The World's Greatest NFL System...

[taxi driver]

NFL Magic Number Powerplay

It was 1983, and at the ripe age of 20 I was in my 4th year of learning the art of Handicapping the NFL. Some of my earliest impressions were formed from articles by Ed Horowitz and Gary Austin. Although Ed 'the Professor' was primarily a statistical capper, it was his technical articles that intriqued me the most. It took me a while, but I eventually grasped the idea that the final score very rarely reflects the statistics due to the many variables. At that point, I understood that a lot of things go on in a game, and that I was eager to get those things 'on my side'.

Among some of the research Ed and Gary printed was a quite simple system: Go Against any NFL team that scored 30 points or more in each of their last two games. I remember back then it was fairly successful, but given a 20-year data base right now, I doubt it's better than 50%. But it nevertheless shaped my thinking. At that point I began to realize that how teams score is somewhat irrelevant. Likewise, how they give up points was also virtually irrelevant.

Why? Because you don't get paid on yardage, you get paid on the final score. That is what intrigued me about some of Ed's ideas in regard to magic number systems using numbers like 30 and other systems using numbers like 10, or single digits, or zero. Over the years, there have been multitudes of variations of the 30-30 including going against teams that scored 70+ in the last two or 90+ in the last three. Man, the Rams alone would've eaten your lunch with that.

It's a different ball game now, but in 1983, just out of high school and living in a boarding house by choice, I uncovered a nugget through my usual research that lives on today. A simple nugget from the old school, I wondered what would happen if I combined those magic numbers of 30 and 10. In other words, how would a team respond if they scored 30 or more in each of their last two games AND allowed 10 or less in each of those same two games? I called it a "double peak". To me, those numbers respresented a 'team peak', and surely if a team accomplished that back to back, they would let down.

Remember, my thinking was that it is irrelevant how teams scored or how they allowed points. In other words, a statistical analysts would insist that scoring 30 back to back was not necessarity a 'double peak' offensively. Their defense could've scored 3 TD's while their offense actually struggled (and therefore did not peak). Likewise, their defense allowing only 10 points or less did not necessarily mean they "peaked". Heck, they could've given up 400 yards and still allowed only 10 points (and therefore, technically, I could not label it a defensive peak). That was the argument and it made sense, but only for a short time.

Then it dawned on me. If I go against a team off a 'double-peak' (believing somewhat incorrectly that the points reflect actual output), I am not banking on what this team will do (not do) in regard to 'output' in the first place, I am banking on what this team will do (not do) in regard to the SCOREBOARD. When the final whistle blows, that's all that matters. I could care less if the team I'm going against roles up 600 yards. What does the scoreboard say?

So with those thoughts in mind, I re-defined (for myself) what a peak was. It had nothing to do with output. To me, a "Team Peak" was determined by the Points Scored "IN ANY WAY, SHAPE, OR FORM". Now that made sense to me. Forget who outplayed who. Which way did the ball bounce? Who got the breaks? If a team gets the breaks and if the ball bounces in their favor two week's in a row to the tune of 30-10 or better, I'm going against them. Since then, I put in a provision that would allow for a safety. So the actual numbers I use are 30-12.

From 1984 through 2000, the System is 17-7 for a healthy 71%. However, I have found that the system is only 50/50 (7-7) in division games, leaving it at a perfect 10-0 in non-division games. That makes perfect sense to me. After a couple peaks, the team is now playing a game of lesser importance. Stands to reason that they would be more inclined to let down in that role.

***GO AGAINST AN NFL TEAM THAT SCORED 30 OR MORE POINTS IN EACH OF THEIR LAST TWO GAMES WHILE ALLOWING ONLY 12 OR LESS POINTS IN THOSE SAME TWO GAMES IF THEY ARE NOW PLAYING A TEAM OUTSIDE OF THEIR DIVISION***

I marketed this sytem in 1987 and printed it in a Nationwide Publication in 1989. I caught some flack from a few services for putting it in print back then. They called me at my house and gave me a hard time, claiming, "that kind of stuff is for 'US'". I said, "Who's 'us'?" And the individual (whom you probably know) hung up. Anyway, I have posted it over the years and especially since 1998 when I started posting on the internet.

You may remember seeing it in 1999 at xxxxxxxxx.com when the Rams -8 fell victim to the system by losing straight up to the Eagles 38-31. Or last year when the Bucs -7 lost straight up to the Jets 21-17. The 10-0 system recently moved to 11-0 a few weeks ago when the Rams -11' were coming off a double peak. They almost lost straight up at home as the Giants gave them all they could handle in the Rams narrow 1-point victory 15-14.

It doesn't come up often (about 1/2 of 1% of all games), but when it does, it's hammer time. Here's the games since 1984...


1985: *DOLPHINS (+4 ) 38, Bears 24 Win
1988: *Eagles (+6) 21, VIKINGS 23 Win
1992: *Oilers (+6) 17, VIKINGS 13 Win
1992: *CHIEFS (+2) 24, Eagles 17 Win
1993: *Bengals (+24) 8, 49ERS 21 Win
1996: *49ers (+6) 20, PACKERS 23 Win
1996: *COWBOYS (-5) 12, Patriots 6 Win
1999: *COLTS (-4') 25, Chiefs 17 Win
1999: *EAGLES (+8) 38, Rams 31 Win
2000: *Jets (+7) 21, BUCCANEERS 17 Win
2001: *Giants (+11') 14, RAMS 15 Win


There you have it...the World's Greatest NFL System...

Enjoy,
dave

 

[plasticman]

Thank you.

 

[coldbeer4soul]

The rams in week #6 won 34-14 vs. jets. How did that fit in your system?

 

[count zero]

Have known dave (taxi) for quite a while now, and respect his experience and hard work. We posted on the same board the last 2 years. But I do have a few negative things to say about this approach.

I spent the years from 1976 or so to 1989 using these kinds of methods. I learned to program in 1980 and wrote my own research software. I compiled a huge library of angles and tracked every imaginable subset year after year (and still do). In the seventies and early eighties, before everyone had computers, I did OK, although basically most of the angles I tracked ultimately seemed to come down to playing the motivated home dog. After the mid-eighties, though, I began to do poorly, and, after suffering with the old methods for a few years, eventually moved on to statistical handicapping, which has worked a lot better for me.

I subsequently tried to analyze what had gone wrong. The essential problems I found with angle-based or situational handicapping are:

1) angles don't take the line into consideration. They tell you to play or don't play, period. This is clearly wrong. Will dave play this angle next time it comes up if it finds him a favorite giving 25 points? 50 points? Of course not. And in refusing to make such a play, the handicapper would be covertly acknowledging that his angle doesn't tell the whole handicapping story. That is, if the angle can't be configured in such a way as to produce a line that can then be compared to the book's line, then it can never address the most important handicapping question of all, that of value.

2) The relentless tightening that angle-shooters love to engage in typically reduces the number of trials to the point where the angle actually becomes worth less and less statistically even as the record improves. The 10-0 record dave cites would barely be significant in the unchanging world of the physical sciences; in sociological (i.e., "real world") applications, 10-0 is utterly meaningless. Most statistics texts that address this issue suggest that the z-score ( a measure of statistical reliability) for such real-world applications would have to be at least 5 or 6 to imply 99% reliability, and the z-score of 10-0 is only 3.2. The past is littered with dead angles that once seemed like they'd never lose; you may need to experience this with your wallet before you believe it completely, but ask yourself: could it possibly be this easy -- bet every team that meets such-and-such criteria and win? No way.

3) the game changes. Scoring 30 points in 2 straight games is a big deal when 20 points is the league average, but less so when the league average is 25. Playing in your third straight road game, or changing to grass after several weeks on turf, is tough, but less so when your coach has come to recognize the potential problem and prepares you for it. How can the angle shooter tell when the world has caught up to what he's doing? (and it has -- these two situations used to be big winners, but have sucked for many years now).

I know this is the type of post that draws flames -- "who the fvck do you think you are, I've used that system for years and I win every time, assh0le." Please don't waste our time. I'm not trying to put down anybody's way of 'capping, and I'm not saying anything bad about taxi. On the contrary, as I said, I'm pals with dave. He comped me to his newsletter. I've run many concepts he's developed through my DB. Everything is cool. I just think from hard experience that this is a very dangerous way to bet the game.

 

[loophole]

great discussion guys, a continuation of that age-old gambling dialectic in search of our avocation's holy grail - an angle with a high probability of success that generates enough events to pass a reliable test of statistical significance. thanks for sharing your thoughts.

 

[statman]

There IS a way to use angles that gets beyond the flawed approach of just tracking records vs the line. And, that is to track the record against an unbiased (for example power ratings based) prediction. It's crucical to do so... here's why (and NBA example).

Years ago, you could print money week after week playing against teams that played 4 games in five nights. But eventually, the linesmaker caught on and adjusted the opening numbers- removing the value. Here is the problem that created:

1) Angle cappers saw their w/l % go to 50% and stopped playing the method.
2) Pure stat players started seeing overlays on these games (since the lines were preadjusted for these "angles/situations") and started playing games that had no edge.
3) Teams playing 4 games in 5 nights still did underperform their power ratings, but no one knew what to do with them.

So, if you can, use a computer to make an unbiased stat based prediction on every game. Then, test your angles against that prediction and not the line. If there is a consistant result because of the angle (the teams fail to cover 70% of the time, by an average of 4.5 points...), then start to add these adjustments to your numbers.

INHO you cannot use numbers alone to win consistantly, nor angles... you must combine the two. Thoughts?

 

[GM]

Agreed. You will have trouble succeeding if you do not combine methods of handicapping. Numbers are only part of the story. Injuries, fatigue/travel, extra (or lack of) motivation, team discord, hot and cold streaks, matchups, look-aheads...all of it has to be examined to some degree (though I believe in some of those things a lot more than others).

[This message has been edited by GM (edited 11-02-2001).]