Performify's Pigskin Picks - Week Ten

[Performify]

Thremp,

Thanks for the suggestion. That's actually what I've always done.

I've got my original model that i've been using for 7 years, that I got from my original betting mentor who used the same model himself for ~5 years prior. Then i've got my "tweaked" model that I run side-by-side with the original, a model which I've made what I consider improvements over the years.

The difference sbetween the original and my tweaked model are usually very rare, mostly the results are on the strength of bets which allow me to (usually) better tweak my units, i.e. last year where I had a 48.5% record on total picks but a ~60% win rate on my game of the week which carried me.

So then anytime I consider a change to the model, I run a third model side-by-side, with whatever proposed changes. Last year in about week six I started a proposed change model to apply significantly less weight to underdogs (which was a MAJOR fundamental change), and am still evaluating that model for this year (i.e. i'm letting it make all its picks, and keeping track of them). its doing better than my current model right now, so i'm about one more week away from getting murdered to making the switch to using its picks exclusively. Either that, or flipping my picks and fading my system [img]/images/graemlins/grin.gif[/img]

I'll now consider making another tweak, but this shouldn't be a major one, pulling out the fumbles lost (and in fact i think i may pull out turnovers entirely in yet another model) and see if that changes things at all.

It really isn't that much work, it takes me a little bit of time to build the framework but the input files are the same for each model so taking the time to grab the data and input my own capping on the variable non-statistical inputs (and doing the research that I use to come up with those inputs) is what takes me the most time each week.

Of course, there's always the possiblity that my system is doing just fine and I'm messing up the human factor. My system is not a purely mathematical model, its really more of a framework, a structured way to include quantitative and qualitative data and a regular, repeatable approach to handicapping.

I have ten years of electronic records and four years of paper records for my own capping, and so far I'm on pace for my worst NFL season ever. But I do intend and expect that the numbers will start to turn around here shortly, and i expect that I can still turn a profit on the regular season season from here in fact.

/rambling

[zOrO2k6]

(Chiefs/Dolphins) Chiefs -1
(Broncos/Raiders) Under 33
(Rams/Seahawks) Seahawks -3.5
(Chargers/Bengals) Over 49
(Bears/Giants) Giants Pick
(Bills/Colts) Under 45
(Saints/Steelers) Steelers -4
(Jets/Patriots) Jets +10.5

any thougts?

[Punker]

[ QUOTE ]
Saints +4 (-102) at Steelers

This is a popular pick this week among some cappers I respect, and my system likes it a little as well. The Steelers are the worst team in the NFL at turning the ball over, at -11 on the year. The Saints aren't great in that category, sitting at -3 on the year, but still I expect the Steelers will not reverse themselves here and play mistake-free football. The Steelers have struggled protecting Roethlisburger all season, and the Saints match up well defensively with an aggressive pass rush. The Saints have allowed only one quarterback to top 250 yards this season, and I do not expect Big Ben to join that elusive club. The Saints defense has also only allowed two running back touchdowns all season, which should mean a rough day for Willie Parker.

On the other side of the ball, the Steelers have allowed some big games for opposing QBs this year, and the combination of Colston and Brees should be able to capitalize to keep things close. The Dynamic Denver Duo of Jake Plummer and Javon Walker tore up the Pittsburgh secondary last week, and you’ll see Brees making big throws between big target Marques Colston and speedster Devery Henderson. Unlike some others I don't think the Saints win here often enough to be worth a moneyline bet, but I do like taking the four points. 24-21 Pittsburgh for the Saints cover.


[/ QUOTE ]

You say 100% negative things about Pittsburgh here and 100% positive things about New Orleans; (paraphrasing) The Pitt offense can't protect the passer and the NO defense is great against the run. The NO offense is based on the pass, and Pitt can't defend the pass. I agree with all of this, btw.

However, this is pretty inconsistent with your final conclusion of "I don't think the Saints win here often enough to be worth a moneyline bet". After reading your preliminary thoughts, I wonder how you expect Pittsburgh will ever be winning this game?

[Performify]

I think New Orleans could certainly win this game, and I can't argue extensively with a moneyline play here. If this were a 3u play in my system it would be worth a 1u moneyline play.

But at a 1u play, I'm almost always going to take 4 points instead of a moneyline play. Its one of my fundamental tenets, if I don't have the team winning on a 1u play but just covering I'm going to take the points. If I have the team winning on a 1u play its discretionary. If the team is a 3u play then it becomes a good option for 2u on the spread and 1u on the moneyline or even 3u spread 1u moneyline.

Despite my writeup, I think there are positives for the Steelers in this game. Primarily, the Steelers rush defense versus the Saints rushing attack. I expect a very poor rushing day for the Saints, which will hurt them on their entire offensive attack by forcing them to go pass first.

I also think the Steelers will be able to get pressure on Brees as well.

The Steelers have some intangibles here, as they're going to be under a lot of pressure to win this game. But I think the Saints have the matchups (strong passing attack, speed receiver, a big receiver , a fast defensive front to outrun the slower offensive line) which should help them keep this close, if not win it.

[Punker]

[ QUOTE ]
Primarily, the Steelers rush defense versus the Saints rushing attack. I expect a very poor rushing day for the Saints, which will hurt them on their entire offensive attack by forcing them to go pass first.

[/ QUOTE ]

The Saints are currently 27th in the league in rushing. I don't think they go into many games expecting to have big rushing days, and always go pass first.

[ QUOTE ]
I also think the Steelers will be able to get pressure on Brees as well.


[/ QUOTE ]

Although sacks are not the be all and end all of QB pressure, it's at least something.

The Saints have given up 8 sacks through 8 games. Conversely, Pittsburgh has 23 sacks.

New Orleans has played 4 games against teams with 20 or more sacks this year; in those 4 games, they have given up 7 sacks (4 to Green Bay, 0 to Philly, 2 to Baltimore, 1 to Atlanta).

For Pittsburgh's 23 sacks, 12 came against two teams with obvious offensive line issues and/or immobile QBs: Cincy and Oakland (23 and 37 sacks allowed respectively). Take those out, and you have Pittsburgh with 11 sacks in 6 games. In fact, if you take the game against Oakland out, they have 3 sacks in their last 3 games (vs KC, Atlanta, Denver). I don't see any evidence they'll be able to get serious pressure on Brees - maybe 2 sacks at best.

[Performify]

[ QUOTE ]
(Chiefs/Dolphins) Chiefs -1
(Broncos/Raiders) Under 33
(Rams/Seahawks) Seahawks -3.5
(Chargers/Bengals) Over 49
(Bears/Giants) Giants Pick
(Bills/Colts) Under 45
(Saints/Steelers) Steelers -4
(Jets/Patriots) Jets +10.5

any thougts?

[/ QUOTE ]

I'm on the other side of the Saints/Steelers obviously, per my picks.

This probably isn't what you want to hear, but if what you're doing here are eight-team parlays, you know that's such a tremendous longshot that you're pretty much buying lottery tickets...

[easternbloc]

do you have any thoughts on the CHI/NYG game? the line started at CHI +2.5 and has moved to PK.

[Performify]

I'll be back tomorrow with some thoughts one potential one unit plays.

Chicago falls in that borderline 1u play area in my system, highly dependant upon my handicapping inputs, but its something I'd probably stay away from, because there are so many unknowns on both sides - primarily the injuries and the effect they'll have.

I did grab some of it in a teaser at Chi +8.5 early though.

I think the Giants are going to be in a bit of trouble with their injuries on defense, but I also think Chicago's offense will struggle without Berrian. Its going to come down to Chicago's ability to run the ball versus the Giants depleted defense, and whether or not Chicago's defense can perform at the highest level and really keep the Giants offense in control...

[NanaAnna]

It seems your system really loves road teams, I know some of your big unit bets have been road teams and have lost recently, so I hope my thoughts are all wrong and you win, but here are some things you may want to think about with regard to your Jets pick.

Brady has had monster games after having 3 and 4 Int' games.

Belichick's Pats have historically crushed teams the second time they see them in the same year. And those numbers are even more impressive when he sees a QB for the 2nd time in a year.

The Pats also play great after a loss, just look at how well they played in Cincy after looking anemic in their 17-7 loss to Denver this year.

You say the Jets have a high powered passing offense, but their passing offense ranks 23rd in the NFL in yards/gm.

I hope I'm wrong but this game just screams a total blowout by the Pats.

Good luck,

Anna

[dixieflatline]

Some really great info in your posts here Performify. As a noob starting out I can only hope to have an advanced system like yours up and running in a few years.

I know the dangers of tuning a system to previous year's data, but here if you were interested in testing your system's reliance on fumbles lost/recovered I think you could do that. If you took a handful of teams that were very lucky/unlucky recovering fumbles and reran those lines with different weights on the fumbles it should come out basically flat.

If your system did come out flat that would provide a consistency check and that should give you more confidence in your system. If it wasn't flat then you have a problem but the curve would give you a clue to how much error it might be causing. That would be an interesting test but maybe too time consuming.

Anyway, the next thing would be to determine if the actual line is affected by these lucky/unlucky fumbles. If it is altering the lines a lot then that could be an important ineffiency in the market.

I do have some more questions about your system but as you have already revealed a bunch about it don't answer them if you don't feel comfortable.

1) You mentioned that you enter statistical input and then input your own capping. Does your capping include things that the statistical input doesn't know like injuries, weather, etc? Do you add that in as a big lump sum and just say something like add three points to this line, or do you let the model merge the two in a more complex way?

2) Do you use the current and/or starting line as inputs? Is the output what the model thinks is the true line or something different?

3) You mention several changes to the model in the past. Are you often tweaking it and running several test models on the side? Do you ever go back and rerun a tweaked model on previous games?

4) Does you model just provide info on the line or does it spit out an over/under as well? Do you run the output through any statistical programs to check for significance or correlations?