At the beginning of October, I wrote an article titled, "Lessons from the Hilton Contest." I pointed out that keeping track of the Hiton NFL Contest provides an excellent lesson in short term vs. long term results.

There are, however, two more lessons that we can learn from examining the contest:__ LESSON #1__: Despite what you've heard, you can't win by betting against against the public because in terms of individual game win percentages the public will win more than they lose.. .

__The general public wins a higher percentage of their selections than the average sports service.__

**LESSON #2:**__Despite what you've heard from lawmakers and linemakers about the spread being the great equalizer, the football spread can be beaten.__

**LESSON #3:**At the time of my first article, the top selector had a record of 87%, and four contestants were tied for 2nd with 80% wins. There were 60 of the 329 contestants with a win-record of 67% or higher.

As predicted, those numbers have come down drastically. Just 4 weeks later, after 35 selections, the #1 conteatant no longer has an 87% record. The current top winning record is 26-9, 74%.

Far from there still being 60 contestants with a record over 67%, there are now a mere 7. My prediction is that there will be no more than one contestant above 67% when the contest ends, and it is more likely that there will be no one above the 67% mark. Only 53 contestants out of 329 (16%) are now above 60%. A mere 141 (43%) are above the 52.5% needed to overcome the vig.

You are probably now asking yourselves exactly how those numbers teach the lessons stated above. To understand, we must look deeper into the numbers. Before we begin, we will need to make some assumptions.

**The 329 contestants in the Hilton Contest represent a cross-section of the general sports-betting public.**

Theorem #1:

Theorem #1:

**Theorem #2:**To win by going against the public money, there must be more members of the general public below 50% than above it.

**Theorem #3:**If the spread equalizes each game and creates a 50-50 proposition, we should find 1/3 or more of the contestants in the Hilton Contest at 50% wins, and the balance of the contestants should be equally split above and below the 50% win line.

Since we are dealing with a pool of 329 separate bettors, we should find the above percentages over any period, long term or short term in number of games. That is true because the 329 separate bettors create their own long term result if we consider that there are 329 x 5 = 1645 separate game selections per week. At the end of the past 7 weeks there have been 11,515 selections against the spread. Thus, even though each individual contestant is still in the short term with only 35 selections, the macro percentages of wins and losses for the entire contest have now been determined over more than 11,500 selections, and can be considered to be long term.

Through the first seven weeks, there have been an odd number of selections (35) per contestant. I considered any contestant with a record of 18-17 (51.5%) or 17-18 (48.5%) , or 17-17-1 to be at 50%. We expected to find 33% of all contestants (110 contestants) at the 50% wins to losses mark. Instead, there were just 74 contestants (23%) with a 48.5% to 51.5% record.

Even if the number of contestants at 50% wins is lower than expected, we still should find that the balance of contestants are equally split above and below 50%. Instead, there were 141(43%) with a record above 50%, and just 114 (35%) below 50%. The number of contestants with less than a 50% win record is only slightly higher than the expected 33%, but the number of contestants above 50% wins is significantly higher than expected. It is obvious that the 10% of contestants who are missing from the expectated number at 50% wins are almost entirely distributed above the 50% win line.

All of the 141 contestants above 51.5% had records over 52.5%, which is high enough to produce profits after vig.

Without calculating the exact win and loss numbers for all Hilton contestants, it cannot be known if, despite the macro win percentages of the general public as represented by the Hilton Contest, the vig would make the bookmakers net winners. But it can be stated with certainty that the bookmakers are not winning the expected 4.5% of all the money crossing the table, and they may be net losers on straight bets from the general public.

If we assume that, if there were no vig, the 114 contestants losing more than 50% balance would balance 114 of their brethren who won more than 50%. There are, however, still 27 extra contestants, or 8.2% of the total contestants who are unbalanced winners. Whether or not their wins exceed the vig paid by all 329 contestants, it is clear that the extra winners would cut into the bookmaker profit from the vig.

We know we cannot win following the 74 people at 50% because we will lose the same vig whether we follow them or bet against them. The balance of the contestants, however, are net winners to the tune of 55% winners over 45% losers. That later record means that if your total losses each season are 4.5% or more of all the money you bet, you will do better by following the public.

Since the straight bets of the public will eat into the bookmaker's profits, going against the public will result in losses greater than the losses produced by a coin flip. Those additional losses will be in an amount exactly equal to the amount that the public is winning. Thus, one would be better off simply flipping a coin to make selections than betting against the public.

After 11,515 selections, it is very significant to find that winners exceed losers by 23.7%, and that winners constitute the 43% lion's share of all contestants, with losers being 35%, and those at the expected 50% being just 23% of all contestants. Such results in a sample 11,515 selections are not likely to be random. It is proof that the spread can be beaten, and that it is being beaten by the general public.

Each year, when the Las Vegas sports books provide their profit numbers to the Gaming Commission and file them with the SEC, we find that the average winnings of all the sports books range from 3.9% to 4.7% of the money bet though the window. If the public can pick more winners than losers, and perhaps even beat the vig, the question arises as to how the bookmakers manage profits that are near the 4.5% vig. The answer lies in the way the public bets.

Not all bets carry a 4.5% vig. When the public buys a half point, they increase the vig from 4.5% to 8.3%. The value of that half point is often less than half the extra 3.8% vig paid. The average increase in value for the bookmaker from buying half points in football on numbers from 1-13, excluding the numbers 3 and 7 in which still more is charged, is just 2.1%. The net to the bookmaker averages an additional 1.7% of all the money bet.

When the public bets a 2-team parlay, the vig is a huge 10%. The vig on parlays with more teams is higher, and the vig on the popular parlay cards is higher still.

On a 2-team, 6-point teaser in the NFL, laying 11-10, the vig is a huge 47.5%. In football, the amount the bookmaker gives back with the 6-point line move depends on the numbers being purchased, and whether the teaser is in the NCAA or the NFL. Nevertheless, the value returned to the player can be as little as 7.5% when turning a favorite into an underdog by teasing across pick'em. The net profit to the bookmaker in such circumstance is 40% of all the money crossing the table.

With such large profits emanating from poor betting technique, the real question is not how the books make money, but why they don't make more than the average 4.3% that they report each year. The most reasonable answer is that the straight bettors, and those using advantage techniques, both public and wiseguy, eat up all the extra profits..

It is intersting to note that not only does the public do well, but they do better than the average sports service as measured by the records of The Sports Monitor of Oklahoma City. Last year, of the 218 sports services being tracked by the Sports Monitor, only 64 (29%) were above the 52.5% needed to win after subtracting the house take. That compares to 44% of the public currently above that percentage in the Hilton Contest. In fact, the Sports Monitor percentage of winners for the 2008 NFL Season is close to the random results we would expect to get if all contestants were flipping a coin, or if the spread caused game results to be strictly a matter of luck. .

It might be argued that the Sports Monitor results are full year, and therefore more long term and perhaps more representative than the seven week record in the Hilton Contest. That argument is probably spurious.

The Sports Monitor handicappers averaged only 72 selections each for the entire season. Several of the handicappers had 40 or fewer selections for the entire season. Based on the top 64 Sports Monitor handicappers, I extrapolated the total number of season selections at the Sports Monitor last year to be 15,700. A sample of 15,700, as compared to a sample of 11,500 in the Hilton Contest, is not a sufficient percentage difference to explain a 15% difference in the pool of winners between the two groups. A statistical sample of 11,500 selections is more than sufficient to validly extrapolate that there will still be more winners than losers in the Hilton Contest when it reaches 15,700 selections.

If the spread makes sports selection a 50-50 proposition, or if the contestants were flipping a coin, then we would expect more and more people to get closer to 50% as the sample of games gets larger. If, on the other hand, the selectors are skilled handicappers and have an edge, the larger the sample the larger the percentage of winners will become and the smaller the percentage of losers will shrink. Thus, if the sample of the Sports Montior is larger and more representative, and we assume that the pool of handicappers are skilled, the Sports Monitor winning percentages should increase and become larger than the results in the Hilton Contest.

Further, any advnatage that may have been gained by the effect of luck on the smaller 11,500 sample in the Hilton Contest is balanced by the fact that the contestants at the Hilton must select 5 games each week, while the handicappers with Sports Monitor can pick and choose their spots, perhaps selecting only one game one week, and 6 games the next. The handicappers at Sports Monitor also have the aability to pick and choose their line based on the time of release of their play. The Hilton contestants must bet based on the fixed line set by the Hilton.

Sports Monitor handicappers are also given the huge advantage of an assumption of the best line when more than one line exists. That means that if a game is selected against a line that is either 3.5 or 4, all the handicappers who selected the underdog are given the line of +4 for the push, while all the handicappers who selected the favorite are given the line of -3.5 to log the win. Hilton Contestants have no such advantage.

Given the advantages of the Sports Monitor handicappers, we would expect to find the professional handicappers to have the same or a higher percentage of winners, and if they are skilled, the extra 4000 selections would create an even higher probability that the skill would show increasing the number of handicappers who are above 52.5%. The low comparative percentage of winners at the Sports Monitor is an indication of a skill level equivalent to flipping a coin.

It should be noted that the above conclusions are based on averages of all 218 handicappers submitting selections to Sports Monitor, and do not reflect the ability of any individual handicapper at Sports Monitor, who may be very skilled and far more skilled than the average Hilton contestant.

It should also be noted that the conclusions are based on the assumption tha the Hilton contestants represent a statitically valid cross-section of the sports betting public, and are not a pool of particularly skilled sports bettors..

To check the validity of our conclusions and extrapolations, I will review the numbers again when the Hilton Contestants reach 70 or 75 selections, as well as at the end of the Contest.