Fun with Massey | The Boneyard

Fun with Massey

Status
Not open for further replies.
Even more fun ... you can simulate a seven game series (or three or five) - prediction 7-0.
Change to Stanford v South Carolina - Stanford
Change to TN v Louisville - Louisville
 
Careful, though. Run those series simulations and you are going to occasionally get some surprising results. You will sometimes get the overall favored team losing some of those series.

Well, maybe except in UConn versus Prairie View. I've run seven simulations of 7 game series and UConn is still undefeated in them. Huskies won one of the games by 119-34. The 34 seems a little high to me though.
 
UcMiami said:
Even more fun ... you can simulate a seven game series (or three or five) - prediction 7-0.
Change to Stanford v South Carolina - Stanford
Change to TN v Louisville - Louisville
Expand ...

These kinds of simulations are usually run over and over; I ran the 7-game series a number of times and ND won the series in a few instances. I'm not sure if the distribution he displays at the bottom is how things look after N (many) simulations or what.

Definitely the simulation follows the expectation that UCONN would be favored. But not a clean sweep in all the simulations.

Edit: Dobbs beat me to it , while I was playing with the simulations :)
 
Massey data is impacted by KML health issues. Given that she should be pretty healthy by F4 I'd give the Huskies an extra 2 or 3 pts vs ND for a spread of about 10
 
Massey geeks: i always thought Massey estimated pt spread came from the diff b/w the two PWR ratings. Playing with the simulator that doesn't appear to be the case. I guess I had it wrong.
 
Suebird10fan said:
Massey geeks: i always thought Massey estimated pt spread came from the diff b/w the two PWR ratings. Playing with the simulator that doesn't appear to be the case. I guess I had it wrong.
Expand ...
They might just set the spread as the high point on a normal distribution curve.
 
Suebird10fan said:
Massey geeks: i always thought Massey estimated pt spread came from the diff b/w the two PWR ratings. Playing with the simulator that doesn't appear to be the case. I guess I had it wrong.
Expand ...
Actually it is true that the prediction for an upcoming game is based on the power differential. It shows what would happen in the most normal situation under standard conditions (UConn picks up 12 PFs, 23 assists, shoots 50%, etc.).

But the series simulator uses a probability-regression based processor that gives results based on all manner of possible situations, and if the normal power-diff score is 80-65, the series score will spit out scores that might be 84-61, 73-71, even 70-72. There are some who claim that say Baylor would have beaten Louisville 99 out of 100 times last season (sounds crack-brained to me), so if the Massey simulator was run 100 times for Baylor and Louisville last season, maybe only the one time would the score show the Cards on top, and by an 82-81 score. Or maybe not.
 
DobbsRover2 said:
Actually it is true that the prediction for an upcoming game is based on the power differential. It shows what would happen in the most normal situation under standard conditions (UConn picks up 12 PFs, 23 assists, shoots 50%, etc.).

But the series simulator uses a probability-regression based processor that gives results based on all manner of possible situations, and if the normal power-diff score is 80-65, the series score will spit out scores that might be 84-61, 73-71, even 70-72. There are some who claim that say Baylor would have beaten Louisville 99 out of 100 times last season (sounds crack-brained to me), so if the Massey simulator was run 100 times for Baylor and Louisville last season, maybe only the one time would the score show the Cards on top, and by an 82-81 score. Or maybe not.
Expand ...
You can still run it with the 2013 information. We don't have a way of knowing what it would have been before they played, but as of the end of the season, Massey lists Louisville as having a 9% shot over Baylor.
 
At one point way back when the data in the archive was still available, I checked the 3/31/2013 predictions and I believe that Massey had Baylor about an 18 point favorite. How that would have worked out for a 100 game simulation I am not sure. You could probably run simulations on two teams that have say an 85-67 power diff and get a pretty similar reenactment.
 
Status
Not open for further replies.

Online statistics

Members online
229
Guests online
1,932
Total visitors
2,161
Totals may include hidden visitors.

Share This Page

Forum statistics

Threads
164,078
Messages
4,381,345
Members
10,177
Latest member
silver fox


.
..
Top Bottom