July, 2007

The most obvious new feature in version 5.0 of Rainbow Analyst Professional is a completely redesigned front end. Nearly all Rainbow’s functions are now accessed via a dialogue box similar to the one shown here, which lets you change the selected workbook/ worksheet and the output medium (colour, map, report) as well as any function-specific options.The option to generate a worksheet map is also new in Version 5.0. Instead of applying colour directly to a worksheet (e.g. to indicate blocks of unique formulas), this option applies the same colour to the corresponding cells in a separate mapping worksheet. And because you can map any number of Rainbow functions to the same mapping worksheet, you can build up your own cumulative picture of the worksheet(s) you are analysing.

And in version 5.0 you can now put all your reports and maps in the same workbook (instead of each being a separate single-worksheet workbook). If you wish, you can even append them as additional worksheets in the workbook you are analysing.

Version 5.0 also includes important enhancements to Rainbow’s core functionality, of which the following are just a selection:

• The new Summary Report gives you a high-level view of a workbook, including e.g. data on Visual Basic code modules and average and maximum formula length per sheet, so that you can assess potential problem areas before starting detailed analysis.
• The Inter-Sheet Logic reports can now group relevant copied formulas together, producing a much more compact and readable report. And this technique has enabled us to speed up the inter-sheet precedents report by an order of magnitude, so that you can now quickly and easily generate a complete inter-sheet precedents report even for a model of some size.
• Nearly all Rainbow’s functions can now generate a report (grouping copied formulas where appropriate), rather than simply applying colour to the relevant cells. This means that Rainbow can now provide a much more comprehensive set of tools for use in a structured model audit process.

Version 5.0 is the first full version of Rainbow to include the Cell Inspector, an interactive function which displays the precedents and dependents of the currently selected cell, and also provides a best-guess description of the meaning of each cell. It will take you to each precedent or dependent (double-click on the relevant line) so that you can explore the logic and check out the meaning of each cell. Then click the back-arrow button to return to where you were before. The Cell Inspector title bar shows how many steps you are from the original cell (#1) where you started.

Cell Inspector uses a complex probabilistic scoring system to generate the cell descriptions; the screenshot on the left is a good example of how this works. Although of course the descriptions cannot be guaranteed to be 100% accurate in every single case, they have been proved to be highly reliable across a range of financial models, and they can provide a very fast indication of possible faulty logic.

As an example, suppose you run Cell Inspector on cell H74 in sheet Operations, and Cell Inspector shows this as Availability Fee - 30/09/2013, with the formula “=Inputs!H101*(1-$C$26)”. If you then look at Cell Inspector’s list of Precedents and see that Inputs!H101 is Availability Charge - 30/09/2012, you can immediately spot that there may be a problem with misaligned columns in different sheets referring to different time periods.

If you come across a complex formula, perhaps with several nested IF() statements, you can click on the Parse Formula button to see a structured layout of the formula which will quickly identify which are the relevant components of each IF() statement. And at any time you can click the Minimise button so that you can see the parts of the worksheet which may be hidden by the Cell Inspector dialogue box.

Many studies over the past ten years have shown that up to 90% of models contain significant errors. Generally these errors go undetected, but there have been some well-publicised cases where they have resulted in financial loss and embarrassment.

For example, an employee at a US investment fund omitted the minus sign on a figure of $1.2 billion while transferring data between spreadsheets, thus “creating” a $2.3 billion gain instead of the actual $0.1 billion loss. The company announced to shareholders a distribution of $4.32 a share; a month later, when the error was detected, it had to announce that there would be no distribution (further information at Dr. Panko’s website; see “Other sites”).

The vast majority of such errors could have been avoided by following a small number of key quality principles in building models, as explained below. (References to spreadsheet features assume the use of Microsoft Excel^TM, but are equally applicable to other spreadsheet software.)

The hallmark of a good model is that it is clearly laid out and easy to understand, so minimising the possibility of errors such as that mentioned above. Some key steps to achieving this are:

• Design, then develop, then test
• Separate the inputs from the calculations
• Use only one formula per row
• Avoid balancing figures
• Use protection and locking

Read the rest of this entry »

There are many issues that can present a challenge in modelling. Discussion of these can be found at some of the websites listed on our Links page, and there are also many books covering spreadsheet software packages, although these tend to focus on technical points rather than quality and business issues.

Some of the most frequently encountered issues are:

• Circular formulas
• Interest calculations
• Currency and inflation
• Mixed time periods
• Large models
• Visual Basic^TM for Excel^TM

Read the rest of this entry »