A way of evaluating predictive models reasonableness

[tsty]

you did not answer any of my questions

because your questions are stupid and irrelevant

this whole discussion was about making a model based on past results

then you chimed about odds once again

do you even know how they make the odds in the first place?

[danshan11]

no I dont can you please tell me how they make the odds?

[Charlies_Hustle2]

because your questions are stupid and irrelevant

this whole discussion was about making a model based on past results

then you chimed about odds once again

do you even know how they make the odds in the first place?

This guy is an obvious troll, I called him out in the other thread and his response confirmed it. Just put him on ignore

[danshan11]

called me out, I dont know what that means but you are like so many before, it is because i say so. That mentality does not teach any of us anything.

[danshan11]

no I dont can you please tell me how they make the odds?

how tsty?

[Charlies_Hustle2]

called me out, I dont know what that means but you are like so many before, it is because i say so. That mentality does not teach any of us anything.

Oh I'll teach you a lot. Just commit to the deal I have pending!

[danshan11]

commit to what stop asking questions because you are gonna show me a link of someone who wins and we dont know how he wins just that he does, DRH kills the line and wins all the time, he openly says his method is to beat the line. I dont really know what you are saying, honestly

[tsty]

no I dont can you please tell me how they make the odds?

From past results? R u acting dopey on purpose?

[danshan11]

generally sure, they look at past performance and design power rankings from those past results and they ALSO use past lines, my example on series shows that past games have a very small influence on the tomorrow lines.

warriors at home -9.5 they win that game today by 30 and tomorrow guess what they are home against the same team with same players with both teams on the same rest and magic line is -9.5 again. so you tell me do the lines come from past results or more from past lines. do you really believe we dont know the closing lines or the really closing line is one of your insider secrets that pinnacle shares with you only

[danshan11]

Opening lines are made from power rankings and any data on weather, injuries and a dash of salt from what the public has done in this spot in the past to get a jump on the juice. Yesterdays games results have very little impact on tomorrows games, look at the teams playing tonight that play tomorrow those lines will not move because they won big or lost big tonight and if they do it will be very slightly and probably will snap back once the sharp money starts rolling in to flatten out the line.

Closing lines are the size dam necessary to direct the water where they need it and they are usually extremely sharp because they have the latest and most accurate information just seconds prior to the game start.

[tsty]

So you think opening lines are made from power rankings? Lol

When sharps close the line what do you think they use to make their model?

Have you even made your own model before?

[tsty]

I dont think you realise how far you are from even the most basic shit

[Charlies_Hustle2]

I dont think you realise how far you are from even the most basic shit

It's quite obvious that he's just trolling, doing his best to turn the think tank in to player's talk. Just ignore him.

[danshan11]

So you think opening lines are made from power rankings? Lol

When sharps close the line what do you think they use to make their model?

Have you even made your own model before?

answer my questions, you just keep asking questions

everyone technically models differently
yes I have my own model on baseball and basketball, although my baseball model comes up slightly short, my basketball model is very good.

[danshan11]

I dont think you realise how far you are from even the most basic shit

help me then, answer my questions and help me.

I think openers come from models that use power rankings adjusted for injury, weather and a small tiny dash of sentiment on certain scenarios

here are my NBA power rankings currently

Team	SAG	SAG	MY PR	BET/SAG	Combined
Rockets	92.97	-2.97	-7.79	-4.84	-5.82
Warriors	95.94	-5.94	-9.07	-7.22	-7.84
Raptors	96.76	-6.76	-7.63	-6.22	-6.69
Jazz	92.36	-2.36	-5.58	-2.92	-3.81
Trailblazers	92.05	-2.05	-3.68	-2.33	-2.78
Nuggets	95.06	-5.06	-4.87	-4.08	-4.34
Celtics	95.16	-5.16	-5.26	-6.88	-6.34
Thunder	94.33	-4.33	-6.73	-6.35	-6.48
Seventysixers	92.16	-2.16	-5.40	-4.05	-4.50
Cavaliers	83.37	6.63	5.21	6.97	6.39
Bucks	95.83	-5.83	-6.80	-6.82	-6.81
Wizards	87.49	2.51	0.09	0.62	0.44
Spurs	91.07	-1.07	-3.34	-1.17	-1.89
Clippers	90.24	-0.24	-3.44	-1.40	-2.08
Heat	88.95	1.05	0.90	0.58	0.69
Timberwolves	89.93	0.07	-3.92	-1.38	-2.23
Pacers	93.53	-3.53	-4.29	-3.79	-3.95
Pelicans	91.45	-1.45	-3.66	-1.68	-2.34
Hornets	89.05	0.95	-3.67	-1.84	-2.45
Lakers	92.19	-2.19	-4.09	-3.53	-3.72
Pistons	88.88	1.12	-1.03	2.42	1.27
Mavericks	89.12	0.88	-1.96	-0.38	-0.91
Nets	87.62	2.38	0.50	1.22	0.98
Kings	87.67	2.33	0.37	1.63	1.21
Magic	86.38	3.62	0.80	2.38	1.86
Bulls	82.05	7.95	3.87	7.00	5.95
Knicks	83.58	6.42	4.52	5.61	5.25
Hawks	81.42	8.58	4.40	7.70	6.60
Suns	84.07	5.93	5.56	6.61	6.26
Grizzlies	89.3	0.70	-1.22	1.02	0.28

My PR column is my power rankings and they are what I use to model games! you can add roughly 3 for home team and you will see how close my rankings are!

here are my totals rankings its a plus minus system which honestly right now has been working rough but historically has done real well

Warriors	6.73
Rockets	1.11
Raptors	-0.14
Celtics	-3.41
Spurs	1.61
Cavaliers	-6.78
Thunder	2.67
Wizards	4.37
Timberwolves	0.26
Trailblazers	-2.05
Jazz	-4.86
Bucks	5.88
Nuggets	-5.47
Pistons	-3.22
Pacers	-7.78
Seventysixers	4.80
Pelicans	8.86
Heat	-5.45
Clippers	4.17
Hornets	2.82
Knicks	1.98
Mavericks	-1.65
Lakers	5.04
Grizzlies	-13.83
Magic	-5.53
Nets	-0.07
Bulls	-5.69
Kings	9.82
Suns	-3.06
Hawks	7.61

to use my totals its pretty easy just take scoring average and adjust for the scoring avg example
Rockets 1.11
Heat -5.45
so league total avg is 220.6 subtract -4.34 and you get 216.26 and then you round so 216.5 and then adjust for injuries and as you can see Heat have lots of key injuries so it will drop accordingly, I aim to be within 3 points of the total and 2 points on the sides
I average on the season 2.49 off on totals and 1.37 on sides

I also dont just use power rankings to get a line, I adjust using a comparable analysis of the past 4 opponents for each team in the game and add that to the power ranking numbers to get a line

[danshan11]

my way of doing it is definitely different but that is because my work background is in comparable valuation not data science but apples oranges really!

[danshan11]

also here are my lines for 12-21 tomorrows NBA games, I have not adjusted for injuries yet, what I do is I wait for lines to come out with decent limits and then if I am off a ton on a game before I bet it I check and adjust for injuries on that game before I bet!

12/21		Pistons	219.5	221.5
12/21		Hornets	-4.5	-5
12/21		Cavaliers	211	205
12/21		Raptors	-13.5	-15
12/21		Pacers	-2	-2
12/21		Nets	212	207.5
12/21		Hawks	230.5	235
12/21		Knicks	-1.5	-1
12/21		Bucks	222.5	224
12/21		Celtics	-1	-1
12/21		Magic	-2	-2
12/21		Bulls	205	202.5
12/21		Timberwolves	222	225.5
12/21		Spurs	-2.5	-2
12/21		Grizzlies	215.5	213
12/21		Kings	-0.5	0
12/21		Jazz	212	213.5
12/21		Trailblazers	-1	0
12/21		Pelicans	231	231.5
12/21		Lakers	-2.5	-3

first is my actual mode numbers, second is a "check" I made which uses past lines to get an estimated line.

[Waterstpub87]

What does "comparable valuation" mean? Is that like real estate?

[danshan11]

my injury adjustments are done by the following
player team value
add up value of all active players on team
then adjust for the opponent
example Lebron is worth 6 as a laker now if he played for Warriors he would be worth 4.5 to the side
Booker is worth 6 to the Suns but if he was a Warrior he would be worth 2.5 to 3

[danshan11]

What does "comparable valuation" mean? Is that like real estate?

for assets, usually yachts, businesses, FFE, some real estate, I mainly worked for courts as an expert witness for suits and criminal cases of seizure. I even did a Ferris Wheel Comparable market analysis one time for bankrupt fair group. Its not easy to get the value of unique items like a handmade unique size Ferris wheel

[dpark80]

Hey Dan,

How long have you been using your power rankings to bet on NBA games? How many units have you been up each of the past seasons? It's been rough for you so far this year. Are you looking to tweak your rankings or do you think it will revert back as the season progresses?

help me then, answer my questions and help me.

I think openers come from models that use power rankings adjusted for injury, weather and a small tiny dash of sentiment on certain scenarios

here are my NBA power rankings currently
My PR column is my power rankings and they are what I use to model games! you can add roughly 3 for home team and you will see how close my rankings are!

here are my totals rankings its a plus minus system which honestly right now has been working rough but historically has done real well
to use my totals its pretty easy just take scoring average and adjust for the scoring avg example
Rockets 1.11
Heat -5.45
so league total avg is 220.6 subtract -4.34 and you get 216.26 and then you round so 216.5 and then adjust for injuries and as you can see Heat have lots of key injuries so it will drop accordingly, I aim to be within 3 points of the total and 2 points on the sides
I average on the season 2.49 off on totals and 1.37 on sides

I also dont just use power rankings to get a line, I adjust using a comparable analysis of the past 4 opponents for each team in the game and add that to the power ranking numbers to get a line

[danshan11]

I have 1 NBA season last year I won and my line value was almost 2.2% subtract 1.1 for margin made 1%ish for 2017
this season I have been getting killed on NBA my totals seem off, but the NBA totals seem crazy to me no support of this in data just seem crazy all over the place with big swings that implies to me I am not the only one strruggling with totals this year. I have one WNBA season using the same model and it killed it I had nearly 3% above margin CLV and my record was nuts like a 61% win rate but that is just luck of the draw but the model is very very very solid dude. I know people think I am an idiot cause I dont just agree with senior players here but I am far from a rookie in comparative analysis and that is all this is. Look at my lines and see once BOL or the closers and you will see I know what I am doing. I always ask any of the touts or smart guys to give me lines on games before they come out if they cant, I usually call bullshit on them and rarely put any weight on them. There are 1000s of guys like charlie 2 shoes or whatever that win because they create 200 accounts and once one starts winning, they start touting that one, its complete scam BS, if you know this game you can tell me very accurately what the line will be on tomorrow's NBA games for most games, of course injuries change that but we all understand that. I will win in the NBA. I started out like 5-20 but since then been at 52% or something about where it should be honestly

[danshan11]

so far this year in the NBA

W	L	P	Win %	CLV	CLV +	CLV -
44	56	2	44.00%	1.01%	40	26
21	19	0	52.50%	0.96%	18	9
23	37	2	38.33%	1.05%	22	17

total
sides
totals

[Waterstpub87]

for assets, usually yachts, businesses, FFE, some real estate, I mainly worked for courts as an expert witness for suits and criminal cases of seizure. I even did a Ferris Wheel Comparable market analysis one time for bankrupt fair group. Its not easy to get the value of unique items like a handmade unique size Ferris wheel

I guess I know who to ask for advice if I want to buy a roller coaster or a yacht.

[danshan11]

yes anytime and it surely applies to sports as well!

[danshan11]

I guess I know who to ask for advice if I want to buy a roller coaster or a yacht.

what is your expertise in?

[tsty]

answer my questions, you just keep asking questions

everyone technically models differently
yes I have my own model on baseball and basketball, although my baseball model comes up slightly short, my basketball model is very good.

This is factually incorrect

There is only one way to do it

The difference maker are the variables and the co efficient you give to each one

[danshan11]

This is factually incorrect

There is only one way to do it

The difference maker are the variables and the co efficient you give to each one

I thought you set the bar pretty high for dumb ass shit when you said we dont get the closing line and now you jumped way over that bar with this one. You think that everyone models the same, LMAO, come on dude, is that really what you are saying?

and why wont you answer the questions I asked you?

[danshan11]

@tsty
how is the line created?
can we get the actual closing line?
can you win by only beating the closing line?

also @tsty are you the guy who makes the line and hides the closing line, are you "they" and tsty is just a misspelling of they?

[danshan11]

This is factually incorrect

There is only one way to do it

The difference maker are the variables and the co efficient you give to each one

EVERYONE models the SAME the only difference is what they test and how much weight they give each test
Come on dude, do you see the bad parts of this argument.

yes and we all know power rankings are not the sole basis for linemaking, sentiment goes in and some math some little algos and other stuff, but essentially its PR's with some salt and that salt can be sentiment, betting history, and a few other little things. so if that is your secret sauce, its not so secret

[Waterstpub87]

what is your expertise in?

Work in finance. Originally as a research analyst with a financial software company focused on quantitative analysis. Recently moved to a data provider.

[danshan11]

sounds like a perfect fit for sports data science and creating models!

[danshan11]

waterspub
what do you think of my approach? pros and cons, just your thoughts of course!

[tsty]

That is not what i said

Re read my post

Linear regession is the only thing that works

[danshan11]

That is not what i said

Re read my post

Linear regession is the only thing that works

LOL, you are drunk, that is like saying we are all the same because we drink water. we are all math teachers cause we know 1+1

you are tsty we all model the same cause we all use some form of numbers, so yes you are correct but even in reality it could be possible to model without numbers