Statistics breakdown. Our 13 regular, human
experts overall picked 2130 games correctly out of 3133 games for a grand percentage of
68.0%. Not a bad job considering the winner was at 71.5%. Consensus Curtis, the pooled knowledge
of our group, came in at 71.1%; only one expert was smarter than the collective unconscious of
the whole group (Heater14).
In order to help you strategize for next season, I ran an analysis to see what predicts good
performance. To do so I used the Spearman correlation, which basically tells you to what extent
being ranked first in a statistical category predicts a high ranking in Season Percentage, the
ranking that really counts. A score of 1 means a perfect correlation (wherever you finish in a
category is where you finish in percentage); a score of 0 means no correlation (the statistic is
worthless for predicting where you will finish); a score of -1 means inverse correlation (you
want to finish dead last in a category so you'll be first in the percentage). Here's the lowdown:
Weeks Won correlated .37, but the correlation was non-significant. Conclusion: you
don't need to win weeks, slow and steady wins the race (I'm looking at you, New H Man).
Exact Scores correlated .36, but that was not significant. Conclusion: just because
you're smart (or lucky) enough to pick an exact score (a trick turned an amazing 9 times in
3133 picks, or 0.3% of the time) means nothing in the grand scheme of things.
Close Scores correlated -.11, not significant (266 of 3133 picks, 8.49%). Conclusion:
picking close scores doesn't mean you are a successful picker (I'm glancing in your direction,
ingenue ROBERT).
Upset Index correlated -.01, not significant, meaning picking a lot of upsets does you
no good, although it doesn't hurt either, so why not! (thanks for the crazy picks, Chuck
Taylor).
Upset IQ correlated .91, and this correlation was significant. It isn't picking upsets
that does it, it is picking upsets correctly that does it. Of course, this is like most
statistics (general philosophical point here, people): it told us something obvious that we
already know! It's like if you have a 100 yard runner, you are more likely to win games. DUH!
And for you stats buffs out there, the Pearson correlation told us the same thing (0.86 in
the case of Upset IQ vs. Season Pct.; non significant for all other correlations).
| 1998 |
| Expert |
Weeks Won |
Close Scores |
Exact Scores |
Upset Index |
Upset IQ |
| Heater14 |
5 |
21 |
1 |
0.15 |
51.4% |
| Consensus Curtis |
1 |
-- |
-- |
-- |
-- |
| Dan |
3 |
17 |
0 |
0.16 |
47.5% |
| Matthias |
3 |
19 |
0 |
0.20 |
44.4% |
| Webmaestro |
5 |
19 |
1 |
0.22 |
46.3% |
| Kelly |
1 |
24 |
2 |
0.13 |
40.6% |
| Oz |
1 |
24 |
2 |
0.09 |
36.4% |
| Kevin |
1 |
17 |
2 |
0.14 |
37.5% |
| John |
3 |
23 |
1 |
0.15 |
37.8% |
| ROBERT |
0 |
28 |
0 |
0.12 |
34.5% |
| Saint Daddy |
3 |
18 |
0 |
0.13 |
36.4% |
| schmookie |
2 |
20 |
0 |
0.14 |
31.4% |
| Chuck Taylor |
3 |
12 |
0 |
0.27 |
35.6% |
| The New H man |
2 |
24 |
0 |
0.20 |
32.0% |
| Homer |
1 |
-- |
-- |
-- |
-- |
| Master O |
0 |
3 |
0 |
0.10 |
20.0% |
| Giri |
0 |
3 |
0 |
0.16 |
25.0% |
End Of Season Awards
At the end of each season since 1997, I have handed out at least one award to each expert that completed the season. Some of the awards are tied to statistical performance, while others are a bit more creative. For a description of these awards, see the hall of fame page on awards.
Picks Contest Championship: Heater14
Good Neighbor Award: Consensus Curtis
Comeback Award: Dan
Midwest Region Champion: Matthias
Tennessee Vols Award: Webmaestro
Most Improved Expert: Kelly
Picks Page Expert To Watch: Kelly
Rookie Of The Year: Oz
Freddie Favorite Certificate: Oz
Public Health Threat Award: Kevin
Loyalty Award: Kevin
Browbeater Of The Year: ROBERT
Send Me To Vegas Award: ROBERT
Vannez Gooch Award: Saint Daddy
Fall From Grace Award: John
Misplaced Priorities Award: Schmookie
Renegade Award: Chuck Taylor
Dale Helms Award: The New H Man
Bottom Feeder: Homer
Pulitzer Prize: Webmaestro (For "Musings of a Sore Loser")
©SJS Designs