Define Statistical Operations
The following table defines the Query Tool statistical operations.
Operation | Description |
---|---|
Rolling Operations | |
Geometric Compound | (1 + C) = (1+ A) * (1 + B)… (1 + N) For example, rolling 3-month return. The Geometric Compound Operation is the only operation where the percentage values retrieved from the database are first converted into growth rates, then this formula is applied and the result is converted back to percentage. For annualizing, the query uses a day count of 365 for daily frequency or if business calendar is not used, and 366 if the actual number of intervening days between begin and end date contain 29th of February. Annualized Return = ( ( (linked growth rates) raised to (exponent) ) - 1 ) * 100 Daily frequency or non-business calendar exponent = 365 or 366 / actual number of days) Weekly frequency exponent = ( 52 / actual number of weeks) Monthly frequency exponent = ( 12 / actual number of months) Quarterly frequency exponent = ( 4 / actual number of quarters) Subperiod frequency = 365 or 366 / actual number of days) This is consistent with the way annualization is performed in the Performance Analysis reports. |
Lag Percent Change | C =((A – AN Periods Ago)/ AN Periods Ago)*100 For example, percent change in market value vs. market value one quarter before |
Maximum | Max(A) For example, largest return over the period. |
Minimum | Min(A) For example, smallest return observed over the period. |
Average | Mean(A, B, C) For example, arithmetic average return, or average daily net assets. |
Standard Deviation | Stdevp(A, B, C) For example, standard deviation of 3-years of monthly returns. |
Sum | Sum (A, B, C) For example, sum of the monthly income distributions. |
Growth of Dollar | The Growth of a Dollar Operation has a Configurable Base Value. |
Series Derivation Operations | |
Addition | C = A + B For example, market value + accrued income. |
Difference | C = A – B For example, Fund Return – Index Return |
Geometric Compound | (1 + C) = (1+ A) * (1 + B) For example, fund return – benchmark return |
Geometric Difference | Calculates a single period return that reconciles the difference between the compound to date returns. It is the ratio of growth rates converted back to percentage rates. |
Natural Log | C = LN(A) For example, continuously compounded return. For adding returns over time. The percentage value stored in the database is first converted to growth rate. If this growth rate is positive, only then natural log operation is performed. |
Ratio | C = A/B |
Using these statistics, you can manipulate not only returns and related performance statistics, but also analyze market and economic time series. For example, the following image shows a report showing the quarter on quarter growth in US GDP vs. UK GDP, as well as the annual volatility of these growth rates.
To produce reports like this, store the economic time series in the PERFORM database as a series of index values. You can store a GDP time series in the same way as you store a market index level. In this case, the GDP level was set at 1000 at a particular starting point and then adjusted each period to reflect the growth rate. This data is available from many financial data vendors and can be loaded using any performance returns uploader.