| Chicago Fanatics Message Board https://mail.chicagofanatics.com/ |
|
| The annual 'stats to predict the tournament' failure https://mail.chicagofanatics.com/viewtopic.php?f=101&t=111372 |
Page 1 of 1 |
| Author: | Psycory [ Wed Mar 14, 2018 12:02 pm ] |
| Post subject: | The annual 'stats to predict the tournament' failure |
Long story short, I run a NCAA pool (cheap 1 bracket $2, 2 for $4, 3 for $5; http://crsl.mayhem.cbssports.com/e?ttag=BPM18_cpy_invite_returning) and my graduate school friend and I try to predict the NCAA tournament using a statistical technique called a profile analysis. Every year it fails. So here are the final fours based on the two decision rules we use for ties in profiles: Difference Score (the more 'conservative' method) final four: Cincinnati, Xavier, Purdue, and Michigan State Final game: Xavier over Purdue Absolute value (the more 'liberal' method) final Four: Arizona, North Carolina, Villanova, and Duke. Final game: Duke over North Carolina Now you know who not to pick. |
|
| Author: | leashyourkids [ Wed Mar 14, 2018 12:05 pm ] |
| Post subject: | Re: The annual 'stats to predict the tournament' failure |
I filled out a bracket in your tourney using my proprietary system. If you want the answers, check there. |
|
| Author: | reents [ Wed Mar 14, 2018 8:07 pm ] |
| Post subject: | Re: The annual 'stats to predict the tournament' failure |
Psycory wrote: Long story short, I run a NCAA pool (cheap 1 bracket $2, 2 for $4, 3 for $5; http://crsl.mayhem.cbssports.com/e?ttag=BPM18_cpy_invite_returning) and my graduate school friend and I try to predict the NCAA tournament using a statistical technique called a profile analysis. Every year it fails. So here are the final fours based on the two decision rules we use for ties in profiles: Difference Score (the more 'conservative' method) final four: Cincinnati, Xavier, Purdue, and Michigan State Final game: Xavier over Purdue Absolute value (the more 'liberal' method) final Four: Arizona, North Carolina, Villanova, and Duke. Final game: Duke over North Carolina Now you know who not to pick. Is it good on the liberal method, I have 3 of the Final 4 and that's my title game and pick |
|
| Author: | Psycory [ Fri Mar 16, 2018 9:10 pm ] |
| Post subject: | Re: The annual 'stats to predict the tournament' failure |
reents wrote: Is it good on the liberal method, I have 3 of the Final 4 and that's my title game and pick It's prediction accuracy varies by round. First round it has a 72.3% accuracy and then it drops for rounds 2, 3, and 4. Then at five it picks back up and round 6 it is at 93% accuracy. So if we get the correct final game, we can predict the winner. The liberal method tends to have more upsets...but they tend to happen in the later round and that's why it falls apart in accuracy. |
|
| Page 1 of 1 | All times are UTC - 6 hours [ DST ] |
| Powered by phpBB® Forum Software © phpBB Group https://www.phpbb.com/ |
|