There’s Nothing Worse Than A Chalky Aftertaste: Big Favorites Relax After Halftime | Stat Geek Idol

posted in NCAA Basketball, NCAA Tournament, Stat Geek Idol

This is a Sweet 16 submission in our inaugural Stat Geek Idol contest. It was conceived of and written by Bill Bihn. The statements below reflect the opinions and positions of the author solely, and not those of TeamRankings.

What do the basketball programs of UNC-Asheville, Appalachian State, and Monmouth have in common?

1) You probably can’t name all three of the schools’ mascots.
2) They all played against the North Carolina Tar Heels this season and covered the point spread.

In case you haven’t noticed, the three programs I mentioned above aren’t elite, or even close to it. You probably can’t name a single player from any of the three squads. As you would expect, it turns out that Asheville, Appalachian State, and Monmouth all lost by double digits when they played the Tar Heels. Appalachian State fared the best, “only” losing by 15. Asheville lost by 16 and Monmouth lost by 37. None of these three schools were competitive.

They didn’t have to be. They all covered. Yes, even Monmouth (the spread was 39.5).

Betting on huge favorites, or “chalk,” is often a trap. In fact, home teams favored by double digits during the last two months of the 2011-2012 college basketball regular season only covered the spread 46.5% of the time (190 wins, 219 losses, and 11 pushes against the spread). Even if you consider a margin of error due to sample size, it is still safe to say that betting on these teams isn’t profitable in the long run.

So, why do these superior teams fail to cover the spread so often? Let’s look at it from two different perspectives: oddsmakers’ logic and the statistics.

Looking Through The Eyes Of A Sportsbook Manager

If you are one of the millions of tourists walking through a casino every year, you may be interested in placing a wager on a college basketball game. If this college basketball game features North Carolina against Monmouth, you probably won’t be betting on Monmouth. In fact, the first question out of your mouth will likely be one of the following.

1) “Does Monmouth really have a basketball team?”
2) “Where is Monmouth?”
3) “What kind of idiot would bet on Monmouth?”

Just so you know, the answers are as follows:

1) Yes. Yes, they do.
2) West Long Branch, New Jersey
3) Idiots don’t bet on Monmouth, but winners do.

Since sports book managers know that the majority of casual bettors will gravitate toward the school they actually know something about, they can set the point spread extremely high. If they think the big-name institution will win the game by 15 points, they may consider releasing a point spread of 16 or 17 points instead.

Will that extra point or two change the way the public reacts to betting on the game? Not much. What this does, however, is give the underdog a slightly better chance of covering the spread. Since most bettors choose the favorites, this skewing of the spread means a better “bottom line” for the book itself.

None of this is to say that North Carolina couldn’t beat Monmouth by 50 points if they wanted to. North Carolina led their game against Monmouth by 31 at the end of the first half. Had they wanted to score 120 points, they probably could have. Instead, they decided to let up, take their foot off the gas pedal, and rest. By the time the game was over, head coach Roy Williams had decided to play 15 different players in that matchup.

Ultimately, when betting on favorites, the question should not be if the best team CAN win by 20 points, but rather if the best team CARES ABOUT winning by 20 points. Usually, the answer is no.

Now From A Statistical Point Of View

From January 8 to March 7, 2012 (60 days), I checked Vegas Insider point spreads and results for every game offered by the Las Vegas Hotel and Casino (LVH) sports book. The main reason why I used this data is because the LVH offers both first-half and game wagering.

By studying the difference between first-half spreads and game spreads, I began to see statistical relationships that strongly pointed to my hypothesis that betting against home favorites in certain situations could be rewarding.

First, we need to take a look at the relationships between first-half point spreads and game-long point spreads in games where the home team was a significant favorite.

As you can see from the table, sports books generally create first-half betting lines and game lines using a simplistic formula. The game line is always 1.69 to 1.92 times (we’ll call it 1.8, or 80%) higher than the first-half line. Even in games where the point spread is particularly large, the ratio still holds fairly close. It is assumed that the better team will grow their halftime lead by about 80% by the time the game ends.

From a bettor’s perspective, this makes the lines very predictable. If you know that a game has a point spread of 9, the first-half spread will either be 5 or something close to it, because 5 multiplied by 1.8 is 9. If a game has a point spread of 13.5, the first-half spread will be near 7.5 (7.5 x 1.8 = 13.5). The rule holds steadfast even in games where the home team is favored by 10 points in the first half alone. The game line will likely be 18, because 10 x 1.8 is 18.

How To Use The 1st Half/Game Spread Ratio To Your Advantage

This ratio seems insignificant, but in reality it yields a huge advantage for the betting public—that is, if they are aware of it. When a team is a heavy favorite and takes a commanding lead in the first half, they are far less likely to actually increase their lead by 80%. They are more likely to perform in the manner of North Carolina in our example. They will usually rest their star players and turn in a lackadaisical effort.

Had North Carolina actually increased its first half lead by 80%, they would have won against Monmouth by 56 points. It is ridiculous to think that we should treat one-sided games the same as closer ones from a betting perspective, but as you can see from the table, college basketball linesmakers continue to do it.

Let’s get proof of this from our 60-day 2012 sample.

In the 119 charted instances of a home team being favored by 15 or more points, the average halftime score was 37.21 to 25.71. The home team led by an average of 11.50 points at the halftime break. However, the average final score was 75.67 to 57.70, an average margin of 17.97 points. What you should notice at this point is that the average team didn’t increase their lead by 80%. In fact, 17.97 divided by 11.5 is 1.56.

The average team only increased their lead by 56%! As discussed earlier, this is likely an effect of a team either losing focus or resting star players. Either way, it doesn’t bode well for the betting favorite. In these 119 games, the home team was 55-63-1 against the spread (46.6%).

Teams Up Big Take Their Foot Off The Gas

If the previous result wasn’t revealing enough, look at our 60-day sample from one more point of view. In this final example, ignore the point spread completely and just pretend that you placed a bet on a random team. This team played very well in the first half and is winning by double digits. How high should you expect the lead to grow before the end of the game?

As it turns out, not high at all.

In the sample, there were 388 instances on the charts where a home team led by double digits at halftime. Amazingly, of those 388 double-digit leads, only 212 of the leads grew at all! Not only that, there were only 51 instances out of the 388 where the lead grew by more than 80%. On average, the lead grew by a mere factor of 1.09.

Try to cast that 1.09 figure in a different light. If you are a college student cheering your favorite team to victory and the raucous crowd has helped establish a 20-point halftime lead, what should you expect the final margin of victory to be?

About 22.

Based on that number, I’m guessing the crowd won’t be quite as raucous during the second half. That is, unless they give away free pizza every time the home team scores 100 points or more. Unfortunately though, the study of how pizza giveaways affect home-court advantage is going to have to wait until another day.

In the end, numbers indicate it is certainly possible for a double-digit home favorite to cover the spread—just not likely. Many of these teams start very strong and jump out to substantial first-half leads. However, very few of these teams choose to keep the same scoring pace during the second half. Since betting lines generally assume good teams will keep up the pace and increase their lead by 80% (even though this isn’t actually true), the lines get inflated substantially higher than they should. As a result, betting underdogs in these circumstances has the potential to be profitable.

  • Davidperl

    this may be pretty hard but you can guys find an article that looks at the characteristics of a pool champion like how many final fours did they pick correctly, what was their overall score? that kind of thing

  • Joe McMackin

    This is great analysis but the part where you throw out the point spreads doesn’t make sense.  Mainly the reason being that 270 of those teams aren’t big favorites and more than likely those 270 will be favored meaning the second half deficit shouldn’t increase.

    It would be great to take this one step further and add in first half shooting percentages.   My guess is that big favorites that under shot what they should shoot against a certain team were more likely to cover in second half and if a team shot 70% they were less likely to cover.

  • David Hess


    That stuff can vary dramatically from year to year. For example, last year, nailing 2 Final Four teams made you an amazing psychic. Whereas in 2008, we picked all four correctly, and so did other people who weren’t afraid to back chalk.

    It also depends heavily on pool size.

    However, those are interesting questions. We have a pool simulation engine that might be helpful in answering those. When we rev it up again next year, we should investigate.

  • David Hess

    Joe — you’re right, it would be pretty interesting to look at shooting percentages by half. Unfortunately that data is pretty tough to come by.

  • Larry Gray

    Stat geek idol adds a dimension worth adding to one’s analysis. Luck is still a factor
    and one more element to make us diversify our brackets make sense. I am at 42 out of 58 thus far in using Team Ranking selections or 98% on Yahoo and would be better had
    I not selected St. Mary’s winning early on. I thought of changing Missouri, but I felt
    good about them also. Who could have determined Norfolk State would defeat them?
    That really came out of no where, except maybe for an exceptional, magical effort and
    performance by Norfolk State. That they got defeated by at least 30 points at the next
    level made sense.

    I am still in the hunt, barely, in my smaller $$ pool with Team Ranking Selections, though now that I have learned a little, will enter more pools next year, and take more care in
    alternating selections making it to the final 8 on up. My hunch on Louisville would have
    made a big difference for me. Also on  free entry yahoo bracket I saw Indiana going
    this far and VCU going as far as it did. With modifications like this I could have
    been at least above 99% instead of just 98% in yahoo! In the smaller pool, I would
    have done one more bracket with Kentucky in the final with Ohio State instead
    of just one with Missouri! If Baylor or Louisville gets there, oh well!

  • Joe McMackin

    Agreed, but no reason why you can’t start collecting it now right :)?  It’s obviously nice to have historical data but collecting things like this would be great and set you guys even further apart.  A perfect example was the kentucky/baylor game.  Kentucky was shooting 65-70% at half while baylor was shooting something like 35%.  At half time the line was a pick and baylor was down 20.  Those evened out and Baylor easily covered.  Obviously helping the fact was that Calipari reigned in his team but that usually happens in 20 point ball games.

    Really liking the site.  I couldn’t believe UNC’s first half vs. second half margin.  In comparison to the other higher echelon teams it’s a pretty big difference.