> ## Documentation Index
> Fetch the complete documentation index at: https://alphacastio.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Formula Reference
> Every operator, comparison and function available in the Alphacast formula editor — row, group, rolling window and date functions.
This is the complete grammar accepted by the Alphacast formula editor. It applies to [Calculate Variable](/steps/calculate-variable), the conditions in [Filter Rows](/steps/filter-rows), and the **Apply Formula** step.
[Regroup Entities](/steps/regroup-entities) does **not** use this grammar — it offers a fixed list of aggregations instead.
Reference a column with `@` followed by its name — `@Close`, `@Date`, `@sales`.
Function names are case-insensitive: `SUM`, `sum` and `Sum` all resolve to the same function.
## Operators
| Operator | Meaning | Example |
| -------- | -------- | ------- |
| `+` | Add | `A + B` |
| `-` | Subtract | `A - B` |
| `*` | Multiply | `A * B` |
| `/` | Divide | `A / B` |
| `^` | Power | `A ^ B` |
## Comparison
| Operator | Meaning |
| -------- | --------------------- |
| `>` | Greater than |
| `>=` | Greater than or equal |
| `<` | Less than |
| `<=` | Less than or equal |
| `=` | Equal |
| `!=` | Not equal |
| `<>` | Not equal |
## Logical
| Function | Syntax |
| --------- | ----------------------------- |
| `and` | `and(conditionA, conditionB)` |
| `or` | `or(conditionA, conditionB)` |
| `xor` | `xor(conditionA, conditionB)` |
| `not` | `not(conditionA)` |
| `is_null` | `is_null(@variable)` |
## Shift and IF
| Function | Description | Syntax |
| -------- | ------------------------------------------------------------------------------------ | ---------------------------------------------- |
| `SHIFT` | Shifts the index by a number of periods, giving access to lagging or leading values. | `SHIFT(variable, lags_or_leads)` |
| `IF` | Returns one value or another based on a condition. | `IF(condition, value_if_true, value_if_false)` |
```
Daily percent change of @Close
@Close / SHIFT(@Close, 1) - 1
Return 1 if the change is positive, -1 otherwise
IF(@pct_change >= 0, 1, -1)
```
## Row functions
Calculated across all values of the column, using only the current row — unlike the group functions below.
| Function | Description | Syntax |
| ------------ | ----------------------------------------------------------------- | ---------------- |
| `power` | Power | `power(A, B)` |
| `exp` | Exponential | `exp(A)` |
| `sqrt` | Square root | `sqrt(A)` |
| `log` | Natural logarithm | `log(A)` |
| `log10` | Base-10 logarithm | `log10(A)` |
| `sgn` | Sign: 1 if positive, -1 if negative | `sgn(A)` |
| `abs` | Absolute value | `abs(A)` |
| `trunc` | Truncate to integer | `trunc(A)` |
| `round` | Round to N digits | `round(A, N)` |
| `floor` | Closest integer lower than or equal to the value | `floor(A)` |
| `ceil` | Closest integer higher than or equal to the value | `ceil(A)` |
| `sin` | Sine | `sin(A)` |
| `cos` | Cosine | `cos(A)` |
| `tan` | Tangent | `tan(A)` |
| `hypot` | Euclidean distance — the hypotenuse of A and B | `hypot(A, B)` |
| `multiply` | Multiplies two values, equivalent to `A * B` | `multiply(A, B)` |
| `nan_to_num` | Replaces NaN with 0, and infinities with large finite numbers | `nan_to_num(A)` |
| `all` | True when every argument is true. Accepts any number of arguments | `all(A, B, ...)` |
## Group functions
Calculated within groups. Groups are defined by the columns passed as parameters; by default they are the entities of the dataset, excluding the date column.
| Function | Description |
| ------------ | --------------------------------------------------------------------------------------------------------------- |
| `rank` | Numerical data ranks, 1 through n — `rank(@Variable, @entity1, @entity2, ...)` |
| `cummax` | Cumulative maximum per group |
| `cummin` | Cumulative minimum per group |
| `cumprod` | Cumulative product per group |
| `cumsum` | Cumulative sum per group |
| `cumcount` | Cumulative count per group |
| `backfill` | Backfills NA values, up to a limit of consecutive values — `backfill(@Variable, limit, @entity1, ...)` |
| `pad` | Forward-fills NA values — `pad(@Variable, @entity1, ...)` |
| `pct_change` | Percent change against N leading or lagging values within the group — `pct_change(@Variable, N, @entity1, ...)` |
| `quantile` | Quantile position within the group — `quantile(@Variable, quantile, @entity1, ...)` |
| `count` | Count of values in the group |
| `nunique` | Number of unique values in the group |
| `first` | First value of the group |
| `last` | Last value of the group |
| `max` | Maximum of the group values |
| `min` | Minimum of the group values |
| `mean` | Mean of the group values |
| `median` | Median of the group values |
| `prod` | Product of the group values |
| `size` | Size of the group |
| `sem` | Standard error of the mean of the group |
| `std` | Standard deviation of the group |
| `sum` | Sum of the group values |
| `var` | Variance of the group values |
## Rolling window functions
Calculated over a rolling window, within groups. Groups are defined the same way as above.
| Function | Description |
| --------------- | ---------------------------------------------------------------------------------------------------------------------- |
| `rollingcount` | Count over a rolling window |
| `rollingsum` | Sum over a rolling window |
| `rollingmean` | Mean over a rolling window |
| `rollingmedian` | Median over a rolling window |
| `rollingvar` | Variance over a rolling window |
| `rollingstd` | Standard deviation over a rolling window |
| `rollingmin` | Minimum over a rolling window |
| `rollingmax` | Maximum over a rolling window |
| `rollingsem` | Standard error of the mean over a rolling window |
| `rollingrank` | Numerical data ranks over a rolling window — `rollingrank(@Variable, window_N, @entity1, ...)` |
| `rollingcov` | Covariance between two variables over a rolling window — `rollingcov(@VariableA, @VariableB, window_N, @entity1, ...)` |
## Date and time functions
| Function | Returns |
| --------------------------------------- | --------------------------------------------------------------------- |
| `now` | Current time — `now()` |
| `today` | Current date — `today()` |
| `year` | The year of the datetime |
| `month` | The month of the datetime |
| `day` | The day of the datetime |
| `hour` | The hour of the datetime |
| `minute` | The minutes of the datetime |
| `second` | The seconds of the datetime |
| `date` | The date part of the datetime |
| `time` | The time part of the datetime |
| `dayofyear` / `day_of_year` | The ordinal day of the year |
| `weekofyear` / `week` | The week ordinal of the year |
| `dayofweek` / `day_of_week` / `weekday` | Day of the week, Monday = 0 to Sunday = 6 |
| `quarter` | Quarter of the date: Jan–Mar = 1, Apr–Jun = 2, and so on |
| `days_in_month` | Number of days in the month of the datetime |
| `is_month_start` | Whether it is the first day of the month, as defined by the frequency |
| `is_month_end` | Whether it is the last day of the month |
| `is_quarter_start` | Whether it is the first day of the quarter |
| `is_quarter_end` | Whether it is the last day of the quarter |
| `is_year_start` | Whether it is the first day of the year |
| `is_year_end` | Whether it is the last day of the year |
| `is_leap_year` | Whether the date falls in a leap year |
The same grammar is available to AI agents through the `get_pipeline_formula_reference` tool on the Alphacast MCP server — see [Pipeline tools](/mcp/tools/pipelines).
## Next steps
How to use these formulas in a pipeline step.
Use comparisons and logical functions as row conditions.