Re: Synchronated Updates

There are currently 60+ 4.5-star players in there. No shortage

OK, I ve replaced the old reputation values with 0-49 and replaced the calculation with lookups to the attribute table.
Retrieving may be working on the finer attributes of the J-League data, which is still to be merged with the main sheet, and that s the only thing that still has the old reputation values on it - but that s not a problem as I can convert the reputations once he s done.

The seven attributes on the right are how the skills currently come out, and are a product of the reputation and the ranked attributes.

The 7654321-system I think not flexible enough (though the hole work is fantastic).

Eg Let s see two a player with (just as an example scale) 0-10 attributes. They real values in the example are:
The Specialist: 9822121
The All-arounder: 6654343
Two very different player. But both are 7654132 (or 7654231) in the 7654321-system! The difference between the specialist and the all-around player have disappeared totally, every player become bit similar.

I suggest other system. Eg. a 0-10 or 1-5 for every attribute, which should be produce the real skills after a multiplication with the reputation and so on.

The 7654321-system I think not flexible enough (though the hole work is fantastic).

Eg Let s see two a player with (just as an example scale) 0-10 attributes. They real values in the example are:
The Specialist: 9822121
The All-arounder: 6654343
Two very different player. But both are 7654132 (or 7654231) in the 7654321-system! The difference between the specialist and the all-around player have disappeared totally, every player become bit similar.

I suggest other system. Eg. a 0-10 or 1-5 for every attribute, which should be produce the real skills after a multiplication with the reputation and so on.

The problem is, as soon as you use variable attributes that actually can vary in all possible combinations, you need a large, complex, iterative calculation to actually arrive at the correct number of SWOS points. This calculation exists and works fine, but is bulky and just cannot be used on large amounts of data. Hence the 7654321 thing.

I have been thinking about this problem though, quite specifically, and my solution (so far) is to have a check box for each player which can be used to denote he is an all-rounder , which will make the lookup values come from a whole separate table, on which the templates for players attributes are far closer together, resulting in players like the all-rounder in your example.

Problems I can foresee with this - checkbox might only get applied to more well-known players and will be unused lower down the rankings. Possible solution to this problem - poll Football Manager s data for players with low variance among attributes (possible and not difficult) and give them the all-rounder attribute. So that s my proposal for that.

The problem is, as soon as you use variable attributes that actually can vary in all possible combinations, you need a large, complex, iterative calculation to actually arrive at the correct number of SWOS points. This calculation exists and works fine, but is bulky and just cannot be used on large amounts of data. Hence the 7654321 thing.

I am not sure, the calculation must be complicated. If I will have a small time I try to propose an easy solution. Hope I will find one

I am finished the Excel, and it is very fast for 100 players data. It is 100 not 65k. I know the difference Now I make random data to try with 65k...

Oh, mine is instantaneous for 100 players.
Excel cannot even put the cells in place for 65,000 though.

Ready and for 65k it is fast as hell

I will upload it instantly.

Reduce it to a small sample of data and attach to a post if possible please - can t access file servers at work.