These stats can have different sample size. For example, if SB goes all-in, BB can only fold or call but not raise. If you want the sum of these 3 stats values to be 100, you need also to put "no all-in" filter on SB action.
Another thing. If you want these stats actually mean reaction to RFI, you need to put "First non-fold action on street" filter to SB raise. Right now it can be any raise on preflop no matter how many people already invested the money into the pot before the player on SB.
_______________________________________________
The sum of stats less than 100% can be only when all stats have the same number of opportunities.
Example, 3 hands:
You raise by going allin - Player calls (he has no opportunity to reraise)
You Raise - Player Reraises
You Raise - Player Folds
As a result, we can see the following stats:
Fold = 33%, Call = 33%, Raise = 50%
___________________________________________________________________________________
The sample size includes folds and all opportunities for the stat to occur. In the diagram, only the stat cases which went to SD are shown.
On the last screenshot, you can see that out of 8 hands in the sample there is 1 weak and 4 strong hands in the diagram.
The stat sample is total Opps, not Cases. The stat value formula is Cases\Opps*100, so out of 8 hands in the last screenshot, only 62.5% are actually cases for the stat spot, which is 5
Same for the next screenshot with 3+2+4 hands breakdown by strength, so a total of 9 out of 14 overall samples:
