Bill Barnwell Bothers Me

(NOTE: I wrote the following email to Grantland last week and shared it with a few friends, as well. They so delighted in it that I thought it apt to share with the public at-large.)

I'm sure Bill Barnwell is a fine and very intelligent individual. It's clear that he knows a great deal about football and has a solid understanding of statistics.

However, he appears to know very little about gambling and continues to make elementary mistakes while providing the public at-large with horrible advice.

Allow me to count the ways in his article from last Wednesday, Oct. 26: http://www.grantland.com/blog/the-triangle/post/_/id/7807/the-hedge-the-tease-and-the-life-of-the-nfl-bettor:


1. "My best estimate on the initial bet before the game was that the Ravens would cover the 1.5-point spread 73 percent of the time."

Where did that estimate come from? That might have been the case when the line was 7.5, but when the line was 10? The probability of the Ravens covering the 1.5-point spread was AT LEAST 74.5% (per http://wizardofodds.com/sports/sports_apx10.html) and, since it was crossing the crucial 7 and 3 (while that chart looks at all sides), I think AT LEAST 76% is a much more realistic estimate. Another way to consider this is to look at the moneyline, which Bill said later in the column was +475 on the Jags.

I don't have access to the actual closing Hilton lines, but sportbet.com had -578/+472, so even though Hilton likely had -600 (or even higher)/+475, I'll use -580/+475. The "fair" price would be (580-475)/2 = 52.5 + 475 = 527.5. It should follow, then, that 527.5/627.5 = 84.06% is the frequency with which the Ravens were expected to win the game outright. So even using 76% would seem too  low unless you really believe the game will be decided by 1 point more than 8% of the time. Bill himself even uses an estimate of 1.8% for a one-point win later in the column, which is probably about right, and would of course peg the true likelihood of the Ravens winning at roughly 82%. But I'll go with 76% to be conservative here.

So, this ridiculous guess of 73 percent throws off the math in the whole column, of course ... we'll get to that.


2. "Estimated Value (EV)"

EV = EXPECTED Value. Not ESTIMATED Value. Again, this is an elementary concept for experienced gamblers.

Additionally, it's not even properly used. When Bill says the estimated value is, for instance, $33.83, that's simply incorrect. Expected value calculations need to include the original bet amount so you can properly indicate ROI ... but since Bill would probably think that ROI stands for Return on Interest rather than Return on Investment, maybe this is for the best. (OK, low blow. Sorry.)


3. "So if we bet $100 on our teaser at -120"

Who does that? Anyone with half a brain is betting $120 to win $100 rather than giving the house even more edge with breakage. This is ABSURD.


4. "JAGUARS BETTING BREAKDOWN" table

Using even the low estimate of 76% for the Ravens to cover the 1.5 point teaser, the math on Bill's teaser (as an aside, anyone betting -120 teasers is making it extremely hard on themselves) is as follows:

Ravens win by 2+ points = 76%; Return when win = $183.33; Expected return= $139.33
Jags win/tie/lose by 1 = 24%; Return when loss = -$100; Expected return=$0
EXPECTED OVERALL RETURN = $139.33

This obviously blows Bill's ridiculous hedging middle idea to pieces  straightaway, but this is nothing new -- his ridiculous practice of diluting a good bet and middle for the sake of middling was well-established in his "How to Find the Creamy Middle on an NFL Bet."


5. "That was the worst scenario imaginable, but I still saved some  money. With my hedge, I'd spent $160 and ended up with $117.14, a loss of $42.86. If I hadn't hedged my Ravens bet, I would have lost $100."

Horrible results-oriented justification of a bad bet.


6. "On Monday night, a bettor at the Hilton could have bet on the  Jaguars to win at +475 odds, a line that suggests that the Jaguars would win outright just 16.7 percent of the time."

I racked my brain for a while trying to figure out where this was coming from until I figured out that Bill was implying the Hilton was dealing a -525/+475 spread on this game. In other words, the book's hold would be only 525/625 - 500/600, or .84 - .833 -- less than 1 percent. Right.

As mentioned earlier, I don't have access to the actual closing Hilton  lines, but sportbet.com had -578/+472, so I'll repeat this again: "even though Hilton likely had -600 (or even higher)/+475, I'll use -580/+475. The "fair" price would be (580-475)/2 = 52.5 + 475 = 527.5.  It should follow, then, that 527.5/627.5 = 84.06% is the frequency with which the Ravens were expected to win the game outright."

That, of course, means the Jags should win outright just less than 16 percent of the time. There is a definite and important difference.


7. "Over the past five years, though, home teams that are underdogs  getting between 7.5 and 10 points have won outright 34.3 percent of the time, more than twice that listed price. Our estimate split the difference and suggested that the Jaguars would win outright 25.5 percent of the time."

HORRIBLE HORRIBLE HORRIBLE. I cannot stress enough just how horrible  this is. I don't have access to Spreadapedia, but I can again see from this site (http://wizardofodds.com/sports/sports_apx10.html) that the TOTAL number of away favorites in the point range from -7.5 to -8.5 from 2000 to week 4 in 2010 was 48! So, less than five per season.

Bill's stat quotes half that time period and includes favorites of up to 10 points. I'd still gander that we're talking about a subset that occurs less than ten times per season, or a total of 50 times in the past five years.

Does Bill realize that the difference from the correct percentage of roughly 16% to the observed percentage of roughly 34% is 8/50 vs. 17/50? Does Bill realize how statistically insignificant that is?!?

To make the general claim that "our estimate split the difference" is HILARIOUS. I've tried to refrain from being over-the-top/too incendiary in this email, but does Bill realize what he's implying here?

If true, this claim amounts to saying Vegas is purely wrong with these lines. In other words, Bill Barnwell apparently has figured out how to DESTROY VEGAS, everyone! Just bet on all home dogs from 7.5 to 10 points and you will make a KILLING!

This is terrible, terrible data mining at its very worst.


8."JAGUARS-RAVENS HEDGE-TEASER COMBO PLATTER"

Doing the real-world math here isn't even necessary based on all of the above, but I'd just like to point out the absurdity of switching the Ravens' win percentage from 73% in his first table to 72.6% in this table.


Again, I have nothing personal against Bill Barnwell and actually feel badly that he's in over his head here. And, on one hand, I'm OK with the misinformation he's spreading -- it makes my job as a handicapper and gambler easier when a guy who's obviously knowledgeable about football tries to write with authority on gambling and leads the public astray.

But, on the other, it irritates me to no end. Grantland is a great site that deserves better.