My first NBA model tracking thread

[pico]

spent yesterday morning figuring this out. this model just take the team's relative strength and ability to cover spread and calc the possible outcome of the game. i started with this very simple 2 variable model, and calculated the results for last night and calibrated the parameters.

for last night, my model picked (all spreads)
NYK-PP
LAC-NN
CHA-NP
DET-NP
ATL-NP
DEN-NN
SAC-NP
LAL-NN

overall: 5-2-1, then i looked closer, my model has 3 catagories of results PP,NP, NN (p=positive, n=negative).

here is last night;s results based on the 3 catagories:

PP= 1-0
NP= 1-2-1
NN= 3-0

these sample size are way too small to yield any meaningful conclusion. so i'll live test this prediction model for this week.

For tonight's game, here is the model results:

ORL-PP
OKC-PP
BOS-NP
MIA-NP
MEM-NP

(these are all predictions in respect to the closing line...i suggest you not to take this model too seriously at this point cuz i am still testing it)

[LUNT101]

can you explain the significance of PP NP and NN in more detail please

[pico]

can you explain the significance of PP NP and NN in more detail please

my model outputs an number. it can be positive or negative. positive means that this team has a higher chance of covering the spread based on the matchup, negative means this team has a lower chance of covering the spread based on the matchup.

it is actually two predictions, one for each team based on the game match up. sometimes, i get postive outputs for both teams, sometimes i get one positive and one negative.

i thought about it yesterday, my model makes intuitive sense, so i think it has potential.

[LUNT101]

so, for example, you have NN next to the Lakers game. that means that you put the lakers and rockets into your equation and it said both of them have a low chance of covering the spread? how do you interpret that and how does that lead you to take the Lakers?

[pico]

so, for example, you have NN next to the Lakers game. that means that you put the lakers and rockets into your equation and it said both of them have a low chance of covering the spread? how do you interpret that and how does that lead you to take the Lakers?

yes, for some reason my model said both teams might not cover the spread, but the number for HOU is a very large negative number and the LAL output is a small negative number within single digits. that is why i pick the LAL.

come to think of it, rather than P, N, i should just find the difference between the two predictions.

let's see how tonight goes, and i'll update it tomorrow.

[LUNT101]

good luck, ill be watching

[pico]

For tonight's game, here is the model results:

ORL-PP - loss
OKC-PP - win
BOS-NP - loss
MIA-NP - loss
MEM-NP - win


over all: 7-5-1
by category

PP= 2-1
NP= 2-4-1
NN= 3-0



i found a better way to calibrate the model. i'll run the new version for these games and see how they compare. give me 30 min...

[MickChunky]

Interesting, keeping an eye on this

[pico]

results did not change. so let me put the estimation difference next to the yesterday and today's results:

sunday
UTA-NYK -193 win PP
DAL-LAC -32 win NN
TOR-CHA -58 loss NP
BOS-DET -20 loss NP
ATL-OKC +167 push PN
MEM-DEN -1046 win NN
GSW-SAC +518 loss PP
HOU-LAL -16 win NN

monday
POR-ORL -9 loss PP
OKC-IND +9 win PP
TOR-BOS -68 loss NP
NJN-MIA -152 loss NP
MEM-PHX +65 win PN

[pico]

over all: 6-6-1...hmm not so good. let's see how it does for the rest of the week. i might scrap this if it turn out to be 50%.

[pico]

Tuesday
UTA-PHI +51 win
DEN-CHA -144 loss
MIL-CLE -60 loss
ATL-CHi -68 loss
LAL-DAL +11 win
NYK-SAN -41 win
DET-SAC -466 push
MIN-GSW +13 win

4-3-1

[ChuteBoxe]

Good luck tonight pico.

[pico]

10-9-2 overall

[walbjj]

ill be the first to say this. if anyone finds an algortihm to predict gambling of any sort, then they will be a wealthy person.
so far i have yet to see anything come remotely close. but good luck with it

[peterpan19]

pico,

as of now it seems like plus numbers do very well my friend
3-0 yesterday
2-0 monday
0-1-1 sunday
so 5-1-1 overall for plus numbers
not quiet sure what plus numbers mean, but I am very interested in systems like that
so if you want to share some infos with me pm me or maybe even with all of us post it hear (maybe it has potential)

thx for your work

[MoneySportsGuy]

goodluck to your system pico interesting how its going to go through the season as more data becomes available

[peterpan19]

what about the thursday and friday plays ??