2014 Subscriber Contest is LIVE (2 Viewers)

[BassNBrew]

I don't Dodds projections are the way to go. Historically the line moves around 8-10 points on Monday night alone. Sunday night generally features bigger name teams that would have higher ownership.

That's possible, but that would be accounted for in my calculation since I'm running it on the actual rosters. I could just add 8-10 points and call it a day but I'm trying to account for the actual ownership distribution, etc. If SNF features a bunch of heavily-owned players, that will be captured in the projection.

Last night the Sunday game moved the line more than Dodds projected it to move over two games.

Sure, Dodds wasn't predicting a 43-17 game. Vegas had it at like 23-22. Any time a lot of points are scored, it's going to move the cutline more than was predicted.

I thinning the Dodds fractional TD numbers becoming real 6 pointers is the weak spot.

Well some of those fractional TD numbers become zeroes, too. If Dodds gets the overall team output close to the actual thing, then the fractional stuff should mostly average out. The problem with last night is that the Patriots outproduced what Dodds had them projected for. I don't think that's a failing of the method.

Simplified example but let's say coming into Monday night we have 10 teams in the contest and three are getting cut.

A - 210

B - 180

C - 175

D - 170

E - 160

F - 155

G - 140

H - 128 + Garcon

I - 128 + DJax

15 K Guy - 100 + Lynch

Let's say Dodds has both Garcon and Djax projected at 5-50-.5. Your predicted cut-off will be 141. The reality is that one of those fractional will become a one, the other will become a zero. That means the cutoff will become 144 in reality.

If you played the game 10 times Dodds numbers would likely average out to his projection. His projections smooth out the variances, but a one game sample doesn't do this. Even if the game plays out exactly like he predicts in terms of yards and TDs, the naturally occurring variances will move the cutoff line more. Of course I'm ignoring distribution of the players in this example but over the course of the year this will balance out.

I think if you keep messing with this you'll come up with a formula of Dodds plus x number of points.

 

[cheese]

I'm wondering how much longer I'll be alive in this thing with both AP and Ray Rice stinking the joint up all year long. 4 weeks in, and still kicking somehow. lol.

Same and make it 5 now for me. Benny Cunningham, Alfred Blue, and Fred Jackson are my remaining RBs

 

[QuizGuy66]

Average ownership of tonight's players is low. Average entry only has .856 guys going tonight.

-QG

I have at least 11 guys. The Redskins defense and special teams

That's debatable

-QG

 

[joffer]

Survive on that Wilson TD

 

[Saint]

90.75 ... think I'm leaving this week...

Still have Wilson (-9), White (-0), Sanu (-6.8), Wright (-7.8)

Don't think it's enough.

Yes, Virginia, there is a santa claus.... and apparently he had Wilson, Sanu, and Wright..... so I have slowly edged above the cutline and going to make it... just delaying the inevitable with only 1 RB as I keep taking a 0 for my second RB slot....

 

[Wadsworth]

It is always a bad week when you get knocked out of the FBG contest. That is this week for me. Good luck folks, I hope one of the regular posters in this thread wins it.

 

[QuizGuy66]

It is always a bad week when you get knocked out of the FBG contest. That is this week for me. Good luck folks, I hope one of the regular posters in this thread wins it.

:( Thanks for the well wishes - sorry you're out.

-QG

 

[PatsWillWin]

Wow. I scored 140.25. Cut-off is 140.05.

 

[FUBAR]

when does the o.ffense site update the query? Looked to see how many Graham owners got booted this week, looks like zero but that likely isn't right.

 

[Pots]

Wow. I scored 140.25. Cut-off is 140.05.

How do people know this?

 

[apalmer]

Wow. I scored 140.25. Cut-off is 140.05.

How do people know this?

Iggy's website: http://o.ffense.com/fbg35k/index.php

 

[Ignoratio Elenchi]

when does the o.ffense site update the query? Looked to see how many Graham owners got booted this week, looks like zero but that likely isn't right.

After FBG posts the official results on Tuesday morning, I run a couple of things to reconcile my results with theirs, and then once everything matches up I update the queriers. I haven't fully automated this process yet, so it's somewhat dependent on when I get to it - usually I can do it pretty quickly but the FBG results haven't posted yet this morning and I have a few meetings in the office so it might take a little longer today.

 

[Ignoratio Elenchi]

Unofficially, 598 Jimmy Graham owners were eliminated this week (out of 2,658 that came into week 5 alive).

 

[FUBAR]

when does the o.ffense site update the query? Looked to see how many Graham owners got booted this week, looks like zero but that likely isn't right.

After FBG posts the official results on Tuesday morning, I run a couple of things to reconcile my results with theirs, and then once everything matches up I update the queriers. I haven't fully automated this process yet, so it's somewhat dependent on when I get to it - usually I can do it pretty quickly but the FBG results haven't posted yet this morning and I have a few meetings in the office so it might take a little longer today.

Thanks! That's a lot of work but many of us love what you're doing.

 

[Ignoratio Elenchi]

when does the o.ffense site update the query? Looked to see how many Graham owners got booted this week, looks like zero but that likely isn't right.

After FBG posts the official results on Tuesday morning, I run a couple of things to reconcile my results with theirs, and then once everything matches up I update the queriers. I haven't fully automated this process yet, so it's somewhat dependent on when I get to it - usually I can do it pretty quickly but the FBG results haven't posted yet this morning and I have a few meetings in the office so it might take a little longer today.

Thanks! That's a lot of work but many of us love what you're doing.

It's a labor of love.

I can set up a process to run on Tuesday mornings and update everything automatically, I just haven't sat down and done it yet (I usually really want to do it on Tuesday morning and then my interest wanes as the week goes on. ) Plus, at least for the first few weeks, I wanted to manually review any discrepancies between my results and FBG's to look for errors in the code, etc.

Usually the FBG results are posted by now, but it looks like they're still not up yet, so it may be a few hours before I get around to updating my site with the official results.

 

[bicycle_seat_sniffer]

FBG isnt updated yet?

 

[KellysHeroes]

lost Ball, Stewart and Gerhart and only got Vereen and Blount left. :(

 

[BassNBrew]

Iggy - Are you tracking the owners you saved by catching that scoring mistake a couple of weeks ago?

 

[Ignoratio Elenchi]

Two discrepancies this week:

• FBG has Shayne Graham with 10.4 points. I have him at 12.4. It looks like he had 3 FGs (29, 30, 44) and 2 PATs. That adds up to 12.4 for me, am I missing something obvious, or should I email Drinen?
• FBG has Falcons D with 3 points. I have them at 5. I assume the difference is a play where ATL was on offense, Ryan threw an interception, and then during the interception return the defender fumbled and it was recovered by ATL. My site was counting this as a fumble recovery for Atlanta DST but I suppose it isn't, so I'll correct that.

I'll hold off on updating the official results until I hear back from Drinen.

 

[Dinsy Ejotuz]

Consistent

140.05

145.30

153.60

149.05

147.40

 

[Sinn Fein]

Interesting (to me) fact: I have Cutler and Romo as QBs. Cutler has counted 4 out of the 5 games. But, I would have survived every week if I only had Romo on the roster - and would have been eliminated in week 4 if I only had Cutler on my roster.

 

[Gawain]

Man, Graham, McCoy and Megatron sure sounded like a good idea at the time. Good luck to you folks moving onward.

 

[Lehigh98]

Unofficially, 598 Jimmy Graham owners were eliminated this week (out of 2,658 that came into week 5 alive).

More next week too with him on bye, would have rather had Tim Wright's 22 pts next week, he's my only TE backup to Graham,

That's what I get for not paying closer attention to the rules and TE scoring.

Colin Kaepernick $14 19.15 20.00 21.65 23.70 15.85 bye

Ben Roethlisberger $10 22.05 9.85 17.90 27.50 17.45 bye

Johnny Manziel $8 0.00 0.00 0.00 0.00 0.00

Brian Hoyer $6 15.30 14.20 18.90 0.00 25.80

Teddy Bridgewater $6 0.00 0.00 10.20 24.55 0.00 bye

Blake Bortles $3 0.00 0.00 20.15 17.05 8.95 bye

---------------------------------------------------------------------------------------------------------------------------------------------

Pierre Thomas $17 14.90 6.20 14.10 4.30 31.20 bye

Darren Sproles $16 18.50 30.80 8.00 8.60 5.10 bye

Andre Williams $7 0.90 3.90 1.80 12.60 16.30 bye

Dexter McCluster $7 5.60 4.90 3.50 0.10 1.80 bye

James Starks $5 6.80 0.00 3.80 0.00 6.80 bye

Jonathan Stewart $5 6.70 12.90 8.00 0.00 0.00 out? bye

Benny Cunningham $3 9.10 2.90 4.40 0.00 16.10

---------------------------------------------------------------------------------------------------------------------------------------------

Dez Bryant $27 9.50 26.30 20.90 13.40 23.50 bye

Cordarrelle Patterson $18 21.80 9.60 9.40 5.80 2.80 bye

Jordan Matthews $9 5.70 2.70 25.90 6.80 7.50 bye

Justin Hunter $9 9.30 4.60 6.70 2.20 18.90 bye

Miles Austin $8 4.00 16.40 17.10 0.00 7.40

Steve Smith $7 24.80 13.10 15.10 32.90 8.40 bye

Cody Latimer $6 0.00 0.00 0.00 0.00 0.00

---------------------------------------------------------------------------------------------------------------------------------------------

Jimmy Graham $30 20.20 38.80 14.40 26.60 6.60 bye

Timothy Wright $5 6.00 0.00 3.50 0.00 22.00 bye

---------------------------------------------------------------------------------------------------------------------------------------------

Josh Scobee $3 6.90 4.60 6.10 2.00 11.40 bye

Caleb Sturgis $3 15.80 4.40 9.10 9.10 0.00

Mike Nugent $3 19.30 6.10 7.00 0.00 5.00

Adam Vinatieri $3 6.00 10.60 17.10 11.10 9.20 bye

---------------------------------------------------------------------------------------------------------------------------------------------

Miami Dolphins $3 8.00 0.00 11.00 16.00 0.00

Detroit Lions $3 6.00 4.00 12.00 6.00 10.00 bye

Pittsburgh Steelers $3 3.00 0.00 13.00 3.00 11.00 bye

New York Jets $3 2.00 6.00 6.00 4.00 5.00 bye

 

[FUBAR]

Consistent

140.05

145.30

153.60

149.05

147.40

167.90

184.50

176.00

169.05

166.40

Don't like the downward trend.

Stedman Bailey is now the only player who hasn't counted for me.

 

[Ignoratio Elenchi]

I checked with Doug - he's updating the score for Shayne Graham in the FBG results, and we agreed that his result for the Falcons DST is correct.

So both his and my results have been updated accordingly. Everything on my site should now be current through week 5, let me know if you have any questions.

Size	Count	Alive	Survival Rate18	5667	1967	34.70%19	1922	796	41.40%20	1519	693	45.60%21	1176	586	49.80%22	1015	525	51.70%23	744	399	53.60%24	542	315	58.10%25	459	252	54.90%26	331	179	54.10%27	253	157	62.10%28	219	133	60.70%29	177	107	60.50%30	279	162	58.10%

 

[shader]

I'm doing well...but the CJ Spiller selection is going to rear it's ugly head before too long.

 

[One More Rep]

I was on the wrong side of the cut line yesterday morning. Had Russell last night and he placed me on the right side

 

[cincy11]

Consistent

140.05

145.30

153.60

149.05

147.40

167.90

184.50

176.00

169.05

166.40

Don't like the downward trend.

Stedman Bailey is now the only player who hasn't counted for me.

TOTAL 167.35 167.40 173.65 173.60 169.95

Justin Hunter finally counted for my team this week.

Not used:

Andy Dalton $13

Blake Bortles $3

Danny Woodhead $13

Austin Seferian-Jenkins $5

Brandon Pettigrew $3

Nate Freese $3

New York Jets $3

 

[solorca]

I made it through this week with about 40 points to spare, but I'm on life support with my RB situation. If Rashad Jennings misses, as expected, I'm down to Shane Vereen, MJD and James Starks as my only healty back.

I've managed to use all of my players at this point other than MJD and James Starks, thanks to Justin Hunter, Tim Wright, and the Steelers defense contributing this week.

 

[BassNBrew]

I checked with Doug - he's updating the score for Shayne Graham in the FBG results, and we agreed that his result for the Falcons DST is correct.

So both his and my results have been updated accordingly. Everything on my site should now be current through week 5, let me know if you have any questions.

Size Count Alive Survival Rate18 5667 1967 34.70%19 1922 796 41.40%20 1519 693 45.60%21 1176 586 49.80%22 1015 525 51.70%23 744 399 53.60%24 542 315 58.10%25 459 252 54.90%26 331 179 54.10%27 253 157 62.10%28 219 133 60.70%29 177 107 60.50%30 279 162 58.10%

I really hope they are at least comping you a subscription based on all the errors you have caught.

 

[Gottabesweet]

No Graham or Pierre Thomas this week likely gone.

My RB's are Cunningham, Starks and MJD this week.

TE's are Rivera and Chandler.

 

[davidwb]

21 players and the only one that hasn't been used yet is Tony Romo (Pitt D was used this week -- they were the only other one that hadn't been used yet). Only 1 player used in all of the first 5 weeks: Fred Jackson (although Marshawn Lynch has been used in all 4 of the weeks Seattle has played). Alive by 11 points!

If I'm still alive in week 9, I know I'll be using Romo then because both Rodgers and Cutler will be on bye. That is the only time I'll ever root for the Cowboys to score a lot of points.

 

[Modog814]

I don't Dodds projections are the way to go. Historically the line moves around 8-10 points on Monday night alone. Sunday night generally features bigger name teams that would have higher ownership.

That's possible, but that would be accounted for in my calculation since I'm running it on the actual rosters. I could just add 8-10 points and call it a day but I'm trying to account for the actual ownership distribution, etc. If SNF features a bunch of heavily-owned players, that will be captured in the projection.

Last night the Sunday game moved the line more than Dodds projected it to move over two games.

Sure, Dodds wasn't predicting a 43-17 game. Vegas had it at like 23-22. Any time a lot of points are scored, it's going to move the cutline more than was predicted.

I thinning the Dodds fractional TD numbers becoming real 6 pointers is the weak spot.

Well some of those fractional TD numbers become zeroes, too. If Dodds gets the overall team output close to the actual thing, then the fractional stuff should mostly average out. The problem with last night is that the Patriots outproduced what Dodds had them projected for. I don't think that's a failing of the method.

Simplified example but let's say coming into Monday night we have 10 teams in the contest and three are getting cut.

A - 210

B - 180

C - 175

D - 170

E - 160

F - 155

G - 140

H - 128 + Garcon

I - 128 + DJax

15 K Guy - 100 + Lynch

Let's say Dodds has both Garcon and Djax projected at 5-50-.5. Your predicted cut-off will be 141. The reality is that one of those fractional will become a one, the other will become a zero. That means the cutoff will become 144 in reality.

If you played the game 10 times Dodds numbers would likely average out to his projection. His projections smooth out the variances, but a one game sample doesn't do this. Even if the game plays out exactly like he predicts in terms of yards and TDs, the naturally occurring variances will move the cutoff line more. Of course I'm ignoring distribution of the players in this example but over the course of the year this will balance out.

I think if you keep messing with this you'll come up with a formula of Dodds plus x number of points.

Or they both go to 0 and the cutoff goes to 140.

 

[CalBear]

I don't Dodds projections are the way to go. Historically the line moves around 8-10 points on Monday night alone. Sunday night generally features bigger name teams that would have higher ownership.

That's possible, but that would be accounted for in my calculation since I'm running it on the actual rosters. I could just add 8-10 points and call it a day but I'm trying to account for the actual ownership distribution, etc. If SNF features a bunch of heavily-owned players, that will be captured in the projection.

Last night the Sunday game moved the line more than Dodds projected it to move over two games.

Sure, Dodds wasn't predicting a 43-17 game. Vegas had it at like 23-22. Any time a lot of points are scored, it's going to move the cutline more than was predicted.

I thinning the Dodds fractional TD numbers becoming real 6 pointers is the weak spot.

Well some of those fractional TD numbers become zeroes, too. If Dodds gets the overall team output close to the actual thing, then the fractional stuff should mostly average out. The problem with last night is that the Patriots outproduced what Dodds had them projected for. I don't think that's a failing of the method.

Simplified example but let's say coming into Monday night we have 10 teams in the contest and three are getting cut.

A - 210

B - 180

C - 175

D - 170

E - 160

F - 155

G - 140

H - 128 + Garcon

I - 128 + DJax

15 K Guy - 100 + Lynch

Let's say Dodds has both Garcon and Djax projected at 5-50-.5. Your predicted cut-off will be 141. The reality is that one of those fractional will become a one, the other will become a zero. That means the cutoff will become 144 in reality.

If you played the game 10 times Dodds numbers would likely average out to his projection. His projections smooth out the variances, but a one game sample doesn't do this. Even if the game plays out exactly like he predicts in terms of yards and TDs, the naturally occurring variances will move the cutoff line more. Of course I'm ignoring distribution of the players in this example but over the course of the year this will balance out.

I think if you keep messing with this you'll come up with a formula of Dodds plus x number of points.

Or they both go to 0 and the cutoff goes to 140.

That's true; the expected probability for two .5 chances is .75, not 1.

 

[Ignoratio Elenchi]

I don't Dodds projections are the way to go. Historically the line moves around 8-10 points on Monday night alone. Sunday night generally features bigger name teams that would have higher ownership.

That's possible, but that would be accounted for in my calculation since I'm running it on the actual rosters. I could just add 8-10 points and call it a day but I'm trying to account for the actual ownership distribution, etc. If SNF features a bunch of heavily-owned players, that will be captured in the projection.

Last night the Sunday game moved the line more than Dodds projected it to move over two games.

Sure, Dodds wasn't predicting a 43-17 game. Vegas had it at like 23-22. Any time a lot of points are scored, it's going to move the cutline more than was predicted.

I thinning the Dodds fractional TD numbers becoming real 6 pointers is the weak spot.

Well some of those fractional TD numbers become zeroes, too. If Dodds gets the overall team output close to the actual thing, then the fractional stuff should mostly average out. The problem with last night is that the Patriots outproduced what Dodds had them projected for. I don't think that's a failing of the method.

Simplified example but let's say coming into Monday night we have 10 teams in the contest and three are getting cut.

A - 210

B - 180

C - 175

D - 170

E - 160

F - 155

G - 140

H - 128 + Garcon

I - 128 + DJax

15 K Guy - 100 + Lynch

Let's say Dodds has both Garcon and Djax projected at 5-50-.5. Your predicted cut-off will be 141. The reality is that one of those fractional will become a one, the other will become a zero. That means the cutoff will become 144 in reality.

If you played the game 10 times Dodds numbers would likely average out to his projection. His projections smooth out the variances, but a one game sample doesn't do this. Even if the game plays out exactly like he predicts in terms of yards and TDs, the naturally occurring variances will move the cutoff line more. Of course I'm ignoring distribution of the players in this example but over the course of the year this will balance out.

I think if you keep messing with this you'll come up with a formula of Dodds plus x number of points.

Or they both go to 0 and the cutoff goes to 140.

Right, but then Dodds's aggregate projections will have been wrong (he'd have predicted the Redskins to have 1 receiving TD, and they actually had 0).

BNB is saying that even if Dodds's aggregate team projections are right on the money, the fractional stuff will still cause the projected cutoff to be understated. I have to think about it a little but he may be right about that, in which case I'll have to fudge the projection in some way to account for that effect.

 

[CalBear]

Used every week: Antonio Brown ($22), Travis Kelce ($8)

Not used: Deangelo Williams ($10), Deter McCluster ($7), Rod Streater ($7), Jerricho Cotchery ($4), Malcolm Floyd (surprisingly, $4), Lance Moore ($3), Jordan Cameron ($22)

Of the unused players, Floyd and Cameron are likely to contribute, and possibly Williams later on. Moore or Cotchery may have a lucky week. McCluster looks like a mistake. Streater was a decent pick but won't help after his injury.

Of the used players, I'm now a two-kicker team without Freese, but everyone else at least has some possibility of producing.

The roster's ticking over pretty well, haven't been below 160 yet. Floyd has four double-digit scores but I haven't used any of them because my WR corps is pretty solid. Gonna lose a little bit of production when Marvin Jones comes back (used Sanu 3 times already). Ingram returning would be nice; without him and Deangelo my RBs are thin (I actually used Shonn Green), although Andre Williams should be able to fill in for a few weeks.

 

[jdkapow]

Awards and notable teams of the week. Note that these awards are based on the NFL's Jeremy Maclin correction, which isn't reflected in the FBG report yet.

Strongest Living Team: RHON, Entry 100300, took a little dip into the 170s this week but he's still on top overall, averaging over 203 points per week. Entry 102658, Metropolis Shark (Entry 107887) and Timmah (Entry 105547) are the only other teams averaging 200+.

Strongest Dead Team: Entry 105074, averaging 191 per week, in 76th place overall. Remember that number...

Keep Your Day Job Award: Entry 100869 has a stranglehold here, with an average score of 79, over 6.5 points below the next-worst team. (This is the 15-QB team, by the way).

Safest Team: A new team here: Irish Bombers (Entry 110328) has cleared the cutline by over 50 points every week. AdamAfrica, who used to own this category, dropped into the 170s this week.

Just Skating By Award: Bloody Nine (Entry 107224) is just scraping by, never clearing the cutline by more than 8.1 points. This week, they made it by 0.15.

Icarus Award: Here's that number again: Entry 105074 was in 7th place overall, averaging over 205 points per week, before missing the cutline by 3 points this week.

Woulda Coulda Shoulda Award: Entry 100301 was eliminated in Week 1 but has averaged 195 since then.

Notable Teams:

QBs: All the teams with double-digit QBs are now dead. hankd (Entry 103966) is the only team with 9 QBs who's still around.

RBs: Same as last week: three of the original six 12-RB teams are still around.

WRs: We're down to one 13-WR team, Zorro (Entry 102255).

TEs: Same as the last two weeks: three teams with 7 TEs survive.

PKs: Say goodbye to 15-kicker guy. Say hello to 12-kicker guy! Entry 111066, come on down! After 12-kicker guy, it's a long drop to 8-kicker guy, which somehow doesn't generate the same excitement.

DEFs: The last team with 8 defenses, Entry 113411, is still with us.

And me, you ask? With another strong week, I moved up almost 500 places and am currently sitting in 837th place among living teams.

More to come later...

 

[ochawkeye]

See a few posts noting how many players they haven't used. Would be interested in the inverse. The Ironman team that has used the fewest number of point scorers relying on the same core set of players every week while still scoring enough to survive.

 

[jdkapow]

And now the players:

Eliminated Players: Still none. Dion Lewis is getting close: only 1 of his original 3 owners is still alive.

The Biggest Mistake: A couple contenders here. Matt Prater is down to an 11.6% survival rate (30/258, $7), while Calvin Johnson is at 22.8% (360/1578, $31).

The Survivor: Still DeMarco Murray. 74% of teams with Murray are still alive.

The Bargain: Brian Quick: he cost just $4 and he has the second-highest survival percentage at 69.4%. His 99 surviving owners are quite pleased. Alternatively, Mohamed Sanu, who cost even less at $3, has a 63% survival rate (4th overall), and is owned by over 1250 living teams.

The Parasite: Going with Mychal Rivera here: 26 total points and yet 8th in survival rate, over 60%.

The Martyr: Last week's guy, Justin Forsett, came up huge this week and lost no owners. I'm going with Matthew Stafford here: over 100 points, his worst week isn't so terrible, and yet 70% of his teams have died.

 

[Ignoratio Elenchi]

Or they both go to 0 and the cutoff goes to 140.

That's true; the expected probability for two .5 chances is .75, not 1.

I think this is a matter of interpretation. Let's say you have 2 WRs, each projected for 0.5 TDs. That doesn't really mean they each have a 50% chance of getting 1 TD, and a 50% chance of getting 0 TDs. It's more the expected value from a PMF (e.g. each WR has a 70% chance of 0 TDs, a 15% chance of 1 TD, a 10% chance of 2 TDs, and a 5% chance of 3 TDs...) In that case, the probability that they score (exactly) 1 TD between them is only like 20%. (Not to mention that these aren't independent probabilities, so even that's not right, you'd really have some kind of joint PMF from which you'd derive the individual marginal probabilities, but I digress...)

I treat Dodds's projections not so much as multiplicative probabilities, and more as additive "shares" of an offense. Dodds might project a team to have 250 passing yards and 2 TDs. The QB will probably be projected for all of the passing stats, and those passing stats will be divided up among the WRs. Maybe WR A is projected for 100 yds and 1.2 TDs, WR B is projected for 80 yards and 0.6 TDs, and WR C is projected for 70 yards and 0.2 TDs.

BNB's point is that even if Dodds is exactly right and the TEAM has a total of 250 passing yards and 2 passing TDs, the distribution of those stats among the receivers won't match the projections, so someone is going to end up with more points than they're projected for (either WR A got 2 TDs instead of 1.2, or WR B got 1-2 TDs instead of 0.6, etc.) which will cause the actual cutline to be higher than that which was projected by the stats.

I'm not 100% on board with that thought yet, but I'll give it some thought later.

 

[Modog814]

Right, but then Dodds's aggregate projections will have been wrong (he'd have predicted the Redskins to have 1 receiving TD, and they actually had 0).

BNB is saying that even if Dodds's aggregate team projections are right on the money, the fractional stuff will still cause the projected cutoff to be understated. I have to think about it a little but he may be right about that, in which case I'll have to fudge the projection in some way to account for that effect.

It could have gone to someone irrelevant (D.Young). And you're still at 140. Yes, the fraction stuff can cause the stuff to off, but it could be off either way (over or under). If Dodds projects a QB to throw 1.5 TD's, it's going to off, but it could be .5 TD too high or .5 TD too low. In a 1 game sample, its going to be off, but as an "expected" cut line, you want to factor 1.5 (if using Dodds as a best estimate), which is basically say we expect him to throw 1 or 2 TD's.

If you add a flat amount after putting in Dodds projections, you're pretty much certain to overstate the cut line.

 

[Modog814]

BNB's point is that even if Dodds is exactly right and the TEAM has a total of 250 passing yards and 2 passing TDs, the distribution of those stats among the receivers won't match the projections, so someone is going to end up with more points than they're projected for (either WR A got 2 TDs instead of 1.2, or WR B got 1-2 TDs instead of 0.6, etc.) which will cause the actual cutline to be higher than that which was projected by the stats.

I'm not 100% on board with that thought yet, but I'll give it some thought later.

Just to reiterate my point, WR A could have gotten 1 TD and a 3rd irrelevant player could have gotten the other, resulting the the cut line being lower than projected (WRA and B are both under TD expectations and the other player was irrelevant)

 

[Ignoratio Elenchi]

In a 1 game sample, its going to be off, but as an "expected" cut line, you want to factor 1.5 (if using Dodds as a best estimate), which is basically say we expect him to throw 1 or 2 TD's.

This was my initial instinct and I think it's how I still feel. The impact of fractional projections probably pales in comparison to the fact that Dodds can't predict the future perfectly anyway. Worrying about the fractions ignores the fact that, however good Dodds's projections may be, reality's going to look very different from the projections (e.g. the reason this week's post-SNF cutline was a few points higher than the pre-SNF projection wasn't because of fractions, it was probably because the Patriots scored like 20 more points than they were projected to).

And while BNB's example illustrated how the fractional projections could theoretically have an effect, it was oversimplified, and I'm not sure the effect would hold in the context of thousands of entries with varying combinations of all the different players going in the games.

Anyway, If Dodds consistently underpredicts offensive output or something, I could see a case made for fudging upwards, but I have no idea if that's the case, and without knowing otherwise, just plugging in his projections and running the cutoff calculation seems like the best (simple) method of coming up with a projected cutline. If anyone feels that this method will consistently understate the actual cutoff, then just mentally add a few points to it.

 

[ochawkeye]

Anyway, If Dodds consistently underpredicts offensive output or something, I could see a case made for fudging upwards, but I have no idea if that's the case, and without knowing otherwise, just plugging in his projections and running the cutoff calculation seems like the best (simple) method of coming up with a projected cutline.

IE,

I think what everyone wants is for you to create your own accurate projections instead of the projections from the site that we all pay to subscribe to.

Get to work.

 

[BassNBrew]

In a 1 game sample, its going to be off, but as an "expected" cut line, you want to factor 1.5 (if using Dodds as a best estimate), which is basically say we expect him to throw 1 or 2 TD's.

This was my initial instinct and I think it's how I still feel. The impact of fractional projections probably pales in comparison to the fact that Dodds can't predict the future perfectly anyway. Worrying about the fractions ignores the fact that, however good Dodds's projections may be, reality's going to look very different from the projections (e.g. the reason this week's post-SNF cutline was a few points higher than the pre-SNF projection wasn't because of fractions, it was probably because the Patriots scored like 20 more points than they were projected to).

And while BNB's example illustrated how the fractional projections could theoretically have an effect, it was oversimplified, and I'm not sure the effect would hold in the context of thousands of entries with varying combinations of all the different players going in the games.

Anyway, If Dodds consistently underpredicts offensive output or something, I could see a case made for fudging upwards, but I have no idea if that's the case, and without knowing otherwise, just plugging in his projections and running the cutoff calculation seems like the best (simple) method of coming up with a projected cutline. If anyone feels that this method will consistently understate the actual cutoff, then just mentally add a few points to it.

Here's another over simplified example for you to dwell on. Cutline is 150. 4000 teams at or above the cutline. 1000 teams precisely 1 point below the cutoff line and all 1000 have a WR who scored 15 pts currently counting as their lowest replaceable score. 500 of those owner have Sanders, the other 500 have Thomas. Dodds has both Thomas and Sanders projected at 5-70-.5. Assume Dodds is 100% in predicting Thomas+Sanders yields 12-140-1.

Based on this, when you run the numbers, the cutline doesn't move 100% of the time. The reality is that Sanders and Thomas are going to vary above and below that theoretical 5-70-.5. The further away the variance is from the projection, the more the cutline will move. Regardless, the cutline will move 99% of the time.

Obviously super simplified, but you can see where I'm going this. I think you will find Dodd's projections will fail you over time predicting the cutline and the biggest driver in how much they fail you will be the number of NFL players remaining to play. When there aren't many players remaining like last night the prediction won't be off by much. When you have a huge number of players remaining it will lag the actual outcome.

 

[BassNBrew]

Anyway, If Dodds consistently underpredicts offensive output or something, I could see a case made for fudging upwards, but I have no idea if that's the case, and without knowing otherwise, just plugging in his projections and running the cutoff calculation seems like the best (simple) method of coming up with a projected cutline.

IE,

I think what everyone wants is for you to create your own accurate projections instead of the projections from the site that we all pay to subscribe to.

Get to work.

Here's the challenge for Iggy…run a Monte Carlo simulation based on Dodds projections. We could get a better cutline estimate and estimates of the probability of the cutoff being higher or lower by x number of points.

 

[dagman]

OK. How many guys are left that have put up 5 surviving weeks in a row that have AP and Rice? This is ridiculous. I thought I'd be WAAY out by now.

http://o.ffense.com/fbg35k/index.php?teamid=113666

I thought I'd be done by week 4, turns out not so much. Now I'm following but simply for comedic value!

I can do better, Peterson, Ingram, D. Williams, J. Stewart, Lacy.

 

[dagman]

Awards and notable teams of the week. Note that these awards are based on the NFL's Jeremy Maclin correction, which isn't reflected in the FBG report yet.

Strongest Living Team: RHON, Entry 100300, took a little dip into the 170s this week but he's still on top overall, averaging over 203 points per week. Entry 102658, Metropolis Shark (Entry 107887) and Timmah (Entry 105547) are the only other teams averaging 200+.

Strongest Dead Team: Entry 105074, averaging 191 per week, in 76th place overall. Remember that number...

Keep Your Day Job Award: Entry 100869 has a stranglehold here, with an average score of 79, over 6.5 points below the next-worst team. (This is the 15-QB team, by the way).

Safest Team: A new team here: Irish Bombers (Entry 110328) has cleared the cutline by over 50 points every week. AdamAfrica, who used to own this category, dropped into the 170s this week.

Just Skating By Award: Bloody Nine (Entry 107224) is just scraping by, never clearing the cutline by more than 8.1 points. This week, they made it by 0.15.

Icarus Award: Here's that number again: Entry 105074 was in 7th place overall, averaging over 205 points per week, before missing the cutline by 3 points this week.

Woulda Coulda Shoulda Award: Entry 100301 was eliminated in Week 1 but has averaged 195 since then.

Notable Teams:

QBs: All the teams with double-digit QBs are now dead. hankd (Entry 103966) is the only team with 9 QBs who's still around.

RBs: Same as last week: three of the original six 12-RB teams are still around.

WRs: We're down to one 13-WR team, Zorro (Entry 102255).

TEs: Same as the last two weeks: three teams with 7 TEs survive.

PKs: Say goodbye to 15-kicker guy. Say hello to 12-kicker guy! Entry 111066, come on down! After 12-kicker guy, it's a long drop to 8-kicker guy, which somehow doesn't generate the same excitement.

DEFs: The last team with 8 defenses, Entry 113411, is still with us.

And me, you ask? With another strong week, I moved up almost 500 places and am currently sitting in 837th place among living teams.

More to come later...

I think I'm still close to the ROLLER COASTER REWARD.....

192.00 140.70 165.90 140.15 175.20

CUTOFF 138.15 140.20 134.65 137.60 140.05

If form holds, it's going to be another nail biting week.

 

[dbc925]

Beat the cutline by only 8.4, but still beat it. FLuck! *

*my weekly post disparaging Andrew Luck's INT in week 2, costing Adam Vinatieri an easy FG and knocking me out. Will continue to post "FLuck" weekly until I miss the cut again.

 

[SeniorVBDStudent]

By eyeball method the weekly cutoff is constant at 140 approx. Based on number of enyries at what time does the cutoff start migrating upward and at what slope / increased points per week? Is flat then linear with positive slope the historical and projected cutline trend behavior?