The statistics shown at the bottom right of your screenshot are based on the default sample size of 10,000 records. The count next to the class values shows that all the records fit into the classes created.
The data is classified in Symbology based on the sample size. With a different classification method/dataset, it could be possible to have values outside of the class ranges. This could make the count next to the classes not equal all the records since some records would be excluded.
Something to note is that a warning should initially appear when Show statistics are chosen and the data’s records exceed the maximum sample size.
Increasing the sample size to best reflect the larger dataset should show the proper count/other statistics.