NBA MVP, DPOY, and ALL-NBA First Team Predictions

Dhruv Rokkam

Problem

Tracking the metrics responsable for calculating a players positive impact after the NBA all star game have been difficult for basketball analysts over the past few decades due to the leagues unpredicatbility with injuries and other factors, as well as the differeing opinions of nearly every analyst.

Solution

Who is going to be the NBA's most valuable player, Defensive Player of the Year, and which five players (2 Guards, 2 Forwards, 1 Center) will make up the NBA's first team All-Pro. The goal of this project is to determine which metrics have been most impactful in determining the leagues MVP and use those to find the current seasons MVP as well as DPOY and All-NBA First Team.

Impact

If we are able to create a program that determines the most important metrics when finding the MVP, we will be able to use those selected values to determines this years league MVP as well as the DPOY and All-NBA First Team.

Dataset

Details

We will use the Kaggle 1982-2022 NBA MVP Player Statistics Dataset to determine the metrics most important to calculating the MVP and then use the Kaggle 2022-2023 Player Stats to find the current MVP leader as well as the Defensive Player of the Year and who is apart of the NBA First Team all-pro

Variables

1.Games Played(GP)
2.Minutes per Game(MPG)
3.Usage Percentage (USG%)
4.Turnover Percentage (TO%)
5.Effective Field Goal Percentage (eFG%)
6.True Shooting Percentage (TS%)
7.Points Per Game (PPG)
8.Rebounds Per Game (RPG)
9.Assists per Game (APG)
10.Steals + Blocks per game (SPG + BPG)
11.Offensive Rating (ORtg)
12.Defensive Rating (DRtg)

import pandas as pd
df = pd.read_csv("NBA Stats 202223 All Stats  NBA Player Props Tool.csv")
df.head()

	RANK	NAME	TEAM	POS	AGE	GP	MPG	USG%	TO%	FTA	...	APG	SPG	BPG	TPG	P+R	P+A	P+R+A	VI	ORtg	DRtg
0	1	Dejounte Murray	Atl	G	26.4	56	36.4	24.8	11.1	135	...	6.1	1.5	0.3	2.3	26.5	27.0	32.5	2.4	9.6	114.4
1	2	Trae Young	Atl	G	24.4	54	35.3	33.2	17.2	478	...	10.2	1.1	0.1	4.1	29.9	37.2	40.1	1.6	11.7	116.6
2	3	De'Andre Hunter	Atl	F-G	25.2	51	31.7	19.8	9.1	176	...	1.4	0.5	0.2	1.2	20.0	17.1	21.4	1.3	16.6	110.1
3	4	John Collins	Atl	F-C	25.4	51	31.0	17.0	10.6	128	...	1.2	0.7	1.2	1.2	20.2	14.4	21.5	2.8	18.2	106.7
4	5	Bogdan Bogdanovic	Atl	G	30.5	36	29.6	20.2	9.2	50	...	3.0	0.9	0.3	1.3	18.0	17.6	21.0	2.0	40.7	114.0

5 rows × 29 columns

df_1 = pd.read_csv("NBA_dataset.csv")
df_1.head()

	season	player	pos	age	team_id	g	gs	mp_per_g	fg_per_g	fga_per_g	...	ws	ws_per_48	obpm	dbpm	bpm	vorp	award_share	mov	mov_adj	win_loss_pct
0	1982	Kareem Abdul-Jabbar	C	34	LAL	76	76	35.2	9.9	17.1	...	10.7	0.192	3.8	1.2	5.0	4.7	0.045	4.87	4.37	0.695
1	1982	Alvan Adams	C	27	PHO	79	75	30.3	6.4	13.0	...	7.2	0.144	1.4	2.2	3.6	3.4	0.000	3.45	3.05	0.561
2	1982	Mark Aguirre	SF	22	DAL	51	20	28.8	7.5	16.1	...	1.9	0.061	2.3	-1.6	0.7	1.0	0.000	-4.43	-4.48	0.341
3	1982	Danny Ainge	SG	22	BOS	53	1	10.6	1.5	4.2	...	0.5	0.042	-3.7	1.0	-2.7	-0.1	0.000	6.38	6.35	0.768
4	1982	Tiny Archibald	PG	33	BOS	68	51	31.9	4.5	9.6	...	5.2	0.115	1.4	-1.3	0.1	1.1	0.000	6.38	6.35	0.768

5 rows × 55 columns

Meathod

To solve this problem, we will use the second loaded dataset and find out the stats most highly emphasized when choosing the MVP. This means we will use a KN Classifier with the player who most matches the stats within those of the past MVP's. Some problems we might face could include the stat differences in eras. Scoring now is much easier than in previous years so PPG and APG might be inflated which means we will have to standardize a unit. Also games played is very important so that will have to have more value than a basic stat like minutes per game. In order to find the first team all pros, we will have to find out the values of each guard, forward, and center by creating a ranking system based off their stats and game metrics. Then choose the top 2,2, and 1 for their positions, respectively.