategy against the portfolios' performance analyzed by Analytics+. Currently, the features include:

Value added return based Contribution analysis
Brinson Attribution analysis
Performance analysis against the benchmark defined by the strategy

The portfolio(s) are analyzed against the predefined strategy (from a given portfolio) using Analytics+ functionality. As a result, the strategy is broken down to:

Analysis of the current positions (group,sectors), average share's, performance, volatility and attribution
Analysis of the current strategy (group,sectors), target shares including mix and max, volatility, attribution
Brinson Attribution analysis including timing (allocation), selection, interaction and total
Performance analysis: alpha, beta, tracking error and sharpe

Analysis supports also:

Analysis against the selected strategy (multi strategy support in FA 3.7 and onwards)
Choice between using either TWR Net or TWR Gross
Selecting benchmark rebalancing strategy (NONE, DAILY, WEEKLY,MONTHLY,YEARLY)
Analysis based on frequencies e.g. monthly by month analysis over one year period
Risk free rate: fixed, based on interest rate on an analysis start date or an index performance representing a risk free rate

Currently, the strategy analyzer is available as an API and also as an extension jar, which means in practice, its logic can be used in customer extensions. In addition, there is also a widget that can be used in the dashboard.

Formulas for contribution and attribution

Overview of the logic

The calculation is done from bottoms up - that is, the lowest sector level attribution related figures are calculated first and the aggregated up. Basic steps are:

Analyze the portfolio(s) using Analytics+ with grouping ‘Group’, ‘Sector’, ‘Security’ using the given group code. Alternatively using the ‘Frequency’ to do multiple periods at once.
Each analysis result is then analyzed against the given portfolio strategy

Formulas

Benchmark return

Total, cumulative return over a given period (0…N)

Value added return

The intra-period (e.g. daily) added return on top the cumulative return up until this day. Please note that in this context this refers to the value added return on a cumulative index. This is different from, and not to be confused with, the more commonly known concept of value added return that refers to the return over a benchmark.

Position

Benchmark

Contribution

Defined as the contribution a single constituent (e.g. position) has to the total group of constituents (e.g. portfolio). Interpreted as how much a single factor is contributing to a total, and can as such be done on any level or breakdown. E.g. positions can contribute to the total TWR of sector groups, which in turn can contribute to the total portfolio TWR, when split up into such parts. Also known as absolute return attribution. Contribution itself is not done against a benchmark.

Position

Benchmark

Attribution

Attribution analysis is the study of the active decisions a portfolio manager has made in terms of allocating (in sectors) and selecting (e.g. stock-picking) investments differently from a benchmark, over a given period. The main point of the analysis is to find how much these investment decisions attribute (or “explain”) the over/under return the portfolio had against its benchmark (the total attribution).

The allocation is measured in different components, where sector (e.g. Health Care) is common to analyze. The selection purely analyzes the choice of investments within one sector (i.e. health care stocks picked). Attribution is always done against a benchmark, which ideally should have very similar components as the portfolio since these are the basis for comparison. In the current implementation, holdings-based attribution is used (so not down to transaction level/transaction-based).

Benchmark weights

Benchmark weights are used in:

Attribution calculation
Average weight calculation
Benchmark return calculation

There are two different weights applied

Start weight of the benchmark - Used in attribution calculation and benchmark return calculation
End of the day weight of the benchmark - used in average weight calculation

Start weight of the benchmark

Start weight of the bechmark is equal to the end of the previous day weight of the benchmark with on exception. When the benchmark is rebalanced, the start weight of the benchmark is set to be equal to the new weight defined by the strategy.

End of the day weight of the benchmark

End of the day weight of the benchmark is calculated by allowing the start weight of the benchmark to drift along with the benchmark’s daily return. That is, the benchmark weights are adjust with their value change. There is one exception to this rule: when the rebalance frequency is defined to be Daily, the end of the day weight of the benchmark will be equal to the start weight of the benchmark (ie. no drift).