Season 3 Cup - Stats and Stuff
It is Cup stats time. When we get through to the top 16 there are often useful stats to be gleaned from the league.
Scorpions might be slightly dominant at the mo..
First up ratings. I have adjusted the ratings app to permit bulk entry. Apart from anything else this allows me to run scenarios through to see what the ratings do without saving them very quickly. So for amusement I pushed just the top 64 and top 32 results through to see what it would do to those players ratings (remember the ratings are monthly so also include all pod games so success in the cup can be countered by an appalling run in the pods and vice versa. Also remember that the ratings are more confident the more games they have to view so pushing 2 games max and 1 game for 31 people means they are likely to be marginally more extreme than they would otherwise be)
If we plonk the players into three groups we can see how wins/losses against the range occurs.
First up low ranked players who lost, almost exclusively to very high ranked players. As you would expect this hardly touches their rating at all
Kakita Leniel lost to a player 600 points above and his rating dropped a tiny 7 points. OgreZSan also lost to someone 600 points higher and his rating dropped 9 points (OgreZSans deviation is higher than Leniel so the system is not as confident as to position)
Next up mid ranked players who won their first game.
Theia was just under 1500 and won her first against a higher ranked and then lost the second to an approx equal ranked. Her rating ended up to 1545 so climbed around 70
Her opponent in the second game was 1478 and he also beat a much higher ranked then beat Theia (an equal ranked). At the end of the top 32 GoblnGuides rank had rocketed to 1665 so increased 187 points. This just reinforces the fact that the best way to increase is to beat higher ranked or almost equal ranked players (and not lose to much lower)
Up at the higher ends of the pool we have the current top seed Kingsley beating a much lower oppo then losing to a slightly lower ranked player in MindsDesire. Kingsleys rating would drop to 1942 so 88 points and out of the candidtate master zone. MindsDesires own rating went up to 1974 so 95 ish points. If no pod games occurred and this was the only rated L5R then the top 7 would be
Daidoji Micah 1960
Byron Kran 1957
Isawa Eujin 1930
Considering that all the others, bar Kingsley, still remain in the Cup and have even matches to win and you can see how the cup is the real push to ratings increases/drops for the elite players. If any of these players lose then they will get big drops so at a guess i would say one more round of victory would probably leave their rating much as above (unless they lose to GoblinGuide). We may therefore see a new top player but unusually Kinglsey is in a pod with Byron Kran and Daidoji Micah so has strong opponents. If no round robin win/loss occurs and one of those players beats the other two then that could also cause a huge swing up or down in the respective ratings.
Next let's take a brief look at my other favourite topic which is whether the ratings indicate any truth as to actual strength. In the top 64 there were 9 games that were approximately even (29% of field) . Ignoring those and looking at games with favoured opponents then only 4 games occurred where a much weaker player won (18%) meaning that in 81% of games the stronger player won.
In the top 16 this actually strengthened. 25% of games were even and only two went against the stronger player (13%) and so 83% of games went to the stronger player. The ratings therefore still appear to be a good predictor of success.
Finally we should briefly look at clan participation
Meh. Scorpion has steadily increased their strength until they were 50% of the field in the top 16. It does make an argument (when added to consistent Scorpion success in paper tourneys) as to how strong Scorpion are. When you consider these wins were against the board of the other clans then it makes it more impressive, or scary. Of the rest lion and Phoenix have both had slight increased with only Lion being above the average. Crab dropped off a cliff in the top 16 joining Unicorn in a heap at the bottom.
Due to two in clan top 16 games the Scorpion already have 25% of the field before they even start…