By the Numbers, Vol. 2: Dec. 3 through Dec. 8
I owe you all an explanation. That explanation is in how the current system is being applied to games and therefore predictions. Last Sunday, in a game scheduled for Friday, the computer predicted Maranatha to beat Wayzata 71-70. This highlighted…
Access all of Prep Girls Hoops
Continue reading this article and more.
Continue ReadingI owe you all an explanation. That explanation is in how the current system is being applied to games and therefore predictions. Last Sunday, in a game scheduled for Friday, the computer predicted Maranatha to beat Wayzata 71-70. This highlighted one of the major shortcomings of the rankings as they currently are and has given me an idea to make them more accurate at the beginning of the season. Unfortunately, I do not believe I will be able to make that change until next year. Here is the short version of how the system comes up with its ratings, and a brief explanation as to why it came up with that particular prediction.
Scores → Adjusted for Opponents → Weighted for Blowouts → Averaged
All offensive and defensive scores go through this process and a team rating is calculated from the averages. In the beginning of the season, most of a team’s rating is determined by data from previous years, which is used as a starting point, but is not always accurate due to graduation and other reasons. At the time of the Maranatha vs. Wayzata prediction, 87.5% of Maranatha’s rating fell into this category and since they had been very successful in the past three years they had a very high rating. Which brings me to the other 12.5% and the oversight of the model as it currently exists.
For the current season, the model views every team the same until it has the data to determine otherwise. So when Maranatha wins 107-10 in their first game, the model is unable to adjust for opponent. Also, since it is the first game, it is unable to weight it for a blowout. Therefore, the outcome of that game is still 107-10 regardless of who they played, and Maranatha is most likely overrated for the current ratings due to that. The prediction then takes those offensive and defensive ratings, combines it with a small boost for home court advantage (although I’m not certain that it exists in high school girls’ basketball, it is a potential topic of study for later), and arrives at a predicted score. As of Friday, the predicted score for that game had changed to Wayzata 81, Maranatha 65.
As I said above, I do have a plan to fix this to a certain extent, but predictions in the early season will not likely be as accurate as they will be towards the end of the season. The model can only go off the data that it has, and that data is more limited in November and December.
Biggest Upset
Hill City/Northland 84, Lake of the Woods 52 (0.2% chance of upset)
Everything we saw heading into this game told us that this result was not going to occur, let alone by this score. Lake of the Woods was coming off of two wins in a row and Hill City/Northland was about what we had expected. This result will hopefully right the ship for Lake of the Woods and can be a springboard for Hill City/Northland.
Honorable Mention: Hope Academy 71, Christian Life Academy 50 (0.22%), Cambridge-Isanti 41, St. Croix Lutheran 32 (3.52%)
Biggest Gainers
A – Alden-Conger/Glenville-Emmons (last week Class A #95, currently #60)
A high-octane team, A-C/G-E made the jump this week by going 1-1, which I get seems a little weird. But they outperformed expectations with a 78-41 win over LeRoy-Ostrander and a 76-65 loss to another team trending upward in Lake Crystal-Wellcome Memorial.
AA – Rockford (last week Class AA #55, currently #23)
The Rockets won twice in blowout fashion this week, taking down Class AAAA Minneapolis Washburn 64-19 before beating section opponent Brooklyn Center 69-26. Rockford lone loss this season is to a highly ranked Holy Family team, so expect the Rockets to make some noise this season.
AAA – St. Cloud Apollo (last week Class AAA #39, currently #25)
Apollo began this week with a solid win over an improving Monticello team 65-59, and had that been their only win it still would’ve been an OK week. The Eagles followed that win up with an impressive rout of Albany 70-45 which propelled them up the rankings. They will have another opportunity to beat a solid team this week when they match up with Becker.
AAAA – Rochester John Marshall (last week Class AAAA #48, currently #34)
These Rockets have started the season 4-0, and while they have not played the best of the Big 9, they still have three conference wins. Add on top of that a win at Edina and John Marshall is off to a solid start. Things ramp up on Tuesday when Northfield comes to town.
Biggest Fallers
A – Cedar Mountain (last week Class A #67, currently #109)
Let’s hope this is where the bleeding stops. For the second week in a row, Cedar Mountain is our Biggest Faller in Class A. Although they were more competitive this week, they still weren’t able to be competitive enough, losing to GHEC/Truman/Martin Luther by 18 and Sleepy Eye St. Mary’s by 38.
AA – Pipestone (last week Class AA #19, currently #55)
Pipestone’s early season rating was largely due to previous success, and since they played their first game this week, a large change in ranking was possible. They proceeded to go 1-2 on the week with their only win being a 61-50 victory over Murray County Central. The Arrows finished off their week with defeats to Redwood Valley and St. Peter. In a Big South Conference that has many teams becoming more competitive, the Arrows will need to right the ship if they want to keep their ranking where it is.
AAA – Hermantown (last week Class AAA #10, currently #24)
The Hawks only played one game this week, and it was a loss to section opponent Chisago Lakes by a score of 59-55. Although Chisago Lakes has now won two games in a row, this section loss puts Hermantown in a tough spot in section 7AAA with it being their second in two opportunities.
AAAA – Edina (last week Class AAAA #14, currently #36)
Edina went 1-2 in three home games on the week which started with a 42-24 win over East Ridge in a game that was likely more the Hornets’ preferred style. There is no shame in losing to a solid Farmington squad, but the surprising result was the loss to the Biggest Gainer in Class AAAA Rochester John Marshall. With just one game next week, Edina has a chance to get things back on track with a section opponent coming to town in Chanhassen.
Highest Scoring Games
The highest scoring games section would prefer to only use regulation games, however, points are points, so we will include overtime games as well. Interesting fact, both Tartan and Triton make their second appearance on this list in a losing effort.
- Hopkins 98, Tartan 70 (multiple reports on the score for this game…)
- Blue Earth 94, Redwood Valley 73
- Lakeville North 83, Centennial 81 (2 OT)
- Grand Rapids 79, Bemidji 72
- Grand Meadow 78, Triton 72
- Worthington 85, Windom 65
Predictions for Next Week
WARNING: These predictions still take into account preseason ratings, which include data from previous years. Once a team has played eight games, that data is no longer factored into ratings or predictions. Also, predictions for later in the week will change based on the results of games early in the week, but I put them out there because it would be boring to just put out predictions for Monday’s games. A reminder, the PEV stands for the Potential Excitement Value of the game.
Monday
- Lake City @ Dover-Eyota (Lake City 61-59 PEV: 77.6)
Tuesday
- Minnetonka @ Chaska (Chaska 66-65 PEV: 86.8)
- East Grand Forks @ Crookston (Crookston 53-52 PEV: 83.2)
Thursday
- Hibbing @ Proctor (Proctor 49-40 PEV: 74.2)
- Eastview @ STMA (Eastview 59-55 PEV: 85.8)
Friday
- Shakopee @ Rosemount (Shakopee 47-46 PEV: 82.3)
- Peter @ Marshall (Marshall 53-48 PEV: 78.6)
Saturday
- DeLaSalle @ Grand Rapids (DeLaSalle 66-65 PEV: 85.5)
- Waseca @ Willmar (Waseca 59-51 PEV: 78.8)