King Yao Prop bet article

[MicroBob]

I really enjoyed this article. And am really glad to see him writing for the magazine now.


Fascinating how such a simple bet can become so mathematically involved.


for this game I think there are some other intangibles in there...such as potentially increased likelihood of trying to get the ball to Bettis inside the 10 (because it's his last game they want him to score)....thus perhaps accidentally setting up an 'in-close' TD.


As King pointed out, it doesn't have to just depend on the numbers of previous games.
It can depend somewhat on the nature of the two teams as well as the nature of the Super Bowl (if this changes things from regular-season play somewhat, which I believe it does).


Nice article King.

[Innocentius]

There is one thing in the article I don't understand: Yao calculates the expected percentage of TDs that are 0 or 1 yards from a certain set of games, then estimates the number of expected TDs, and draws his conclusions. Why would this give a better estimate than looking at the same set of games and just calculate the frequency of 0 or 1 yard TDs?

Can someone please explain? (Sorry if this is stupid; I don't know anything about sports betting.)

[Levarkin]

Are you talking about his initial research? Because King does count the actual number of 0- or 1-yd TDs, rather than project them:

[ QUOTE ]
How likely is a 1 yard or 0 yard TD compared to other TDs? . . . The hard working bettor will put in many hours of researching boxscores and database compilations. He will go through the boxscores of each regular season game and find out how many 1 yard and 0 yard TDs were scored in each game. Then he will divide that number by the total number of TDs.

Fortunately for the lazy bettor, I have done this research already, and I am going to provide it for this article. Over the past four regular season games, 15.3% of all TDs were 1 yard or 0 yard TDs.

[/ QUOTE ]

King only projects for future games.

[King Yao]

I calculated the percentage of TDs that are less than 1.5 yards for all games.

Innocentius asked why I didn't calculate the percentage of games that had TDs less than 1.5 yards. This is a good question.

The main reason is I wanted the ability to adjust the probability of a TD less than 1.5 yards based on the total in the game. The line for the total in Super Bowl XL is currently 47. Generally, the Super Bowl has a higher Total than regular season games. The average total for PIT was 39.3 this past year, and the average total for SEA was 43.1. The average total for all regular season games this year was 40.8. So it doesn't quite follow that the Super Bowl's total is 47. If we just looked at the regular season numbers of how many games had TDs of less than 1.5 yards, we are looking at a sample size where there are fewer expected TDs than the expected TDs in the Super Bowl. With fewer TDs scored, there is a lesser chance of a TD under 1.5 yards being scored.

In the section titled "Market Adjustments", I wrote about why I thought it was important to adjust for the Total in the game. Another way to look at this is to look at games where the Total averaged around 47 to see how many of those games had TDs under 1.5 yards. But then the sample size really decreases and to get a higher sample size so one is more comfortable means going back to past years and compiling a greater database. I think that method works fine too, but the database gathering becomes the tough part.

There will be a correction to the article posted soon. Basically, in the article, for the formulas, I assumed the expected TDs is equivalent to the actual TDs scored. Instead, I should have used a distribution of actual TDs and made the calculations for a TD under 1.5 yards based on each of those possible outcomes of actual TDs. The method in the article can overstate the probability of a TD under 1.5 yards by 0% to 5%, and about 2% for Super Bowl XL. The good thing is that the betting line on this prop for Super Bowl XL is about 6% off. I will start another thread when the article is corrected, but you can see a thread on SSB where I was informed about the error and learned from it.

http://www.sharpsportsbetting.com/forums/free/nfl/index.cgi/read/11259

[King Yao]

The revised article is now online.

[21times20]

this is obviously the best 2+2 magazine article ever. thank you king yao, jerome bettis, and ben roethlisberger

[Innocentius]

Thanks for the explanation and an interesting article! Now I think I understand.

[Shandrax]

If we assume that baseball statistics over the last 100 years show that 80% of the pitches were fastballs, would it be smart to bet on it...if Tim Wakefield is on the mound?

In my opinion you can't derive much from the fact that 15,3% of TDs were 1-yard or less on the regular season. It depends too much on individual style. If both teams rely on the passing game, you might not even see a 5-yard TD in the whole game. There are too many overriding factors that can introduce tons of variance to such a bet.

To get a "clear" picture you would have to analyze the Steelers and the Seahawks and not "the NFL 2005". You have to know what plays the respective offense and defense coordinators are usually calling in such a situation and add a deception factor to that, you would even have to consider the time on the clock etc. ....and even then you probably know nothing. The reason is simple: It's extremely tough to predict plays in real time, so it's even tougher to predict them a few days in advance.

The difference between poker and sports betting is that the probability for the cards in the deck is always the same, while you never know what your favorite horse was dreaming about last night.

[King Yao]

[ QUOTE ]
The difference between poker and sports betting is that the probability for the cards in the deck is always the same, while you never know what your favorite horse was dreaming about last night.

[/ QUOTE ]

In the last section titled "Refinements", I go over some of the factors you mentioned. How one deals with those factors, and whether any factor over-rides the 15.3% number is where the 'art' comes into play. Understanding the numbers is important, but so is understanding the art of the game and when to adjust or even throw away the numbers. That is the same in sports betting as it is in the financial market - and it is also true in poker where personalities and characteristics of the players matter a ton.

[bobbyi]

Are you going to be contributing more articles about sports betting in the future, Mr. Yao?

[King Yao]

I hope so. I hope I can come up with an interesting, timely topic once a month, and hope it meets the magazine's requirements.

[StellarWind]

Very interesting article.

How many 1-yard touchdowns are set up be the dreaded pass-interference-in-the-endzone penalty? Is this an important factor? Is it useful to look at how often a team's cornerbacks are penalized for pass interference?

[King Yao]

[ QUOTE ]
How many 1-yard touchdowns are set up be the dreaded pass-interference-in-the-endzone penalty? Is this an important factor? Is it useful to look at how often a team's cornerbacks are penalized for pass interference?

[/ QUOTE ]

I think those issues could be useful, but there will be two problems with them:

1. Getting the data. It will take a ton of work to get the specific data that you mentioned, and all those hours of work may not help much at all, becase.....

2. Sample size will be small. Say two teams both had 40 TDs in the year, with about 15% of them being 1 yard TDs, the expected number of 1 yard TDs will be 6 for both teams. But one could easily have 8 and other 4, and it wouldn't be all that strange. So looking at team specifics for the season gets that small sample size problem. Throw in pass interference in the endzone calls, and it gets tougher, because just one or two calls during the season will skew the percentages for any individual team. There's a fine balance between looking at a team's specific data for the season as showing something reflective of how the team plays, versus just randomness. It's tough, that's where the "art" of making a good bet comes in.