Well hope this continues…
- Thread starter DwaynePurvis00
- Start date
It seems like most of the attention goes to Edey, Smith and Jones (and they deserve it), but these two guys create the spacing to make them much more effective than they would otherwise be.
So out of 12 shots Purdue made 4 last year and 5 this year. Not sure the average number of shots taken and how much variation is in that number. Don’t think we average 24 shots so wild guess is an average increase of 4.5 maybe 5 pts per game on 3 point shooting
And how many close games did we lose last year? Makes a huge difference.
1.7 more three-point shots made per game on around 21 attempts. Purdue shot under 25% 11 times last year. This year once.
5 pts is huge for the exact reason you state. What I don't know to put in proper perspective is points that are NOT 3 pt shots and would need to weigh that against the 3 pt shot for better understanding. The 3pt shooting in a vacuum may not tell as clear a story? However 5 points is huge as you state.
It would be interesting to compare also fouls drawn, Fts attempted and made, FGs attempted and made from last year to this year. Fouls alone can indicate the potential to force the opposition to abandon some of the original defensive approach which can result in more scoring in multiple ways for your team. So many variables or sources of variation in determining the model related to scoring even if not accounting for differences in teams or trends over a season. Sometimes an average is not a single population and can be a bit misleading without seeing the data visually in some frequency chart or histogram.
Another would be the potential relationship of 3 pt shots taken and 3pt shots made. Could there be an inverse relationship (maybe a large range in standard errors (variation) between the two or whether it is just a positive correlation at least in direction and then depending on the data "if" inverse...which came first the chicken or the egg for more or less shots taken relative to the makes.
Lastly, like test questions analyzed not only on its own like a four choice multiple question for too high or too low in the correct and wrong choices as well as the wrong choices being excessive, but that the particular question be highly correlated with the final result which takes me into a topic of concern relative to assimilation of the current culture as a result of DIF (differential item function) being in play, but that is for another day and perhaps a different site.
Lots of questions on the 3 ball, but a 4 or 5 point difference is great.
Last edited:
the difference i have seen this year is that our guards are driving more to the basket and taking mid range jump shots(especially smith).
my guess of 5 was pretty close. 20/21 would be a reasonable guess without knowing(my quick head math used roughly 18 or halfway between 12 and 24) I would want to see the frequency chart on the percent per game, but your data suggest 1/3 of the time Purdue shot 25% or less last year and only around 1% this year. It would still be interesting to see the data as I mentioned in this thread, but it is too much work for anyone unless someone had a class assignment that would allow some statistical application on the various numbers.
Not that I expect the model to be this, but merely as a visual understanding if I confused any with my previous text.
Points scored= .6 (2pt Fgs attempted)+ .38 (3pt Fgs attempted)+.7(Fts attempted)+ .42(fouls drawn)+ .32(offensive rebounds)+.12(defensive rebounds)-1.5(opponent steals)-.27(opponent blocks)
Obviously the constants were just guesses for the visual and you would hope the model variables would be highly correlated enough (large enough sources of variation) to explain most of the model or variables used. Interactions of the main sources of variation may have a larger constant than the variables independently and are not shown above, but essentially this is what I was describing for any that I may have confused previously.
If the error term was large and the model not explaining much of the variation, then did we leave out variables or interactions between main effects or variables...or was the data quite different earlier in the year than later in the year or preconference vs conference or home and away or...
Last edited:
Here you go for the raw data:
Samford - 16/29 (.552)
Morehead State - 8/23 (.348)
Xavier - 7/15 (.467)
Gonzaga (N) - 4/17 (.235)
Tennessee (N) - 4/15 (.267)
Marquette (N) - 10/21 (.476)
Texas Southern - 13/25 (.520)
@ Northwestern - 5/19 (.263)
Iowa - 8/25 (.320)
Alabama (N) - 8/18 (.444)
Arizona (N) - 10/24 (.417)
Jacksonville - 9/29 (.310)
Eastern Kentucky - 6/21 (.286)
@ Maryland - 9/20 (.450)
Illinois - 9/19 (.474)
@ Nebraska - 13/33 (.394)
Penn State - 11/24 (.458)
@ Indiana - 7/19 (.368)
@ Iowa - 9/26 (.346)
Michigan - 14/21 (.667)
@ Rutgers - 5/19 (.263)
Northwestern - 10/21 (.476)
@ Wisconsin - 3/11 (.273)
Indiana - 8/21 (.381)
Minnesota - 9/19 (.474)
@ Ohio State - 3/9 (.333)
Rutgers - 12/23 (.522)
@ Michigan - 7/24 (.292)
Michigan State - 10/20 (.500)
@ Illinois - 9/16 (.563)
Wisconsin - 9/18 (.500)
Home Games (16) - 159/353 (.450)
Road Games (10) - 70/196 (.357)
Neutral (5) - 36/95 (.379)
Total 265/644 (.411)
Median Game for percentage - .417 (Arizona)
looks like it is in order? I'm going to catch some games today, but may try to plot that post it in excel since that is all I have. Trying to think what all to plot?
Yes - chronological order of games played during regular season. Probably won't be the best of help on presentation or additional statistics. My old roommate was a stats major - he probably could......surely, we have some posters with the knowhow and knowledge for suggestions - I'm just a friendly nerd "helping" out where I can.
Is this not a metric people/teams strive to improve on the daily? Don’t these statistics validate that our team as a unit has improved at the top of competition? It was a goal, it’s a complaint fans and talking heads let alone coach Paint stated needed improvement to improve chances for winning tough games. I believe we also improved on decreasing TO/game, both goal. Analyze away, but the team thus far has been on track to improve on weaknesses from last season. Great job by our Boilers! Boiler Up!!!🖤💛🏀
What would be nice, but unable to do as needed with 3pt shots is any relationship it may have to another metric at various times and then its weight toward success. Was an increase in % for the 3 ball due to better shots (includes other variables) better execution of the same shot which also can be related to better timing and such and so there are a lot of variables in play. Doing so would take a lot of work of which nobody wants to do as well as multiple regression and even without all that work we don't know the constant of 3pt attempts relative to scoring on the whole and if there appears to be some correlation to another metric...maybe 2pt fg. Quickly the data suggest a 5 pt improvement over last year. How important was that to other variables when considering all the other variables. My guess is the model leaves room for improvement by not explaining the sums of squares of the variables and that the metrics we see leaves much of what we do not measure. What is that number?
FWIW, in my experience we learn that we don't know as much as we think in studies of this nature and it generates more questions than it answers. I always thought I would study kenpom to see what they do, but in predicting winners they no doubt make a model of similar nature for comparisons to the model of the other teams. Maybe they already provide their model for Purdue somewhere and the weighing of various metrics they use.
Similar threads
