2015-03 to 2015-09 Player One WR (Map Data)

• Data contains ~1800 winners and ~2300 losers

 

I should clarify those are unique counts.

Let's say there are 4 players in the game.

Player A wins all his games, Player B wins 50%, Player C and D never win.

You now have 3 players who shown up under losers (B, C, and D), and 2 who show up under winners (A and B).

 

It's a pretty well known phenomenon in turn-based games that there's a first player advantage. It is why Pox and many other games has given those going first advantages.

This data does not show that high ranked players win more often as first player, however. It shows that when games occur between players of similar ratings, the first turn advantage is more prominent. Additionally, it shows that when players have greater differences in rankings, the first turn advantage diminishes relatively.

 

Um.. sure, but I don't see the purpose.

 

There's no specific reason why the system would make higher ranked players go first. It would have to be something that was specifically coded in on purpose, and there's no good reason to do that extra work in order to do something no one would want.

But I'll check the data to see if there's anything weird going on there.

 

Overall #:


Scatterplot of all games (excluded some small number of games where one player had -1000, 0 or some other erroneous rating):

 

It's statistically insignificant. In over 90k games, the higher ranked player went first just 132 more times. A coin flip has just as much variation as this. So I don't see how you can claim "it's still there."

In fact, depending how we split up the data, it shows something different too:



So if players have greater than 100 difference in ranked, the higher ranked player actually goes second more often (again, statistically insignificant).



Here it is again by month, where sometimes the higher ranked players go first more often, other times they go first less often - which is exactly what we'd expect if it's basically 50/50.

(Note: This is a good way to verify conclusions sometimes, incidentally, is by looking at different slices of your data to see if the conclusion holds true over time, or across variables that you'd expect to not affect the results.)

 

I don't understand.

 

Ah, the colors

 

Strangely, it seems like the opposite is true based on the time-line here. It's almost like it started out pretty even, but as players got used to it, Player 1 started winning more often on it.

 

It did, but the change basically made it harder for P1 to contest one of P2's fonts. and that happened in the middle of/late May.