I'm afraid that's not how you do it. Slicing and dicing in measures is not usually done via putting a filter in them for a specific attribute's value but rather by placing the attributes in columns/rows of matrix/table/visual and then letting the measure work on the filter context set by these selections automatically. On top of that, you should never place the table name in front of a measure, in contrast to column names where you always have to precede them with the table name.
So, your measure should just be:
[seconds/rounds]
and to obtain the matrix above you just put the attribute's values on the columns (Game 1, 2, 3) and drop the measure onto the matrix. A good measure needs to sense what kind of attribute the context has been set to and react accordingly.
If you want your measure to return averages over the months, you have to code the measure accordingly. It won't do it on its own, it has to be coded. I'd say something like this (but don't know the whole story, so it might be a bit off):
total seconds = SUM('Table'[seconds])
[seconds/rounds] =
averagex(
// MonthID must be a unique identifier
// of each month. Not just the month
// number. A unique id across all years.
distinct( 'Table'[MonthID] ),
[total seconds]
)
I kindly suggest you learn the basics of Power BI from here for instance. This looks to be a free course and Enterprise DNA have top-quality material about Power BI.
