Many players calculate Koga DPS spreadsheet incorrectly and frame the conditions to be too small. That being said, if all you're doing is making Set 1 to be Koga+AM vs Set2 to be Tsuru,
you're doing it wrong, making Koga much stronger than it really is. In practice, you will not have aftermath full time and many times, you're going to have to reapply aftermath. The true DPS of Koga is by using the Combined Set function on the spreadsheet.
You need to make 2 identical sets for Set1 and Set2 and it can happen in two ways:
Perfect Aftermath maintenance:
Set1: Aftermath, Set Ratio: 156, Minimum TP WS: 100, Tachi:Fudo/Shoha
Set2: Aftermath, Set Ratio: 24, Minimum TP WS: 300, Tachi: Rana
Ghetto Aftermath maintenance:
Set1: Aftermath, Set Ratio: 180, Minimum TP WS: 100, Tachi:Fudo/Shoha
Set2: No Aftermath, Set Ratio: 24, Minimum TP WS: 300, Tachi: Rana
Now look at the combined DPS: That's the DPS of Koga in an
ideal situation realistically. Remember this number.
Now, adjust the gear sets so that you're using the ideal hit build for your competitive G.kat. Now, vary the evasion, buff situations, rolls, songs, vary the hit builds between the two G.kat's and vary their gear so that they hit their respective hit builds in an ideal way (4 hit koga, 3 hit tsuru, 5 hit koga, 4 hit tsuru).
Now also consider that, the less you fight with AM up, the more the DPS will drop. Every second that you're walking to a mob is a decrease in DPS for Koga (ratio of Set#1 compared to Set #2 is decreasing), but does not affect other G.kats. Instead of 156 full seconds of attacking in aftermath, you might need to make it 140, 120, 100 even maybe. The more you walk, the greater Set 2 will be compared to Set1. This is really your DPS.
It's closer than you think. So close, that it all comes down to the player deciding when to use and not to use each G.kat. The peaks for Koga are great and I'll agree with Kenny that when the situations arise, it's the best. It would be disingenuous for me to label it the best G.kat without considering it's one greatest flaw. It's "one of" the best G.kats is still the most I'd give it.