Planual rules regarding Formulas.

Avoid using multiple IFs. It's better to split the formula into several line items and use LOOKUPs or alternative constructs.

If it takes you longer than one simple sentence to describe what the formula is doing, it's probably too long. Try not to mix expressions.

If the expression is repeated in the formula (or other modules), put it on a separate line item. “Calculate once, reference many times.”

Avoid multiple concatenations. Split commonly used concatenations into separate line items.

Make use of IF ISBLANK() when concatenating text with delimiters and set to BLANK to avoid unnecessarily populating cells . Because the text data type consumes more memory than other formats (Classic only) minimizing their use and avoiding unnecessary calculations is crucial for optimal model performance. 

Engine Applicability: Classic Hyperblock and Polaris

If a text string concatenation is needed, build it in the smallest possible dimensionality. This minimizes the size of the text strings and significantly reduces calculation effort and memory usage. Guards should be used in Polaris to reduce populated cell count.
Engine Applicability: Classic Hyperblock and Polaris

It's faster to check A=B rather than A-B=0.

In Classic, booleans take up 1/8th of the space as a number, so unless a numeric value is needed try to use TRUE or FALSE.

Never combine SUM and LOOKUP within the same formula expression. This combination forces the calculation engine to perform a massive number of calculations and creates slow intermediate relationship mappings, especially when Time is a dimension or when source and target structures differ. To drastically reduce calculation size and optimize performance, you must always split SUM and LOOKUP operations into separate modules and line items.
Engine Applicability: Classic Hyperblock and Polaris

Use these summary methods to minimize the use of additional line items and IF statements.

For long timescales or high cell counts, using the PREVIOUS() function is faster than the CUMULATE() function due to the potential number of cells required for the calculation. Use the expression: 'Calc line item' = 'data line item' + PREVIOUS('Calc line item') rather than CUMULATE('data line item').
Engine Applicability: Classic Hyperblock and Polaris
2.02-10a Short timescalesWhere the number of periods is small (Year granularity with a small number of years), CUMULATE is faster.

TEXTLIST() requires a lot of memory for calculations and should be avoided if possible. Using two dimensional modules and Boolean flags with ANY is a better alternative. Other alternatives are using FIRSTNONBLANK as well as LASTNONBLANK.

Avoid direct references to list item, for example, IF ITEM(list)=list.xx. Instead, use a SYS module with a line item having a Boolean format as this makes the formula more dynamic (multiple members can use the same logic).

Don't use POST for simple data offsets as OFFSET, LAG, or MOVINGSUM are more efficient.

Avoid hard coding using SELECT if possible. Use a constants module and LOOKUP instead.

2.02-14a VersionsIt's OK to use SELECT for versions.
2.02-14b For top-level items on listsIt's OK to reference the Top Level item of a list with SELECT. For other items (actual list members), it's better to use mock parent lists and additional modules.

Performing a FINDITEM() on blank values creates inefficiency, as the engine must traverse the entire list before returning a blank. To optimize calculation performance, evaluate the data's density and use conditional logic to bypass blanks. 

If the data is mostly blank check using ISBLANK: IF ISBLANK(Line Item) THEN BLANK ELSE FINDITEM(List, Line Item) 

If the data is mostly populated check using ISNOTBLANK: IF ISNOTBLANK(Line Item) THEN FINDITEM(List, Line Item) ELSE BLANK By putting the most common condition first, the engine works faster. 

If the source line item will never contain blanks: you can omit the conditional check entirely and simply use FINDITEM().
Engine Applicability: Classic Hyperblock Only

In multiple conditional statements, try and include a conditional to prevent further referencing in the formula if the condition is satisfied.

For formula efficiency, put the conditional with the most common occurrence first in the formula.

The engine works more efficiently when calculations are broken up into separate line items. So, break up formula expressions where possible. This is especially true for calculations that are referenced many times and/or calculations that don't change often.

Always refer back to the ultimate source if possible, to avoid creating more dependencies than necessary. This allows more parallel calculations to be run, increasing efficiency and speeding up calculations.

RANK is a calculation-intensive formula that can't multi-thread. Used in conjunction with large lists, it can lead to poorly performing calculations. The same applies to RANKCUMULATE and ISFIRSTOCCURRENCE.

TIMESUM formulas shouldn't be used in a line item that applies to Time as the calculation will be duplicated for each time period.

If the line item contains time periods, consider using MOVINGSUM or YEARTODATE. If you need the total for the entire timescale, consider using line item[Select: Time. All Periods].