Danshan on Kelly again

[danshan11]

if you are using your average edge are you not now flat betting ?
if avg edge is 5%
and you use 5% as edge in kelly calculator, you are now flat betting again right?

[Alfa1234]

No you are not flat betting. If you estimate your edge on 1 bet to be 5%, Kelly will give you a certain stake. If your edge estimate is too high you'll be overbetting that bet. Next bet your could estimate your edge to be 5% again but it could actually be 10%. You are now understaking. It will even out, you are not flat betting, your kelly adjusted stakes will be averaging out as in sometimes being too high and sometimes being too low.

[danshan11]

No you are not flat betting. If you estimate your edge on 1 bet to be 5%, Kelly will give you a certain stake. If your edge estimate is too high you'll be overbetting that bet. Next bet your could estimate your edge to be 5% again but it could actually be 10%. You are now understaking. It will even out, you are not flat betting, your kelly adjusted stakes will be averaging out as in sometimes being too high and sometimes being too low.

I dont get it sorry
if I use my avg edge over the last 20000000000 games of 5% on every kelly bet, am I not going to get the same result for every bet using kelly now?

[Alfa1234]

I dont get it sorry
if I use my avg edge over the last 20000000000 games of 5% on every kelly bet, am I not going to get the same result for every bet using kelly now?

Yes, you are going to get the same result. That is the point everyone is trying to make, isn't it?

[AsianmanSports]

if you are using your average edge are you not now flat betting ?
if avg edge is 5%

Yes, thats what i was trying to say.

I also choose my bet sizings conservatively for peace of mind and I wont be forced to lower my sizings during a downturn. Is it optimal? Theoreticlly not but people are comfortable with different kinds of risks and react differently to bad runs. I dont want to trade of my emotional well beeing to maximize profit.

Those are things that are important to consider when doing your sizings and everyone should decide for themselves.

[danshan11]

ok I am getting really confused, sorry guys my IQ is really low and I am very slow learner let me lay this out and tell me if I am retarded.

Flat betting
bankroll 10000 betting 1%
$100 per game every game no matter anything else

kelly betting
bankroll 10000 over last 100000000000 bets my edge has been 5%
I would bet $812.50 per game right?

so basically I am flat betting 812.50 right if I am using my average edge? or are we just adding 5% on to the current probability for that particular game

[AsianmanSports]

Except you would adjust your bet sizings constantly based on the developement of your bankroll which why in theory its more effective.

[danshan11]

so the key is not the edge, the key of kelly is the bankroll size, so if I adjusted my flat betting bankroll on every bet I would essentially do the same thing as kelly?

[Believe_EMT]

if your model is accurate over a big sample size the same principle applies.

again, this assumes a static market where bettors and books do nothing to adjust their own models.

outside of player props (due to low limits) are there any 'soft' markets left? i admire the spunk of individuals wanting to make their own model but these markets are so extremely efficient you are going to lose long term. if your model proves successful, by the time you have reached a statistically significant sample size the market has already adjusted.

the smartest thing i ever did was recognize how smart i am not. welcome to sports betting, leave your ego at the door.

[Alfa1234]

again, this assumes a static market where bettors and books do nothing to adjust their own models.

outside of player props (due to low limits) are there any 'soft' markets left? i admire the spunk of individuals wanting to make their own model but these markets are so extremely efficient you are going to lose long term. if your model proves successful, by the time you have reached a statistically significant sample size the market has already adjusted.

the smartest thing i ever did was recognize how smart i am not. welcome to sports betting, leave your ego at the door.

Very few markets besides the really huge, liquid ones are efficient, trust me on that. You said it yourself, markets with low limits aren't efficient. 99% of all markets have low limits. There are thousands of events to bet on every week and barely a few hudred have big limits and can be called efficient. Do not assume there is no money in low limit events, if you can bet a few hundred on each event that is mispriced, there is plenty to be made.

[danshan11]

Very few markets besides the really huge, liquid ones are efficient, trust me on that. You said it yourself, markets with low limits aren't efficient. 99% of all markets have low limits. There are thousands of events to bet on every week and barely a few hudred have big limits and can be called efficient. Do not assume there is no money in low limit events, if you can bet a few hundred on each event that is mispriced, there is plenty to be made.

if those markets are not efficient, how do you track your progress in those markets before you go broke?

[Alfa1234]

if those markets are not efficient, how do you track your progress in those markets before you go broke?

You track the amount of money you have. If it's more than before you started, you are doing well.

No seriously, you generally track the closing line. It's not efficient, but if you beat it by a certain margin overall you'll still be winning. It's just more "off" than the liquid games and sports, but that doesn't mean you won't win if you beat it by say 10% all the time.

[tsty]

I wonder how every other person tracks the effectiveness of their models lol

Maybe we need a market for the weather forecast...

[tsty]

How many games occur in every tennis match?

You dont always need thousands of matches when you can get thousands of games..thousands of serves..

Only retards think you need thousands of match results to know how effective your model is..

[AsianmanSports]

so the key is not the edge, the key of kelly is the bankroll size, so if I adjusted my flat betting bankroll on every bet I would essentially do the same thing as kelly?

That and adjusting your betsizings if you bet moneylines. You would bet more on a -150 fav and less on a 150 dog.

[danshan11]

yeah so you would take your edge and add that to the implied probability to get your probability for the Kelly calculation

[danshan11]

How many games occur in every tennis match?

You dont always need thousands of matches when you can get thousands of games..thousands of serves..

Only retards think you need thousands of match results to know how effective your model is..

how do you know how effective your model is?
and if your answer is net dollars than how do you know when the market has changed, when you are broke?

[danshan11]

read this tsty and let me know what you think?

https://www.pinnacle.com/en/betting-articles/Betting-Strategy/using-closing-line-to-test-betting-skill/7E6JWJM5YKEJUWKQ

[Glitch]

Nice thread. I like your style, Dan. Antagonistic but humble; I knew the responses would flow and i'm glad they did.

1) Most casual fans with or without access to a model they trust can give an answer (correct or incorrect) to the question "what is the percent chance this team/guy wins this game" or that "this guys scores over 13.5 points tonight" etc
a) The question then becomes whether or not each individual can have an informed and close enough to correct answer enough of the time, to be able to identify and quantify an estimate of an edge correctly, regularly.
a1) This edge would be ascertained by cross-referencing your predicted percentage likelihood of occurrence, to the implied percentage chance derived from the given odds. Beating the closing line would just be feeding confirmation bias that the line was, at the time you bet it, in need of correction. But this could be the case whether it's corrected or not.

2) Kelly betting provides 2 distinct advantages over not using it:
-preventing the bettor from regularly over-betting based on the edge which exists (if estimated closely enough to reality to consistently enough know where edges exist) because that would lead to ruin almost every time these patterns continue (whether the edge is above 0 or not.) Also the bet amounts are a portion of your bankroll, so they grow with you continuously and dynamically over time. If you are correct enough, the trajectory is up and it automatically keeps pace with your progress so you're not stagnating or having to change your flat bet size arbitrarily.
-And to bet bigger in situations where 1 or especially both of the following 2 things occur: there is a high likelihood of being correct on the prediction, and secondly- where your edge is larger (Where your adequately perceived or predicted percentage is so far off from, and above the implied percentage (which is based on the given odds) You must be able to take advantage of the really good opportunities to compensate for when you are wrong and/or unlucky. The law of large numbers says the universe will balance it all out as long as your projected percentages are close enough.

3) It's not about how many bets you win or what percentage of them you win, but whether you can identify market inefficiencies and properly quantify how inefficient they are. With kelly betting you bet nothing on a line you think is correct and you bet bigger the more the line seems off from reality (the actual chances as you perceive it.)

All you're trying to do is quantify the magnitude of how off a line is compared to what you think it should be if it reflected the percent chance you think it has. The better you are at this skill, the more money you make. That's all.

If it is not possible to have any idea of the probability of something to occur, then it doesn't make much sense. If it is possible to be able to measure the factors and variables associated with what may happen and quantify or parse it numerically into a prediction, then kelly betting may be the only way that makes sense (though as long as you do consider an odds' implied percentage to be too low compared to reality, flat betting could be fine too as long as you're not over-betting your edge above the fully kelly portion, you'd just be slowing your growth by not adjusting bet sizes to the perceived magnitude of opportunity.

I know i'd much rather be in the business of trying to predict/identify the chances of events to occur than that of trying to be correct 55+% of the time on what many people believe to be coin flips.

Please feel free to read this short article referencing my favorite book by my favorite author for further elaboration on the crux of the matter: https://acquirersmultiple.com/2018/0...e-that-counts/

[danshan11]

people dont know their edge and if you read that article it does not sound like the bettor in that article does either
he says he thinks he has a 2.5% edge but it looks like he has a 6% edge. I am not sure with that info that we can even establish confidently that the closing line is even correct.

[Glitch]

But you don't have to guess the closing line, you ONLY need to be able to semi-accurately and consistently assign a percentage liklihood of something to occur. Some teams are better than others and there are factors which influence this. If your model and/or brain can learn to play this guessing game correctly, while trying to guess the right thing (the percent chance of an event to occur) then kelly betting is what helps you avoid ruin and get the most bang for your buck with efficiency. If we can't project semi-accurately a percentage of something to occur or AT LEAST that the implied percent chance extracted from the odds is too low, then we shouldn't even be betting at all.

You can't know with absolutely certainty precisely what the edge is but this game is all about quantifying the magnitude of the discrepancy. And a ballpark is fine. But if sports are not predictable (for anyone other than the all-knowing divine linesmakers) then we should not be playing at all.

[danshan11]

see that is where we differ. you think the model is efficient and I think the closing line is.

I bet when the line has not gotten right yet and others bet when the line is different from their model.

I think the closest probability is in general the closed line and since we do not have a closed line when we bet, we technically dont know the probability

[Glitch]

Well i'm not stating that any specific model or that vegas/pinnacle etc. is nearly always correct. Here are my 2 pertinent assertions.
1) IF you can somehow accurately project the percentage on your own, without relying on basing the likelihood off what the linesmakers say it is, you can just compare your percentage to theirs and that is the edge by which you can apply some kelly or partial kelly bet sizing practice to in an attempt to maximize growth and avoid ruin.

2) The linesmakers are not always correct and that is the full game here. Because the ability to find out where your percentage chance is higher than theirs, is mathematically definitely the one thing that determines whether or not you have an edge against the odds offered, the only thing that is up for debate is whether or not this is possible. Can you consistently assign a close approximation of a binary bet's chance to win. If yes, some form or portion of kelly is the right way. If no, there is no right way but you can get lucky enough if you somehow mostly bet on games where the percent was higher than purported by the odds. Because over time, that is the only way you can turn a profit.

I see your other questions about beating the closing line by an average of 2% but turning a 6% profit. The fact that 2% is positive kinda indicates that more often than not, you are finding the inefficiencies. Some combination of either or both beating the closing line the majority of the time or by a large amount. You seem fixated on the closing line, and you're not the only one but I believe that is an error and something closer to correlation than causation. If you are correctly betting the right stuff (where the linesmakers are wrong in your opinion) then at least some of those times, the line will change to become less generous and you will have beaten it. The more wrong they were, the bigger chance you have of having beaten the closing line when you look back because that would render them more likely to change the line after you and others bet it.

The only mystery i see in this 2% vs 6% question is why it would matter how much you beat the closing line by if you're not bet sizing based on the perceived edge and if the closing line changing isn't indicative of their thinking the previous offering was too generous and correlating to the size of the edge. ie if you're hypothetically really actually finding some games where your percent chance is 20% higher than the implied percentage converted from the odds offered, then of course at least some of those times the line will move to become less generous by closing once and if adjusted.

How much more correct you are than the bookie's closing line does not matter much unless you are betting your belief according to how much better you think your number is in the qualifying opportunities. (bet sizing according to the magnitude of discrepancy)

[danshan11]

that is where I think overall the closing line is efficient. the closing line is the most efficient method we currently have to estimate probability. I think there are very few people who have a model with higher overall efficiency than the closing line. We know the closing line is a ton of things and pretty much factors in all important factors and is a wisdom of the crowd consensus and adjusted for proven tested winners. I come back again that it would be hard pressed to find someone who can model better than that. Yes in some instances the line cannot overcome volume or publicity of an event ie super bowls but in general there is nothing currently more efficient

[danshan11]

the 2% 6% thing is an issue for me because, if the line is so perfect and this person is beating it at 6% long term, that is for what reason?

bettor has lots of bets, he beats the line for an expected ROI at 2.5% but is actually winning at 6%, how can that be, what is your explanation for this. I know you dont have the specifics but generally what is your opinion?

[Glitch]

But that's the problem. I agree that beating the closing line is a good indicator of your doing the right things. But you can't know the closing line until it closes. And you can't bet something better than it if you wait for the closing line. Beating the closing line is good because if the book finds an inefficiency, or indicators that there may be one, they can and sometimes will adjust the line to be less generous. So you really wanna find the generous lines. That will correlate with beating the closing line. If that is not possible, we should not be betting heavily at all or trying anything besides some light recreation for the fun and action.

If you wanna become some kinda casanova pickup artist or get more action with the ladies or whatever, your course of action can't be bringing more women home. That is in a way the goal, because it indicates and correlates strongly with a higher rate of success with women. (as do other things like going home with her, getting phone numbers, taking her to a hotel or even just getting them to laugh and interact with you all night). But your goal or course of action to getting more women can't be "get more women to go home with me." Sure you want to do that, and sure that'll happen if you do the right and necessary things but that doesn't really help you with a strategy or plan to execute.

[danshan11]

my goal is not to beat the line, my goal is to actually have an edge. the only way I can have an edge is to beat the line, imho

[danshan11]

the 2% 6% thing is an issue for me because, if the line is so perfect and this person is beating it at 6% long term, that is for what reason?

bettor has lots of bets, he beats the line for an expected ROI at 2.5% but is actually winning at 6%, how can that be, what is your explanation for this. I know you dont have the specifics but generally what is your opinion?

what is your opinion on this? I am really curious

https://www.pinnacle.com/en/betting-articles/Betting-Strategy/using-closing-line-to-test-betting-skill/7E6JWJM5YKEJUWKQ

[Glitch]

my goal is not to beat the line, my goal is to actually have an edge. the only way I can have an edge is to beat the line, imho

How often you beat the closing line and by how much you do it means almost nothing except a crude indicator of whether you are or are not generally successful. (flat betting) While everyone knows winning more, will usually win you more, It is just a good thing to look back on to see whether or not you've been getting your money in good.

But if you are trying to beat the closing line, you are basically trying to bet a line that you anticipate getting worse. If you believe lines get "worse" for reasons other than the "they only wanna balance the action" myth, then the way you can beat the closing line more often is to try to guess where the line is off or wrong or less accurate than what your process thinks. However, you still wouldnt know until afterwards whether or not you beat it, you would still hope you beat it or even try to beat it by trying to find where and when the linesmakers are a bit wrong. Why beating the closing line is so important is because the idea is that it indicates your having located the inefficiency. But the process in application should be trying to figure out how to find the efficiency. and you can do that by having your own prediction and comparing it to theirs (which is implied by the odds).

You don't have an edge because you beat the line, you beat the line because you have an edge.
You have an edge because you find where you think they are underselling a probability; A likelihood or chance. You find where the odds are wrong or inefficient and you bet only on those games. Ideally sizing appropriately per the formulate that tells you how to do such a thing correctly. But flat betting would be fine too as long as you're not over-betting and only (mostly) betting when you have an edge

I skimmed that article but need to know more about his bet sizing and odds he was taking to have an opinion. I could not find those quickly. What i do know is that it doesnt matter how much you beat the line by if you're not adjusting your bet size based on your edge. With beating the closing line also being a product of having had these edges. He's betting the inefficiencies or getting his money in good or beating the line. Great. That's the only thing I would expect that to tell me, that he's winning some. I don't see how or why anyone would expect it to be 2% or whatever exact amount he beat them by.

[danshan11]

is a modelers probability more accurate than the closing lines probability?

[danshan11]

he was flat betting and the avg line was +101

[Glitch]

he was flat betting and the avg line was +101

Ok thanks. It sounds like some underdogs were pulling some weight for some other bets but, beating the lines more often than not (+2%) did indicate or support that he was successful (as the 6% actual would indicate or exemplify.) To me it's just not suppose to be directly proportional or exactly the same. He was right a bit so he won a bit. Even if he was using something that was suppose to correlate with the size of the edge more like kelly, you'll never know what the real percent chance was of your 63% prediction you put against their 50% +100. Similarly, if that line moves to -110 or -120 later, the only things that affected your actual profits are how much you bet and what the odds were when you bet them and whether you won or lost. it's nice that he beat it, and maybe some of the factors that helped him beat it, were some of the factors that helped him win that particular bet but the amount by which he beat it had no direct affect on anything.

is a modelers probability more accurate than the closing lines probability?

Not necessarily but that tool or system, or whatever that he is using to ascertain an opinion on the event which he is trying to predict and wager on, is the only thing he can use. Whether it's from him or his brain or tools or someone elses'. Some models will be good or good for a while, some will never be good, some will be lucky. But whatever you have, to be able to predict a percentage of likelihood accurately enough, that is all you can use. If you can pick winners with your brain, then a better way to look at it than "this teams gonna lose or this team's gonna win" is to try to practice getting to this team has a 59% chance to win. Then you compare that to the implied percentage chance of the odds. If yours is higher then you think you have an edge there. It is unfortunate that we can't know an exact percentage likelihood and that some of us have really bad sports opinions and prediction skills. But for those who think they have a way of getting to an approximate percentage likelihood of occurrence, that is how and when a bettor could find something like an edge.

I just think it is incredibly vital to have your own prediction on the game and that when you have one you can quantify or format as a percentage, then if it's accurate enough to reality, it will come out in your favor if you flat bet small enough and mostly on where you have an edge, or especially if you bet some portion of kelly according to the size of your perceived edge. If you're crap at predicting sports outcomes then of course you're gonna lose like with every other method. But every model or system or brain is different.

[danshan11]

you dont think it is necessary for a modeler to be more efficient than the closing line in order to win long term?

[Glitch]

you dont think it is necessary for a modeler to be more efficient than the closing line in order to win long term?

I think they would have to have a better projected percentage likelihood than the one implied by the closing line very often. And I also think that would result in their having a more favorable line than the closing one quite often.

[Glitch]

That is to say, they would have to project a higher percentage likelihood than the one implied by the closing line often. And that this would result in the bettor having bet a better line than the closing one very often as well (if they are accurate enough often enough and thus successful.)