So, thanks everyone for the data!
I've corrected my counting with the convention of ApothecaryGeist (i.e., not counting lands as
Myriad Landscape or
Krosan Verge) and put all the data together. Here's what I've learnt (I'll use the abbreviation CUL for colorless utility lands):
- In 5-color decks, we run on average 0.50 CUL, with a standard deviation of 0.71 and a standard error of 0.50
- In 3-color decks, we run on average 2.41 CUL, with a standard deviation of 1.87 and a standard error of 0.45
- No data for 4-color decks ?
- In 2-color decks, we run on average 3.32 CUL, with a standard deviation of 1.86 and a standard error of 0.40
- In 1-color decks, we run on average 4.27 CUL, with a standard deviation of 1.85 and a standard error of 0.56
I've defined the standard error as standard deviation divided by the square root of the sample size. It's clear that these sample sizes are quite small, so this error should be taken with a grain of salt. Also we don't know if the distributions are Gaussians, although we may expect them to become so in the limit of large sample sizes.
The conclusion is that my heuristic formula "# of CUL in a N-color deck = 5-N" is not so far from what we are doing on average!
I've also learnt that you people like Pramikon a lot
Please let me know if I've missed something!
Here are the data in case you want to analyze them yourself:
Garth 1
Niv-Mizzet-Reborn 0
Doran 1
Galea 1
Haldan-Pako 0
Licia 0
Mishra 1
Vaevictis 5
Breya 4
Merieke 5
Toggo-Thrasios 1
Animar 5
Pramikon 2
Licia 1
Yurlok 1
Mimeoplasm 3
Kathril 4
Atla-Palani 5
Pramikon 2
Meren 2
Imoti 5
Araumi 3
Xenagos 5
Hamza 2
Chatterfang 4
Lathril 0
Brudiclad 3
Edric 1
Scarab-God 2
Sisay 3
Rosheen 5
Gyruda 2
Kediss-Malcolm 4
Wort 1
Volo 3
Athreos 5
Arixmethes 6
Umbris 3
Emiel 4
Grenzo-Dungeon-Warden 8
Asmoranomardicadaistinaculdacar 2
Syr-Konrad 3
Braids 5
Birgi 4
Marwyn 3
Krenko 3
Kami-of-the-Crescent-Moon 5
Mikaeus-Lunarch 2
Ragavan 4
Haunt-of-Hightower 4
Balan 9
Questing-Beast 5