collapse

* '23-'24 SOTG Tally


2023-24 Season SoG Tally
Kolek11
Ighodaro6
Jones, K.6
Mitchell2
Jones, S.1
Joplin1

'22-23
'21-22 * '20-21 * '19-20
'18-19 * '17-18 * '16-17
'15-16 * '14-15 * '13-14
'12-13 * '11-12 * '10-11

* Big East Standings

* Recent Posts

Kolek throwing out first pitch at White Sox game by wisblue
[Today at 05:21:59 AM]


Big East 2024 Offseason by DoctorV
[April 26, 2024, 10:47:48 PM]


Marquette Football Update by Viper
[April 26, 2024, 08:10:52 PM]


2024 Transfer Portal by avid1010
[April 26, 2024, 07:48:11 PM]


Does Bucky NOT have a Basketball NIL? by WhiteTrash
[April 26, 2024, 03:52:54 PM]

Please Register - It's FREE!

The absolute only thing required for this FREE registration is a valid e-mail address.  We keep all your information confidential and will NEVER give or sell it to anyone else.
Login to get rid of this box (and ads) , or register NOW!

* Next up: The long cold summer

Marquette
Marquette

Open Practice

Date/Time: Oct 11, 2024 ???
TV: NA
Schedule for 2023-24
27-10

Author Topic: [Cracked Sidewalks] MU 1- to 3-pt favorite in FS, KenPom, Sagarin, Massey, Value Add; Fischer Breaks top 100 & MSU was 7-point dog in Value Add 3.0  (Read 1429 times)

CrackedSidewalks

  • Guest
MU 1- to 3-pt favorite in FS, KenPom, Sagarin, Massey, Value Add; Fischer Breaks top 100 & MSU was 7-point dog in Value Add 3.0

Denzel Valentine's absence cost Michigan State 6 points of offense and gave Iowa an extra 2 points - enough to change Michigan State from a 73-72 favorite last night at Iowa on KenPom to a 67-74 underdog. In fact, they lost to the Iowa team that looked so good at Marquette by a 70-83 margin.

We just completed the first true run of Value Add Basketball 3.0 (click here if you want detailed explanation of the result of the 1000 plus lines of code programmers completed this break as we calibrated much more advanced calculations). The new data projects a healthy Marquette to beat a healthy Seton Hall 70-69 tonight in the Big East opener - and KenPom, Massey, Sagarin and the Fox Sports projection also all project a one to three point MU win tonight on Fox Sports 1 in what KenPom ranks as the most exciting game of the day.

Value Add 3.0 Adjustment to kenpomMichigan St.Iowa
Pomeroy73.072.0
Valentine's Value Add 12 of 13 games8.9-2.6
Valentine Per Game9.6-2.8
71 trips in Iowa Game / 1000.710.71
Impact of losing Valentine-6.41.8
Pomeroy Projection - Valentine Loss66.673.8
Actual Final70.083.0

To go over the math using last night's game as an example, Pomeroy calculated a 73-72 MSU win based on all results to date. Valentine's current Offensive Value Add is 8.9 and Defense is -2.6 - meaning that is how many points he impacts each team's points per 100 possessions. However, that includes the ZERO rating for the game Valentine already missed to injury, so his per game is actually 9.6 and -2.8 - the fourth best total in the country. Since there were 71 possessions each in last night's game, losing Valentine cost Michigan State 6.4 points total and added 1.8 points to Iowa's score.

When you add those figures to Pomeroy's projection it yields the 67-74 projected loss for the No. 1 ranked Spartans. And lose they did.

SourceSeton Hallat Marquette
Pomeroy70.071.0
Massey70.072.0
SagarinUDby 0.6
Fox Sports ProjectionUDby 3
Value Add Offense24.621.2
Value Add Defense-6.0-5.9
Cross add previous two +82100.797.3
KenPom Tempo/1000.70.7
Neutral Court Score70.568.1
VA Home +2, Away -268.570.1

Because Seton Hall and Marquette are healthy for tonight as far as I know, no adjustment is made. You can go with your favorite projection - Fox Sports' listed spread is a 3-point MU win, Massey Ratings and Value Add (with rounding) says MU by 2, and Pomeroy and Sagarin have MU by 1.

Value Add 3.0 continues to be focused on individual player values, but to give a better understanding of how the sum of values add up to a total team performance, I ran the math above. The sum of the Offensive Ratings for Seton Hall players is 24.6, over three points better than Marquette's 21.2. The defenses are virtually even at -6.0 and -5.9. The assumption behind Version 3.0 is that a team of replacement players would score about 82 points per 100 trips going up against an AVERAGE D1 defense (this total will always be 20 points less than the national average moving forward).

Therefore both teams "start" with 82 points per 100 trips, and Seton Hall has enough offense to add 24.6 points per 100 trips when healthy, while Marquette has enough defense to take 5.9 back away - leaving Seton Hall projected to score 100.7 points in 100 trips against Marquette and Marquette scoring 97.3. Of course, neither team will get 100 possessions - Pomeroy projects 70 possession each, so we multiply by .7 and get a 71-68 Seton Hall win on a neutral court. However, teams average scoring 2 points better than average at home and 2 points less on the road, so with the game being played at Marquette the projected final is 70-69 Marquette (actually a 2-point favorite due to rounding of 70.1 - 68.5).

SourceSeton Hallat Marquette
Without Desi Rodriguez would lose6.4-1.8
Value Add Offense18.321.2
Value Add Defense-4.2-5.9
Cross add previous two +8294.499.0
KenPom Proj 70 Tempo/1000.70.7
Neutral Court Score66.169.3
VA Home +2, Away -264.171.3

Where Value Add 3.0 is more versatile is in quickly adjusting if a player misses a game from either team. Pirates small forward Desi Rodriguez is among the leaders in steals, 2-point shooting and 3-point shooting to rank as the 87th est player in the country with a 6.35 Offensive Rating and a -1.77 Defensive Rating. (other adjustments for position and another factor make the total slightly different than the basic Offense - Defense total).

If Rodriguez did not play (just for illustration, I do not know of any player expected to miss the game) then you would rerun the math with his contributions out, and Marquette would be a 71-64 favorite against a Seton Hall team without Rodriguez.

LSU's Ben Simmons is easily the top player in the country in the first run of the new system with a 15.31 rating, with Oklahoma's Buddy Hield in second. Luke Fischer breaks into the top 100 at 98th to be just behind Rodriguez for best player on the court tonight, while Henry Ellenson keeps skyrocketing as the third best player on the court tonight despite ranking outside the top 2000 after his first few games.

RnkPlayerTeamHtYrOffenseDefenseTotal ValuePGPerNBA?
1Simmons, Ben 25LSU6'10"Fr11gms9.99-4.1015.31SF*1.01596%
2Hield, Buddy 24Oklahoma6'4"Sr11gms10.87-2.0513.92SG*157%
87Rodriguez, Desi 20Seton Hall6'6"So12gms6.35-1.779.25SF*1.015
98Fischer, Luke 40Marquette6'11"Jr12gms7.17-0.799.07C*1.015
258Ellenson, Henry 13Marquette6'10"Fr12gms4.28-1.747.11C*1.01592%
304Carrington, Khadeen 0Seton Hall6'3"So12gms4.77-0.846.61SG*1
350Delgado, Angel 31Seton Hall6'9"So12gms3.18-2.066.32C*1.015
398Cohen, Sandy 5Marquette6'6"So12gms4.07-0.916.05SF*1.015
423Nzei, Michael 1Seton Hall6'7"Fr12gms4.22-0.635.93PF*1.015
470Sanogo, Ismael 14Seton Hall6'8"So11gms1.97-2.565.60C*1.015
640Whitehead, Isaiah 15Seton Hall6'4"So12gms2.30-1.364.65PG*1
867Johnson, JaJuan 23Marquette6'5"Jr12gms1.59-1.123.75SF*1.015
895Wilson, Duane 1Marquette6'2"So12gms2.12-0.513.63SG*1
1113Cheatham, Haanif 25Marquette6'5"Fr12gms1.77-0.132.90SG*1
1398Carter, Traci 21Marquette6'0Fr12gms0.21-0.872.08PG*1
1450Gordon, Derrick 32Seton Hall6'3"Sr12gms0.49-0.481.97SG*1
1672Ellenson, Wally 22Marquette6'6"Jr10gms0.00-0.451.46C*1.015
1798Singh, Veer 33Seton Hall6'7"Fr11gms0.450.361.23PF*1.015
1812Anthony, Rashed 25Seton Hall6'9"So8gms0.410.861.21PF*1.015
1825Soffer, Dalton 21Seton Hall6'5"Fr5gms0.390.231.20SF*1.015
1879Anderson, Braeden 4Seton Hall6'9"Jr10gms0.00-0.131.13PF*1.015
1952Carter, Myles 23Seton Hall6'9"Fr3gms0.072.431.04PF*1.015
1981Anim, Sacar 2Marquette6'5"Fr8gms0.010.291.01SF*1.015
1988Heldt, Matt 12Marquette6'10"Fr8gms0.000.341.00PF*1.015

Source: MU 1- to 3-pt favorite in FS, KenPom, Sagarin, Massey, Value Add; Fischer Breaks top 100 & MSU was 7-point dog in Value Add 3.0

mileskishnish72

  • All American
  • *****
  • Posts: 4551
And the books range from -1 to -3.5. Could be a good night to try for a middle.

Jay Bee

  • All American
  • *****
  • Posts: 9062
A couple of thoughts:

This is a good reminder to be aware of the limitations of systems and to remember what they truly are.

KenPom is a predictive system, however it is ---by design--- inflexible with regard to dealing with day to day realities. If you don't adjust for its inherent flaws, you're doing yourself a disservice.

Similarly, you'd want to make adjustments to ValueAdd... however, with a simple rec of ValueAdd I think you still have issues dealing with the reality of an Eron Harris getting more time in replacement of Denzel, as Eron is legit and underrated pre-game by ValueAdd..

End of the day, ranking systems, predictive systems, etc.... many of them are wonderful (KenPom and Value Add included), but without understanding - and adjusting for - their limitations, you done messed up.
Thanks for ruining summer, Canada.

bamamarquettefan

  • All American
  • *****
  • Posts: 1299
  • pudner-at-aspen-ideas-festival.jpg
    • Value Add Basketball
Similarly, you'd want to make adjustments to ValueAdd... however, with a simple rec of ValueAdd I think you still have issues dealing with the reality of an Eron Harris getting more time in replacement of Denzel, as Eron is legit and underrated pre-game by ValueAdd..
Excellent observation - this was the first feedback I got in private when the system was first published and NBA teams had me come meet. The best I can do to summarize the way I believe we have moved toward this fact is:

1. The numbers are being calibrated with the result seeming to be that this "domino" impact is what is really happening. When the best player on a team is out we don't really mean that all his shots and minutes go to the 9th best player on the team - we really mean the 2nd best player expands his minutes and shots some, then the 3rd best player does etc, so all are slightly less efficient because they are having to take tougher shots - and yes the 9th man would get minutes at the end of the dominos.

Also to your point, my actual advice is to look at the teams 9th man and subtract his points from the Value Add of the player not playing.

Finally, even if this 3.0 system worked PERFECTLY and was pinpointing exact spreads - Vegas would find it and start adjusting their spreads like they do with every analytical system that works. They start with these numbers, and then really just tweak a couple of points if there is other info they do not feel is incorporated in their systems.

The biggest reason their spreads on college games are so much more accurate now is systems get them so close as a starting point so they can really focus on pinpointing extra items like matchups from former players who understand the internals matchups, etc.

I don't gamble and like that because I feel less emotional evaluating and tweaking the system. The last year I did bet on college games AGES AGO analytics were not advanced - and I actually won the last 14 regular season games of the year against the spread Back then I could pick a game or two every night in the middle of nowhere and be confident I actually knew more than Vegas about the particular matchup and dynamics. Then March Madness came, and I quickly realized Vegas can track 16 opening round Thursday games and be much more precise - so I lost it all back and realized I didn't want to place bets on games any more. So now it's just a couple of friendly March Madness brackets - though I actually did get mad a few years back when I entered a poll and didn't realize there was a $20 fee - but when I paid it late and they revealed the rankings to me I was in line to win $6300 until the first two Elite 8 games knocked me out of first in a winner take all.

I am more interested in teams getting credit for pulling off wins against the odds, but obviously understand a lot of interest is driven by people who do play the spreads.
The www.valueaddsports.com analysis of basketball, football and baseball players are intended to neither be too hot or too cold - hundreds immerse themselves in studies of stats not of interest to broader fan bases (too hot), while others still insist on pure observation (too cold).