Note that
@Jackluyt's isn't quite correct. The raw numbers are accurate, but it isn't always right to pick the chest with the "best" option because it may be too expensive when you have bad choices. A simple example to cite what I mean:
Say you have only 3 chests: a chest costing 50 that gives 5 event currency; a chest costing 201 that gives 5 event currency; and a chest costing 1 that gives 0 event currency. You get offered two at random each time.
If you rank them by how much event currency each one gives, you'd rank them ABC (or 50, 200, 1).
Clearly, if you get the choice of chest A (50) you should take it. But if you are given the choice between chest B (201) and chest C (1), you should take chest C. Not because it gives more event currency than chest A (it doesn't) but because it is cheap so you have more keys left to buy future chests. For the 200 keys you 'save' by not buying chest B, you would be able to pick chest A a little less than 4 times (on average).
This tends to affect choices "further down" the table -- the best overall chest should still be your first choice.
This means that the best way to optimize is (italicized entries are different from Jackluyt):
Grand Prizes: 70, 138, 79,
62, 37, 49, 115 [For 5000 currency, expect 11.9 grand prizes, 11.3 daily prizes, and 342.9 KP.]
Daily Prizes: 49, 35, 37, 70,
30, 62, 79 [For 5000 currency, expect 10.1 grand prizes, 13.9 daily prizes, and 98.2 KP.]
Knowledge Points: 138, 62, 79, 35, 37, 30, 115 [For 5000 currency, expect 11.4 grand prizes, 10.8 daily prizes, and 495.7 KP.]