- Joined
- Nov 3, 2011
- Messages
- 6,077
Hey Guys, posting this from the WEZ with permission from poster.....below is a pretty interesting analysis of the point spread by sookdd. All credit to him.
Largest Point Spread In History: An Analysis I realize there have been a few good threads regarding the point spread. I am starting a new one to provide an analysis.
I start with an interesting nugget: 21.5 points is the largest point spread ever between two top 10 undefeated teams. You read that right. Undefeated teams, both in the top 10, have never played a game in which the line was this large. Translate that to there has never been as big a supposed mismatch between two top 10 undefeated teams. So what gives?
Some background. I have done plenty of work on Vegas lines (statistical and otherwise). I also made a career of exploiting market anomalies/pricing inefficiencies.
How is it that this game has the largest point spread in history between two undefeated top 10 teams?
Lines are "Right"
Vegas spends a lot of money making sure the lines are right (including hiring MIT "consultants" to crunch numbers). The lines are right. If you looked at all the games that have ever had a line put on them, you will find that the favorite covers almost exactly 50% of the time, and the underdog covers almost exactly 50% of the time. You will also be able to plot a bell curve of results of all games, with the peak of the curve being right at the line. Of course, you will have massive underdogs that win games. That is the tail of the bell curve. It exists, but it is rare. And it occurs almost the exact amount of times that the statistics tell you it should. What all that means is if you are going to bet on games, assume that over the course of your betting career, you will win 50% of the time. And if you pay a vig, at the end of your betting career, you will be down 5% of the total volume of bets you have made. The lines are that good.
Further, the market does not dictate lines, rather the lines dictate the market. Vegas of course wants half the people to bet on one side, and the other half on the other. And the best way to make that happen is to get the line right, and get a lot of volume. Yes, they will adjust lines to compensate for market directionality - but that is for internal risk management. If in aggregate they get their lines right, they will make 5% of the volume.
How are the lines made - why are we 21.5 point underdogs?
Lines are made by very smart people - who make lines for a living. Some can simply look at a game and say: Canes-Noles, 21 points. Others use a computer. Others use both. What are they mostly looking at? Scores of previous games. Next is personnel changes. Next is conditions (home/away, weather, etc.). This is why the best opportunities for anomalies come in the early part of the season. It's not to say these lines are wrong. In aggregate, they are right. There is just greater variance of results (the bell curve is flatter). It's also why the bell curve is flatter in college vs. the NFL. No lines are better than the NFL, there is simply more data, and less turnover of personnel. College teams change much more substantially from year to year.
Scores of previous games have a tremendous impact. FSU destroys #3 Clemson on the road 51-14. They obliterate a #25 ranked Maryland team 63-0. These weigh heavily on the line. They weigh heavily on voters, opinion what have you. There is some analysis of the why - sure. But the bottom line is, FSU destroyed Clemson. Remember in 2000 - we barely beat Louisiana Tech. We dropped in the polls. And most likely, that single poor performance, at home against a pathetic opponent, cost us a National Championship. Score a few more touchdowns in that game, and we almost certainly have a sixth ring. Is that fair? Of course not. But it's reality.
What does the line say about our chances of winning?
5%. 21.5 point underdogs win roughly 5% of the time. Play this game 20 times, we win once. Again, never in history have two unbeaten top 10 teams played a game in which one team was perceived to be a near certainty of winning the game to this degree.
Is this line right?
Well, as already stated above, Vegas lines are right. However, I also claimed expertise in pricing anomalies, and that sometimes the bell curve can be flatter. Let's take a closer look at the factors that create the line, and ways in which this line might be skewed.
FSU Wins
FSU beats Pitt 41-13. First game of the year on the road. Winston nearly perfect. Throws a TD in the fourth quarter - an unnecessary one. Can you ask them to stop playing? Of course not. The game at the half is 28-10. It's lopsided no matter how you slice it.
FSU beats Nevada 62-7. It is 17-7 at the half. They score 31 in the third. They add 14 more in the fourth with backups.
FSU beats B-C 54-6. It is a blowout at the half 33-0. Their backups add another 11 points.
FSU beats BC 48-34 on the road. This is their lone "blemish". And it is still by 12 points to a not horrible BC team. If all you saw was the score, you might thing FSU handled them. BC was up 17-3. FSU gets a big play (a 56 yard TD). Then there was the Hail Mary to end the first half - giving FSU a 24-17 lead. BC still fought in the second half - but that was a heartbreaker (the Hail Mary).
FSU beats Maryland 63-0. That score gets computers and lines makers going. Keep in mind, by this point, FSU is now being considered in the hunt for a title. And impressing pollsters matters. The score matters - as opposed to simply winning games. They are up 21-0 at the half, and Maryland has no life. In the third quarter, FSU is still passing, and Winston getting three TDs. He adds another one in the fourth. Is FSU 63 points better than Maryland? The score indicates so. But at this point, we are beginning to see some possible skew.
FSU beats Clemson 51-14. This game is likely the largest determinant of the line - or at least having the largest impact. Clemson has some solid wins, it was at Clemson. #3 vs. #5. Is FSU 37 points better than Clemson? Now, it is ridiculous to say that Florida gave us the game - just like it would be ridiculous to say Clemson gave it to FSU on all the early turnovers. However, we are simply making a cold blooded analysis of the score. If this game was played 10 times, what would the average score be? More likely, FSU would win by an average of 1-2 TDs. This game was over in the first quarter when it was 17-0. Quite possibly, it was over when Clemson turned it over early on their first drive. Mentally, Clemson was out early. And that is a recipe for a blowout, when otherwise it might not have been. If FSU wins this game 35-28, then the line this week would be a TD.
FSU beats NC State 49-17. Here it was 42-0 at the half. NC State was the one that put up 17 "easy" points. This score is likely a good indicator of the differential.
Miami Wins
Canes beat FAU 34-6. Not a massive blowout. Game is over 20-3 at the half. Backups playing for Miami in third quarter add 7 more points to differential. Shut it down for the fourth.
Canes beat Florida 21-16. Win by turnovers? Sure. They were almost all forced. No need to rehash - what mattes is the score. Recall that Florida scored a TD with 2:08 to go to make it 21-16. Not to say this was a meaningless TD, it was legit. But while many argue that we could/should have lost, a case could also be made that Canes were more like 12 points better in the game (yes, factoring in the turnovers, which count). If that was the final, the line this week is likely 7 points tighter (from that alone).
Miami beats Savannah State 77-7. It is 49-0 at the half. There was no 4th quarter scoring.
Miami beats USF 49-21. This game is over at the half - 35-7. Ryan Williams in at QB. They get two fourth quarter TDs (one off pick-6). If this game ended 49-7, the line this week is probably 3-4 points tighter.
Miami beats GT 45-30. Two TD differential to a decent GT team. This was a back and forth dogfight all the way through - and it was ied at the half. Canes took care of business in second half (particularly fourth quarter) and pulled away. Note a last second TD for GT to make it a 15 point game rather than a 22 point game. This win looks similar to FSU's BC win on paper. Though most likely it is a better win for Canes.
Miami beats UNC 27-23. This game has big impact on line. UNC with a terrible record. Only a 4 point win. Score most likely accurately reflects the game - it was a dogfight all the way. Is UNC better than their record? Possibly.
Miami beats Wake 24-21. A three point win against a reasonably solid team. Again, a dogfight.
One thing you might note - no common opponents. This adds to the possibility of an anomaly. In the bond market, there is a pricing anomaly when a 9.5 year bond trades at a large spread to a 9.75 year bond. This is a rarity. They both sit next to each other on the yield curve, easy to compare. Let's say that FSU pummeled UNC 75-0, the 4 point win for us would look much worse. How badly will the Seminoles beat the Gators? Vegas would surmise that game to likely be a 28 point spread. But had the Noles already played them, and won by less than a TD, then the line this week would be much tighter.
I am not saying there is an anomaly in the 21.5 line. It could very well be right, and we have less than a 5% chance of winning this game. But to accept its correctness, you would have to presume that the scores of all the games for both teams to this point accurately reflect their strength.
Also keep in mind that lines are cold blooded. They do not typically factor things in like - this is a rivalry game. Yes, it might be factored in slightly - based on a history of this game in particular. But that factor pales in comparison to the 51-14 Clemson win.
Personally, I believe that the previous scores do not accurately reflect the relative strengths of the teams on November 2nd. I would make Noles probably a 10-12 point favorite (including the home field factor). That would be mostly based on my feeling that the final Clemson score was an outlier.
And that translates into us having roughly a 20% chance of winning (winning the game 1 out of 5 times played).
Having said all of that, I am left with one conclusion: We will win this game. Go Canes!
Largest Point Spread In History: An Analysis I realize there have been a few good threads regarding the point spread. I am starting a new one to provide an analysis.
I start with an interesting nugget: 21.5 points is the largest point spread ever between two top 10 undefeated teams. You read that right. Undefeated teams, both in the top 10, have never played a game in which the line was this large. Translate that to there has never been as big a supposed mismatch between two top 10 undefeated teams. So what gives?
Some background. I have done plenty of work on Vegas lines (statistical and otherwise). I also made a career of exploiting market anomalies/pricing inefficiencies.
How is it that this game has the largest point spread in history between two undefeated top 10 teams?
Lines are "Right"
Vegas spends a lot of money making sure the lines are right (including hiring MIT "consultants" to crunch numbers). The lines are right. If you looked at all the games that have ever had a line put on them, you will find that the favorite covers almost exactly 50% of the time, and the underdog covers almost exactly 50% of the time. You will also be able to plot a bell curve of results of all games, with the peak of the curve being right at the line. Of course, you will have massive underdogs that win games. That is the tail of the bell curve. It exists, but it is rare. And it occurs almost the exact amount of times that the statistics tell you it should. What all that means is if you are going to bet on games, assume that over the course of your betting career, you will win 50% of the time. And if you pay a vig, at the end of your betting career, you will be down 5% of the total volume of bets you have made. The lines are that good.
Further, the market does not dictate lines, rather the lines dictate the market. Vegas of course wants half the people to bet on one side, and the other half on the other. And the best way to make that happen is to get the line right, and get a lot of volume. Yes, they will adjust lines to compensate for market directionality - but that is for internal risk management. If in aggregate they get their lines right, they will make 5% of the volume.
How are the lines made - why are we 21.5 point underdogs?
Lines are made by very smart people - who make lines for a living. Some can simply look at a game and say: Canes-Noles, 21 points. Others use a computer. Others use both. What are they mostly looking at? Scores of previous games. Next is personnel changes. Next is conditions (home/away, weather, etc.). This is why the best opportunities for anomalies come in the early part of the season. It's not to say these lines are wrong. In aggregate, they are right. There is just greater variance of results (the bell curve is flatter). It's also why the bell curve is flatter in college vs. the NFL. No lines are better than the NFL, there is simply more data, and less turnover of personnel. College teams change much more substantially from year to year.
Scores of previous games have a tremendous impact. FSU destroys #3 Clemson on the road 51-14. They obliterate a #25 ranked Maryland team 63-0. These weigh heavily on the line. They weigh heavily on voters, opinion what have you. There is some analysis of the why - sure. But the bottom line is, FSU destroyed Clemson. Remember in 2000 - we barely beat Louisiana Tech. We dropped in the polls. And most likely, that single poor performance, at home against a pathetic opponent, cost us a National Championship. Score a few more touchdowns in that game, and we almost certainly have a sixth ring. Is that fair? Of course not. But it's reality.
What does the line say about our chances of winning?
5%. 21.5 point underdogs win roughly 5% of the time. Play this game 20 times, we win once. Again, never in history have two unbeaten top 10 teams played a game in which one team was perceived to be a near certainty of winning the game to this degree.
Is this line right?
Well, as already stated above, Vegas lines are right. However, I also claimed expertise in pricing anomalies, and that sometimes the bell curve can be flatter. Let's take a closer look at the factors that create the line, and ways in which this line might be skewed.
FSU Wins
FSU beats Pitt 41-13. First game of the year on the road. Winston nearly perfect. Throws a TD in the fourth quarter - an unnecessary one. Can you ask them to stop playing? Of course not. The game at the half is 28-10. It's lopsided no matter how you slice it.
FSU beats Nevada 62-7. It is 17-7 at the half. They score 31 in the third. They add 14 more in the fourth with backups.
FSU beats B-C 54-6. It is a blowout at the half 33-0. Their backups add another 11 points.
FSU beats BC 48-34 on the road. This is their lone "blemish". And it is still by 12 points to a not horrible BC team. If all you saw was the score, you might thing FSU handled them. BC was up 17-3. FSU gets a big play (a 56 yard TD). Then there was the Hail Mary to end the first half - giving FSU a 24-17 lead. BC still fought in the second half - but that was a heartbreaker (the Hail Mary).
FSU beats Maryland 63-0. That score gets computers and lines makers going. Keep in mind, by this point, FSU is now being considered in the hunt for a title. And impressing pollsters matters. The score matters - as opposed to simply winning games. They are up 21-0 at the half, and Maryland has no life. In the third quarter, FSU is still passing, and Winston getting three TDs. He adds another one in the fourth. Is FSU 63 points better than Maryland? The score indicates so. But at this point, we are beginning to see some possible skew.
FSU beats Clemson 51-14. This game is likely the largest determinant of the line - or at least having the largest impact. Clemson has some solid wins, it was at Clemson. #3 vs. #5. Is FSU 37 points better than Clemson? Now, it is ridiculous to say that Florida gave us the game - just like it would be ridiculous to say Clemson gave it to FSU on all the early turnovers. However, we are simply making a cold blooded analysis of the score. If this game was played 10 times, what would the average score be? More likely, FSU would win by an average of 1-2 TDs. This game was over in the first quarter when it was 17-0. Quite possibly, it was over when Clemson turned it over early on their first drive. Mentally, Clemson was out early. And that is a recipe for a blowout, when otherwise it might not have been. If FSU wins this game 35-28, then the line this week would be a TD.
FSU beats NC State 49-17. Here it was 42-0 at the half. NC State was the one that put up 17 "easy" points. This score is likely a good indicator of the differential.
Miami Wins
Canes beat FAU 34-6. Not a massive blowout. Game is over 20-3 at the half. Backups playing for Miami in third quarter add 7 more points to differential. Shut it down for the fourth.
Canes beat Florida 21-16. Win by turnovers? Sure. They were almost all forced. No need to rehash - what mattes is the score. Recall that Florida scored a TD with 2:08 to go to make it 21-16. Not to say this was a meaningless TD, it was legit. But while many argue that we could/should have lost, a case could also be made that Canes were more like 12 points better in the game (yes, factoring in the turnovers, which count). If that was the final, the line this week is likely 7 points tighter (from that alone).
Miami beats Savannah State 77-7. It is 49-0 at the half. There was no 4th quarter scoring.
Miami beats USF 49-21. This game is over at the half - 35-7. Ryan Williams in at QB. They get two fourth quarter TDs (one off pick-6). If this game ended 49-7, the line this week is probably 3-4 points tighter.
Miami beats GT 45-30. Two TD differential to a decent GT team. This was a back and forth dogfight all the way through - and it was ied at the half. Canes took care of business in second half (particularly fourth quarter) and pulled away. Note a last second TD for GT to make it a 15 point game rather than a 22 point game. This win looks similar to FSU's BC win on paper. Though most likely it is a better win for Canes.
Miami beats UNC 27-23. This game has big impact on line. UNC with a terrible record. Only a 4 point win. Score most likely accurately reflects the game - it was a dogfight all the way. Is UNC better than their record? Possibly.
Miami beats Wake 24-21. A three point win against a reasonably solid team. Again, a dogfight.
One thing you might note - no common opponents. This adds to the possibility of an anomaly. In the bond market, there is a pricing anomaly when a 9.5 year bond trades at a large spread to a 9.75 year bond. This is a rarity. They both sit next to each other on the yield curve, easy to compare. Let's say that FSU pummeled UNC 75-0, the 4 point win for us would look much worse. How badly will the Seminoles beat the Gators? Vegas would surmise that game to likely be a 28 point spread. But had the Noles already played them, and won by less than a TD, then the line this week would be much tighter.
I am not saying there is an anomaly in the 21.5 line. It could very well be right, and we have less than a 5% chance of winning this game. But to accept its correctness, you would have to presume that the scores of all the games for both teams to this point accurately reflect their strength.
Also keep in mind that lines are cold blooded. They do not typically factor things in like - this is a rivalry game. Yes, it might be factored in slightly - based on a history of this game in particular. But that factor pales in comparison to the 51-14 Clemson win.
Personally, I believe that the previous scores do not accurately reflect the relative strengths of the teams on November 2nd. I would make Noles probably a 10-12 point favorite (including the home field factor). That would be mostly based on my feeling that the final Clemson score was an outlier.
And that translates into us having roughly a 20% chance of winning (winning the game 1 out of 5 times played).
Having said all of that, I am left with one conclusion: We will win this game. Go Canes!
Last edited:
