Ranking Analysis – Wildcard Round
Hello folks, sorry I’m late and posting in the middle of the divisional round, but here is the analysis of the rankings after the Wildcard round. One big mover as a result of the Wildcard games, and that is Denver—a good illustration of how seemingly unrelated games can affect the system’s overall assessment of a team.
Despite all of last week’s games being counted as upsets by Beatpaths, the ranks don’t change a great deal. The ranks remain fairly stable and the variance is down, despite Denver’s rapid ascent from 19th to 5th. Given that there were only four games, we might expect that the rankings would be more stable. We’ll see if the ranking changes after the four games of the divisional round are comparable.
Pick Confidence – Divisional Round
TT already gave the official Beatpaths picks for the Divisional round. Below is the ‘confidence’ of the Beatpaths system in its picks, based on a comparison of the ‘EdgePower’ scores of each team. Compared to last week, the system is about twice as confident in the picks for this weekend.
| Matchup (Winner-Loser) |
“Confidence” (out of 100) |
EdgePower comparison (predicted winner – predicted loser) |
Result |
|---|---|---|---|
| New Orleans-Arizona | 38.31 | 83.47-45.16 | ✓ |
| Indianapolis-Baltimore | 35.88 | 81.85-45.97 | ✓ |
| San Diego-NY Jets | 21.78 | 84.68-62.90 | ✗ |
| Dallas-Minnesota | 13.71 | 77.82-64.11 | ✗ |
As usual, I’ll return to fill in the results.
January 18th, 2010 at 10:20 am
At this point I don’t think there’s going to be any more conflict picks, so I did a quick run-through of the whole season to compile a comparison of iterative and standard.
Because I didn’t want to compare MOOSE’s tiebreaker to your tiebreaker, I only looked at games where iterative had a beatpick that conflicted with your pick. (There were no cases, as far as I can tell, where standard had a pick and iterative did not.)
Here’s the summary:
Week 7: Iterative got CIN→CHI right.
Week 11: Iterative got MIA→CAR right, and SD→DEN wrong.
Week 12: Iterative got NYJ→CAR right, and BAL→PIT wrong.
Week 13: Iterative got NYG→DAL right, and PHI→ATL wrong.
Week 14: Iterative got MIA→JAC right, and PHI→NYG wrong.
Week 16: Iterative got PIT→BAL wrong.
Week 17: Iterative got DAL→PHI and NYJ→CIN right.
Week 18: Iterative got DAL→PHI and NYJ→CIN right.
Overall, Iterative went 9-5 in conflict picks. It’s intereating to see the patterns here – iterative clearly thought more highly of Miami and thought less of Carolina, and benefited from that. On the flip side, iterative thought less of Philly, and went 2-2 on those conflict picks.
The most humorous, I think, is how iterative picked the road team in the PIT/BAL series both times, while standard picked the home team. Good thinking there, standard algorithm.
So, anyway, my arguments for iterative, in a nutshell:
1) The graph is more vertical and displays more of the data in a visually accessible way.
2) Significantly better retrospective pick record.
3) Slightly better predictive pick record.