- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
Posted on 8/21/15 at 11:43 am to moneyg
quote:
That's a really bad argument.
Your argument should be that historically, when looking at the data of betting lines vs. actual outcomes...the win % is your chart.
Is that not where you derived the chart you are using?
Can you describe where you got the chart of winning % vs. betting lines? Is it not based on historical data?
Posted on 8/21/15 at 11:44 am to moneyg
quote:No. That's a really good argument. Smart money tends to move the line to the most accurate position. Thus, a betting line is a good indicator of the potential result.quote:That's a really bad argument.
Betting lines have everything to do with money being bet on games. Most money is smart money. Thus, betting lines are some of the best indicators of the actual ability of a team.
quote:That is a DIFFERENT argument than the quoted statement.
Your argument should be that historically, when looking at the data of betting lines vs. actual outcomes...the win % is your chart.
Is that not where you derived the chart you are using?
Your proposed argument merely notes the correlation between the amount of the betting line and the percentage chance of victory.
However, the quoted argument explains WHY one would expect to find a high correlation between the amount of the betting line and the percentage chance of victory.
In short, your arguments expresses the correlation, the quoted argument explains why the correlation exists.
Posted on 8/21/15 at 11:46 am to Salviati
quote:
That is a DIFFERENT argument than the quoted statement.
Your proposed argument merely notes the correlation between the amount of the betting line and the percentage chance of victory.
However, the quoted argument explains WHY one would expect to find a high correlation between the amount of the betting line and the percentage chance of victory.
In short, your arguments expresses the correlation, the quoted argument explains why the correlation exists.
That's a whole lot of discussion that ignores the only thing that matters...is your chart based on actual data or not?
Posted on 8/21/15 at 11:47 am to jonboy
quote:No.
Without ties, win's and losses are absolutes. Shouldn't anything above 50% be counted as a predictive "win" individually, instead of cumulative?
If you take a bent coin that has a 52% chance of landing on heads and flip it 100 times, your analysis would count all flips as a predictive "win" individually.
Thus, your analysis would predict the coin landing on heads 100 times.
However, the more likely result would be the coin landing on heads 52 times.
Posted on 8/21/15 at 11:49 am to Salviati
quote:
Thus, a prediction based upon betting lines would also change over time viewed from the day of the prediction.
Doesn't that basically render a pre-season prediction based on betting lines useless? LSU will be 9-3 based off of preseason betting lines, but the lines will shift, sometimes drastically, after each game.
Posted on 8/21/15 at 11:55 am to Guava Jelly
quote::smh:
his is EXACTLY what you're doing when you average the odds of success.
I am NOT averaging the odds of success.
quote::smh:
Again, I'll try to explain it. LSU's odds of success vs McNeese according to this are 100%. On a later event, the Bama game, LSU's odds of success are 30%.
When viewed as individual events, as they should be, LSU would win vs McNeese and likely lose vs Bama. But, when you average the odds, this changes. Your odds of success among all events is 65%.
This is flawed because your odds of success vs McNeese have no bearing on your odds of success vs Bama.
quote:It is NOT an average. It is the cumulative total of the odds.quote:What do you think this is?
That works out to 8.6 wins for the season; thus, the lines suggest a 9-3 season, perhaps an 8-4 season.*
Now, the funny thing about averages is that if you multiply the average times the total, you will come out with the same answer. However, in your analysis of averaging the odds, YOU are the one who is attempting to conflate all 12 games. I am not. You will get to the same result, but your analysis incorrectly conflates all of the games.
PLEASE, do me ONE favor.
Please tell me HOW you would do it, and PROVIDE us all with YOUR result.
HINT: You will find that I am right.
Posted on 8/21/15 at 11:57 am to Salviati
damn it I'm confusing myself, need the season to get here.
This post was edited on 8/21/15 at 12:01 pm
Posted on 8/21/15 at 11:57 am to Dan Bilzerian
quote:Does P mean Positive?quote:The line for LSU/Ole Miss as a P.
Ole Miss. . . . . . . . . . 4
Are you trying to tell me that Ole Miss is favored?
I know that. Yes, Ole Miss is favored against LSU.
That is why the OP states that LSU has only a 45% chance of winning that game.
Posted on 8/21/15 at 12:05 pm to atltiger6487
quote:quote:I did. 9-3. Took me about 20 seconds after considering our roster and schedule.quote:Uhhh . . . which games do you suggest that I watch right now.
way too much time on your hands. Let's just watch the actual games. Then we'll know.
So I guess you didn't participate in any season prediction threads?
So YOU spent time predicting the season, yet you go to the time and effort of posting in this thread MULTIPLE times to tell me NOT to predict the season.
quote:
All your mathematical analysis, while somewhat impressive, just doesn't mean squat.
I'm sorry that you didn't know how to do the analysis in the OP.
I'm sorry that you don't know how to measure it's value.
But don't bash me because I didn't "just trust my guts."
quote:
Sorry to put it that bluntly, but it's true.
Ohhh, since you put it that way, it MUST be true.
Posted on 8/21/15 at 12:08 pm to Salviati
Saying "smh" to strong arguments is not a refutation.
You keep using the "bent penny" analogy, but earlier you tried to say that you don't see games as individual events... which is what the coin flip analogy is meant to illustrate.
When you view each game as an individual event, LSU is more likely to win in 10 of those individual events.
However, when you add all of the odds of success together, then divide (read, "find the average") your likelihood of success is pulled down by an outlier.
By totaling the odds, you are the one who is trying to view the season as a whole. I am viewing each individual game, as it well should be.
This post was edited on 8/21/15 at 12:10 pm
Posted on 8/21/15 at 12:13 pm to Salviati
quote:
I am NOT averaging the odds of success.
No you didn't. You simply took the sum of each of the probabilities...
quote:
100% . . McNeese St
. 55% . . Mississippi St
. 55% . . Auburn
. 90% . . Syracuse
100% . . Eastern Michigan
. 70% . . South Carolina
. 80% . . Florida
100% . . Western Kentucky
. 30% . . Alabama
. 65% . . Arkansas
. 45% . . Ole Miss
. 70% . . Texas A&M
860%
That works out to 8.6 wins for the season; thus, the lines suggest a 9-3 season, perhaps an 8-4 season.*
Let's take a look at where one sums probabilities to arrive at an overall probability for a given series:
You sum the probabilities, i.e., P(A or B) = P(A) + P(B), when the events are mutually exclusive. The games are NOT mutually exclusive.
So basically, you summed the games. Which you then conclude means we will win 8.6 games. However that is not true, what your analysis did say is that there is a probability of .86 that we win any single game in your series. That isn't true, because the events are NOT mutually exclusive.
I suggest you look up non-mutually exclusive and conditional probability and try again.
Posted on 8/21/15 at 12:18 pm to CptBengal
His problem is that he's trying to say he both believes games are not independent events (which in this example I believe they should be) while also using the "coin flip" analogy.
In this example, he's saying that we win 8.6/12 games. Which means that there is a 71.6% chance we win an individual game.
This is patently false because his own numbers bear out that our probability of success vs Bama is far lower, and vs McNeese far higher.
Thus you can't view the odds as an aggregate.
In this example, he's saying that we win 8.6/12 games. Which means that there is a 71.6% chance we win an individual game.
This is patently false because his own numbers bear out that our probability of success vs Bama is far lower, and vs McNeese far higher.
Thus you can't view the odds as an aggregate.
Posted on 8/21/15 at 12:18 pm to Guava Jelly
quote:Dear Guava Jelly,quote:This is the polar opposite of the purpose of the coin flip analogy.
I tend to agree with both of your statements. A team reacts to how if performed earlier in the year. Your second statement is reflected in the fact that betting lines change over time to reflect these affects.
The analysis in this thread appears to be over your head.
My conclusion is based soundly on three of your posts. First, you incorrectly defined and used the "gambler's fallacy." Second, you accused my analysis of averaging odds when, in point of fact, it did not; rather, it was you who was averaging odds. Third, you fail to recognize the distinction between the cumulative effects on a coin and a football team based upon the frame of reference of the prediction.
The coin flip demonstrates the mathematical accuracy of my analysis. And while a coin has no memory, a football team's chance for victory is affected by its previous games. HOWEVER, the prediction, at the time it is made, whether at the beginning of the season or midway through the season, cannot foresee the results of future games; THUS, from the frame of reference of the person making the prediction, the games have no cumulative affect on the football team, AND each game must be considered an independent event. THEREFORE, like the coin flips, the football games odds can be added together to predict the outcome for the season, or remaining games as the case may be.
Posted on 8/21/15 at 12:21 pm to studentsect
quote:Nobody ever said that it was. I did not take 86% of 12 games. I did NOT even use 86%. (BTW, 86% of 12 is 10.32.quote:Among other things, 86% of 12 things is not 8 of them.
That IS how it works.
I said that 860% = 8.6. And it does.
Posted on 8/21/15 at 12:21 pm to Salviati
The fundamental flaw here is expected value of random variable only indicates large amount of repetitions of an experiment. Using your coin toss analogy, for each match you are tossing each coin exactly once, not 100 times. Thus the outcome might not be close to expected value at all.
Additionally, although each match can be seen as independent event, the odds/lines for games are highly correlated to the outcome of previous games. Thus the pre-game odds can be a lot of different from current odds which makes the prediction inaccurate.
Additionally, although each match can be seen as independent event, the odds/lines for games are highly correlated to the outcome of previous games. Thus the pre-game odds can be a lot of different from current odds which makes the prediction inaccurate.
Posted on 8/21/15 at 12:28 pm to Salviati
You all are very smart and this has been much more fun to read than the plain ole prediction/guessing threads. In the end I hope that our QB's have developed, our players are pumped and we don't sustain anymore serious injuries. these are the things that stats and guesses can't foresee or control. Because of these, stats and guesses are replaced by hope and I hope that we get at least 10 or more wins rather than 8 this year.
Posted on 8/21/15 at 12:28 pm to moneyg
quote:Okay, bear in mind that there are TWO different arguments.quote:Can you describe where you got the chart of winning % vs. betting lines? Is it not based on historical data?
That's a really bad argument.
Your argument should be that historically, when looking at the data of betting lines vs. actual outcomes...the win % is your chart.
Is that not where you derived the chart you are using?
You are CORRECT that the correlation between betting lines and victory is based on historical data. I acknowledged that the correlation exists and that is why I used it.
The other argument about why there is correlation at all is based upon the fact that smart money tends to exploit discrepancies between a current betting line and where the betting line should be. Thus, to the extent a line is perceived to be inaccurate by smart money, additional bets will be made that will the betting line to an accurate position.
Posted on 8/21/15 at 12:29 pm to Salviati
quote:
Ohhh, since you put it that way, it MUST be true.
I'm not trying to pick a fight here, all I'm saying is it's a lot of time and energy to put into a (in my opinion) worthless analysis. Look, I'm all for a mathematical approach to performance because there's a lot to be learned there.
But this thing, trying to predict our record based on betting lines before we've played a single down this year, is wasted effort. Just my opinion.
Posted on 8/21/15 at 12:29 pm to Salviati
I'll say it again. You can't view probability as an aggregate.
8.6/12 = a 71.67% probability that we win an individual contest.
But as I said above, the likelihood that we win could be far greater or far lower in a given game.
The relative odds of success in one game (when viewed individually) has no bearing on the relative odds of success in another.
8.6/12 = a 71.67% probability that we win an individual contest.
But as I said above, the likelihood that we win could be far greater or far lower in a given game.
The relative odds of success in one game (when viewed individually) has no bearing on the relative odds of success in another.
This post was edited on 8/21/15 at 12:32 pm
Popular
Back to top
Follow TigerDroppings for LSU Football News