F.INV.RT and FINV in ONLYOFFICE Spreadsheet Editor

• Introduced: Original introduction release is not established by the sources checked.
• Available: Documented in current unversioned ONLYOFFICE function help, checked October 18, 2026.
• Implementation inspected: Official SDKJS v10.0.0.169, commit 3679e0375a758d7dc59d31337a45461be3525ed8. This establishes a source snapshot, not identical behavior in every installed editor.
• Retired or deprecated: Retirement or formal deprecation is not established by these sources. A legacy spelling is not automatically a removed function.
• Testing: Examples are independently calculated or labeled source predictions. No ONLYOFFICE application or calculation-engine execution was performed. No native screenshot or general import/export guarantee is claimed.

Use English names and comma separators as shown; adapt separators outside quoted text if your locale requires semicolons. Keep full precision in calculations and round only the displayed summary.

=F.INV.RT(probability, deg_freedom1, deg_freedom2)

All three arguments are required. Probability is a decimal area. deg_freedom1 belongs to the numerator of the F ratio, and deg_freedom2 to its denominator. The output is a threshold on the F scale, not a p-value or an interval width.

F.INV.RT and its legacy FINV counterpart solve an upper-tail probability directly. For a 5% upper tail, supply 0.05.

Degrees of freedom are ordered. Reversing numerator and denominator without also applying the appropriate reciprocal/tail transformation changes the question. Keep the model’s variance ratio and its two df labels visible beside the formula.

1. Put 0.05 in A2 and 2 in both B2 and C2.
2. Enter =F.INV.RT(A2,B2,C2) in D2.
3. The independent expected result is 19.
4. Check it with =F.DIST.RT(D2,B2,C2); the mathematical target is 0.05.

For df=2 and 2, left cumulative probability is x/(1+x), while right-tail probability is 1/(1+x). Solving these equations gives:

left inverse at p = p/(1−p)
right inverse at p = 1/p−1

Thus 1/0.05−1 equals 19. This is an independent algebraic check, not an observed solver output. Format D2 as an ordinary number; a percentage format would mislabel the threshold.

For df=2 and 2, probability 0.5 gives threshold 1 in either tail. At the median, equality can conceal a tail-selection mistake; keep the 0.05/0.95 example as a second check. The small opposite-tail cutoff is 1/19, approximately 0.0526315789474.

With numerator df=2 and denominator df=4, the upper tail is (1+x/2)^−2. An upper-tail probability of 0.05 therefore gives 2*(1/sqrt(0.05)−1), approximately 6.944271909999. Use probability 0.05 directly in F.INV.RT. The result differs substantially from 19 even though the probability is unchanged.

For interior probabilities and suitable df, the mathematical complement identity is F.INV(p,df1,df2)=F.INV.RT(1−p,df1,df2). A reciprocal identity also exchanges df1 and df2 while exchanging tails. These are mathematical relationships, not promises of identical rounded outputs from every application or every extreme input.

Both inspected inverse paths parse the df arguments as integers before calculation. For example, the source-derived expectation for =F.INV(0.8,2.9,2.9) follows the df=2,2 equation and is 4. By contrast, the inspected F distribution wrappers do not explicitly integer-truncate df. Use integer model degrees of freedom for a predictable inverse/distribution check.

The source rejects probability at or below 0, probability above 1, df below 1, and df at or above 10^10, using its numeric-error category, conventionally #NUM!. Probability 1 passes this guard and corresponds mathematically to an upper-tail threshold of zero. This guide does not claim a native endpoint output.

Preserve probability precision. Rounding a very small upper tail to zero or a large left probability to one turns a finite quantile request into an invalid or boundary problem. Do not work from a rounded percentage label when the original numeric probability is available.

The inspected inverse helper uses up to 1000 bracket expansions and up to 500 interpolation/bracket refinements. It flags failure to bracket the target. The helper does not automatically set that same error flag merely because refinement reaches its loop bound. Source inspection therefore does not establish a universal convergence or precision guarantee.

For important thresholds, substitute the unrounded result into the matching tail function and compare with the input probability. Use a reasonable numerical tolerance rather than exact equality, and independently check at least one simple case. The rational df=2,2 examples above are useful for that purpose. Large df, extreme tails, nonfinite cases and imported array coercions remain outside this guide’s verified scope.

These are independent mathematical expectations or identified source-error predictions, not native application results. Preserve full input precision when reproducing them.

The legacy name FINV is covered in this guide. The modern name inherits FINV’s calculation method. This source relationship does not establish identical metadata, file serialization or every imported-workbook behavior. Legacy does not by itself mean formally deprecated or removed.

Use the exact probability direction, parameter order and data meaning above when moving a workbook between products. Similar names are not a general cross-version or file-format compatibility guarantee.

SQRT supports the unequal-df calculation; ROUND can format a reported threshold after the unrounded value has been checked.

• Official F.INV.RT help, checked October 18, 2026.
• Pinned SDKJS implementation, including parameter validation and numerical helpers.
• Official FINV help.