Question for LTprofits & other model-based cappers

[Miz]

by the way, don't start modeling with baseball. there is too much data manipulation to do, park factors to adjust for, weather and other things. pick a small market sport like AAF or something else first. If you can beat small market openers then you can move on to harder challenges.

[pavyracer]

You don't need modeling to win. Use the hours wasted on modeling stats to watching film of games, try to understand why teams win without looking at betting lines, study coaches, players, team tactics and then you can pick winners easily whether they are chalk or dogs.

That's how I made my first $100,000 in sports gambling.

[Miz]

bonus tip.... before you use any type of regression or neural network to relate input to output you may need to adjust the data first (aka transform it first).

[Larkman]

You don't need modeling to win. Use the hours wasted on modeling stats to watching film of games, try to understand why teams win without looking at betting lines, study coaches, players, team tactics and then you can pick winners easily whether they are chalk or dogs.

That's how I made my first $100,000 in sports gambling.

You don't *need* modelling to be profitable, but it does make betting much more consistant, repeatable, quantifiable, and automatable, which will save a lot of time and effort in the long run.

[Miz]

if you bet with no model, you better have excellent information ahead of the rest of the market, which is unlikely. Information might make me over-ride a model bet and make it a no play, but no way i bet without a model to quantify the edge.

you can make money chasing steam with no model, but you'll get the boot.

pavy, do you keep records at all? if so, i'd be curious to know your win pct using your methods. Do you vary your staking at all (i.e. kelly variants) I am not saying you can't win without a model... but that just isn't how I do it.

[danshan11]

my same question I keep asking it over and over. how do you know what to bet if you dont create a fair line for the game?

[tsty]

You can create it in your head..

[u21c3f6]

my same question I keep asking it over and over. how do you know what to bet if you dont create a fair line for the game?

You posted a link in a previous thread on football (soccer) wagering. Did you read it? If not, read it and you will find out one way of doing that. If you did read it, what don't you understand?

Joe.

[danshan11]

You can create it in your head..

as long as you have a fair line and if you can create it in your head that is cool!

[danshan11]

You posted a link in a previous thread on football (soccer) wagering. Did you read it? If not, read it and you will find out one way of doing that. If you did read it, what don't you understand?

Joe.

Hey Joe, that was more of a system play not individual lines of an independent game. I dont like system plays, I prefer fair line play against it.

[u21c3f6]

Hey Joe, that was more of a system play not individual lines of an independent game. I dont like system plays, I prefer fair line play against it.

But it answers your question. Whether you like it or agree with it is irrelevant. The focus is in the concept, not the methodology.

Joe.

[danshan11]

But it answers your question. Whether you like it or agree with it is irrelevant. The focus is in the concept, not the methodology.

Joe.

I am talking about people picking games randomly because they see value without actually having a line they created

[nash13]

I am talking about people picking games randomly because they see value without actually having a line they created

First of all we have to be clear that we are not betting vs the bookies here, we are trying to beat the market and esp. doing this on timing.
A successful bettor know WHEN to BET and on what PRICE and which AMOUNT.
There is no reason to believe that we can outsmart the lines, but we can outsmart the market.

[danshan11]

First of all we have to be clear that we are not betting vs the bookies here, we are trying to beat the market and esp. doing this on timing.
A successful bettor know WHEN to BET and on what PRICE and which AMOUNT.
There is no reason to believe that we can outsmart the lines, but we can outsmart the market.

why do we need to be clear on that, I dont understand the importance of that, can you explain why it matters?

how does a successful bettor know when to bet, what price and which amount without having a fair line?

[nash13]

why do we need to be clear on that, I dont understand the importance of that, can you explain why it matters?

how does a successful bettor know when to bet, what price and which amount without having a fair line?

Simple reasoning:
The so called fair line helps not very much, if you can't forecast what the market will do.
Let's take a hypothetical example.
NBA:
Dallas vs Portland
Opening Line: Portland -6

a) Your fair line tells you that this is too low and Portland should be favored by 9. So you take Portland -9 and buy points. But the market moves in the opp way and the closing line moves further away from your fair line and ends up at Portland -4. So your fair line told you at the start that it would be a smart move to take Portland -9. But the market created a bigger value when the line moved. So what is the choice here? I know a lot of so called services who fabricate their profit by moving the lines to the best odd available. Which leads us to the second problem in the equation.

b) You have a good predictive system or are good with bets, but the amount you bet leaves a lot of potential profit on the table.
If you can analyze the value better than others, you should be able to maximize your profits by betting the right amount.
This can have two reasons: first your bookmaker who has the best odds limits your bets and so the potential profits are erased. And secondly the bookies where you can bet bigger amounts don't have the lines to yield you enough profit.
If you believe in your system you shold value the profit.
Betting 1 Dollar on a system which yields 5% or betting 10 Dollar on the same leads to 10 times more profit. Evaluating the risk and timing of the bet is as important as analyzing the bet itself.

[danshan11]

the fair line is a forecast of what the market will do. if your fair line is not accurate overall its not a fair line. the goal of the fair line is to get as close as possible to the closing line since we by consensus agree is the most accurate line overall.

if I model -6 and it closes -9 without any injury, weather or other significant factors, my fair line was not fair.

still leaves the question of how do you know to pick a spot if you dont have a fair line.

say Kentucky -5 -110 is current and for you that is a good bet, would you do -5 -140 or -5 -190 or -5 -380, I mean i dont understand how you can derive value without having a fair value.

if I want to buy Dinkledonks on ebay and resell them. I can't know how much to pay for them without knowing what they are actually worth, what is the difference here, Kentucky -5 -110 is that a good deal (value) or is it not and the only way I can know if it is a value bet, is to know what it is actually worth.

How else can you determine value?

if the argument is pick your spots and price does not matter, you could literally bet every game
say you like Kentucky -5 but its -10 just buy points if that is the problem. even though the points cost you big and now you are minus -450 on the bet, who cares if that is a winner to you and you supposedly win on the spots you pick.

[nash13]

the fair line is a forecast of what the market will do. if your fair line is not accurate overall its not a fair line. the goal of the fair line is to get as close as possible to the closing line since we by consensus agree is the most accurate line overall.

if I model -6 and it closes -9 without any injury, weather or other significant factors, my fair line was not fair.

still leaves the question of how do you know to pick a spot if you dont have a fair line.

say Kentucky -5 -110 is current and for you that is a good bet, would you do -5 -140 or -5 -190 or -5 -380, I mean i dont understand how you can derive value without having a fair value.

if I want to buy Dinkledonks on ebay and resell them. I can't know how much to pay for them without knowing what they are actually worth, what is the difference here, Kentucky -5 -110 is that a good deal (value) or is it not and the only way I can know if it is a value bet, is to know what it is actually worth.

How else can you determine value?

The Fair Line getting as close to the closing line still leads to the question what happens in the process.
We se a lot of games which circulate around lines all the time. Catching Kentucky -5 at -110 may create value while the closing line will end at -160, but if you had the chance to take it at +100 because of fluctuation, you will leave money on the table.

[jjgold]

I have gotten requests for LT's MODELS

THEY ARE NOT FOR SALE!!!!!!!!!!!!!!

STOP!!!!!!!!

[danshan11]

this is true but still how can you make a play without an accurate fair line to determine if there is value?

[nash13]

this is true but still how can you make a play without an accurate fair line to determine if there is value?

we are on the same page here. i agree that creating a fair line is the first step to making a bet. but the market and lines move because the bookies adjust to the money, they want the best outcome no matter how the game ends. the best move for a good bookie is to balance the money and create their own "sure bets".

[danshan11]

agree but most books dont get even money ever. what they usually do is end up crushing the line and in the long run that is a winner. books adjust to avoid over risk on one side but being on one side with juice is a good place to be and where they are 9 out of 10 times

[danshan11]

its called an api or scrape

[Miz]

i create fair lines. that is the first step in any model. the bet amount varies based on the fair line vs posted line difference. your model should increase in success the larger that difference is. If it does not, then the model is probably not functional.

[JacketFan81]

i create fair lines. that is the first step in any model. the bet amount varies based on the fair line vs posted line difference. your model should increase in success the larger that difference is. If it does not, then the model is probably not functional.

If there is a sizeable difference between the fair line and the posted line, I know 100% the model is not functional and is just a semi-random number generator.

[danshan11]

exactly, our goal as modelers is to get as close as possible to closing lines to know our model is efficient

[HeeeHAWWWW]

exactly, our goal as modelers is to get as close as possible to closing lines to know our model is efficient

My goal is to be as accurate as possible, in subsets of the market I can beat by more than the juice.

Closing lines aren't really relevant to that.

[vinnykal2323]

I would like to see Danshan weigh in on this thread

[danshan11]

I would like to see Danshan weigh in on this thread

what question or questions do you have for me, I would be glad to chime in?

[Miz]

If there is a sizeable difference between the fair line and the posted line, I know 100% the model is not functional and is just a semi-random number generator.

I always have an upper limit filter enabled. Once the delta is too big it is likely missing something. Good models have outlier predictions that can be filtered out.

but yes, if the model is way off on average then it is spitting out trash

[vinnykal2323]

what question or questions do you have for me, I would be glad to chime in?

I would just like to see some more input from you. I believe you can explain more

[HeeeHAWWWW]

I always have an upper limit filter enabled. Once the delta is too big it is likely missing something. Good models have outlier predictions that can be filtered out.

It makes sense - there will always be the occasional major factor it's impossible to know, and/or assess properly.

One approach if you have a very long betting history: plot estimated edge against yield. In the ideal world it'd just be a linear slope. In reality it will increase less than linearly, slowing as the edge increases - the larger the edge you predict, the more it's likely to be an error. Eventually it'll stop increasing, and fall away to zero and below.

One approach is to use this relationship as a mask to adjust estimated edge. I do this too, although not beyond the peak of the curve.

[danshan11]

My goal is to be as accurate as possible, in subsets of the market I can beat by more than the juice.

Closing lines aren't really relevant to that.

when you say accurate as possible, accurate to what?

[tsty]

agree but most books dont get even money ever. what they usually do is end up crushing the line and in the long run that is a winner. books adjust to avoid over risk on one side but being on one side with juice is a good place to be and where they are 9 out of 10 times

No true at all or they risk going bust

[danshan11]

can you explain how they go broke when they have hundreds upon hundreds or thousands of different bets a day at a juiced line?

[Miz]

It makes sense - there will always be the occasional major factor it's impossible to know, and/or assess properly.

One approach if you have a very long betting history: plot estimated edge against yield. In the ideal world it'd just be a linear slope. In reality it will increase less than linearly, slowing as the edge increases - the larger the edge you predict, the more it's likely to be an error. Eventually it'll stop increasing, and fall away to zero and below.

One approach is to use this relationship as a mask to adjust estimated edge. I do this too, although not beyond the peak of the curve.

Interesting thoughts. Thank you for sharing them.