Week 0 Ranking Explained

When I made my Week 0 ranking, I didn’t explain my method. There were several reasons:

1. I was just trying to get the content going, so the point of my post wasn’t really the method used.
2. I was trying an experimental method, which I knew was not particularly sound. It was for my own amusement. I didn’t feel like defending myself from people who wouldn’t get it.

In the past, I would just adjust player mmrs using my proprietary methods, and then see which teams had the most mmr. The fact is, starting resource value is too important. I firmly believe that in the long run, mmr-based methods will outperform completely subjective ones. If I ranked teams according to total BoX-mmr, I would’ve gotten this:

crap
lucky
jabinski
princemeerkat92
moca
burrahobbit
box
doomcow33
steven
aeos
clare
ultragunner
skaterx7
clueless
MGHappyhobo
stonedspirit
deadprez
Charles the Baboon
velvelvel
JIVEE
lemonardo
polo5885
khillasaurus
flyingnightmare
themantis
bbtm
mappo
sharingan
seven thirty
bsvhsvj
fatsloth
jonsquid
jenny death
give up dude
maximus
♂K[A]livunkka♂
starend 2.0
spidez

A problem I was thinking about this season is, “what is the ideal mmr distribution for the players of a team?” For example, let’s use an average mmr of 3845. If one player has 8000 mmr, and the remaining four players are 2800 mmr, this is probably not ideal. Having all five players with 3845 mmr is likely a stronger team. Can this perfectly equal team be improved? What if we transferred some mmr from the 5 position to the mid-player? That’s probably better.

What if we tried to set our mmr distribution up, such that of the three Dota lanes, we could win two of them?

For my Week 0 power ranking, I divided each team into three groups: mid player, highest and lowest remaining mmrs, and 2nd and 3rd highest mmrs. These three groups were my “three lanes.” If I mis-assigned some player roles, I figured it wasn’t a big deal.

Then I sorted teams by the mmr of the mid player, and assigned a score from 1 to 38. I did the same for my 1+4 and 2+3 groups. I then added up the 3 scores, and ranked teams from lowest to highest. This was my final power ranking.