Dirty Weasel
Footballguy
Is there a new link to the contest stats anywhere?
Fantasy Football - Footballguys Forums
This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
2008 $35,000 Subscriber Contest (2 Viewers)
- Thread starter Makavity
- Start date
Dirty Weasel
Footballguy
I'm encouraged by my 95.3% survival probability rate for this week. I've only got Portis on a bye, but 6 other RB's to pick from, including MJD vs Det. Let's see how it shakes out. Sadly, I see my team becoming less unique as weeks pass.
OK, I couldn't help but extend my simulation into future weeks. I can now play out the entire rest of the season and estimate what everyone's probability of winning is.
Here is the problem. In practice, I basically have no choice but to treat the future weeks as being independent. But they're not. And I think it makes a real difference.
Take Ray Rice for example. The forward projections have him at 68 fantasy points for the rest of the season (after this week). So that's seven games and 68 points, roughly 10 points per week. That seems pretty reasonable as an expected value. What my program does right now is to assume he is a 10-point-per-game player every week from week 11 to week 17. Then I simulate each week using the procedure described here.
But that's not really what we're looking at with Ray Rice, is it? It's more like he has a 20% chance of being a 20-point-per-game player (if they shelve McGahee) and a 50% chance of being an 8-point-per-game player and a 30% chance of being pretty worthless. In other words, if Rice gets 25 points in week 11, then we should most definitely adjust our estimates for him in future weeks. And building that into the program intelligently is, at least at this point, too complicated to be realistic.
Another issue is that the forward projections also don't attempt to project injuries. They (and hence my program) essentially assume everyone who is currently healthy will stay healthy. But that assumption could throw off the results. Teams with a couple of stars are more susceptible to an injury than teams with fewer stars and more depth.
I'm sure there are other issues that I'm not seeing.
The point is: I don't have nearly as much confidence in this as I do in the weekly estimates. And it's not just because future projections are more uncertain than this week's projections. It's because of fundamental complications that aren't going to be fixed this season (maybe ever). So big grains of salt are in order. Bigger than usual. This is just for fun.
That said, there is still some interesting stuff to be learned from this exercise. In particular, I wish I had done this 6 weeks ago. It would have been very useful for ferreting out teams that had serious bye week problems in certain weeks. This also led me to discover my favorite sleeper team and some others that might have a hard time in week 10 but potentially look really good for the long haul.
Let me explain what the output is. Here is a team picked at random from the middle of the pack:
---- survival % ---- wk10 wk11 wk12 wk13 top10 top50-------------------------------------------------------- 1179. Team 107781 | 804 520 248 113 1 6I ran 1000 runs of the rest of the season. 804 times (80.4%), this team survived the week 10 cut. [This team's number was 82% on the weekly sim. I'm going to assume/hope the difference is due to random variation. Unless there's a bug, the algorithm is identical for week 10.]
52% of the time, this team made it past the week 11 cut.
24.8% of the time, they made it past the week 12 cut.
11.3% of the time, they made it to the final 250.
Once out of 1000 runs, this team's overall finish was in the overall top 10. Six additional times, it was in the top 50. So a total of 7 times, this team finished in the overall top 50.
I guess it's the teacher in me, but I don't want anyone clicking the link unless they understand what they're looking at, so I have buried the link to the full result set inside this sentence. It's the comma after the word 'at'. EDIT: In the light of morning, this schtick seemed just a little over the top.
Full result set
Enjoy.
Here is the problem. In practice, I basically have no choice but to treat the future weeks as being independent. But they're not. And I think it makes a real difference.
Take Ray Rice for example. The forward projections have him at 68 fantasy points for the rest of the season (after this week). So that's seven games and 68 points, roughly 10 points per week. That seems pretty reasonable as an expected value. What my program does right now is to assume he is a 10-point-per-game player every week from week 11 to week 17. Then I simulate each week using the procedure described here.
But that's not really what we're looking at with Ray Rice, is it? It's more like he has a 20% chance of being a 20-point-per-game player (if they shelve McGahee) and a 50% chance of being an 8-point-per-game player and a 30% chance of being pretty worthless. In other words, if Rice gets 25 points in week 11, then we should most definitely adjust our estimates for him in future weeks. And building that into the program intelligently is, at least at this point, too complicated to be realistic.
Another issue is that the forward projections also don't attempt to project injuries. They (and hence my program) essentially assume everyone who is currently healthy will stay healthy. But that assumption could throw off the results. Teams with a couple of stars are more susceptible to an injury than teams with fewer stars and more depth.
I'm sure there are other issues that I'm not seeing.
The point is: I don't have nearly as much confidence in this as I do in the weekly estimates. And it's not just because future projections are more uncertain than this week's projections. It's because of fundamental complications that aren't going to be fixed this season (maybe ever). So big grains of salt are in order. Bigger than usual. This is just for fun.
That said, there is still some interesting stuff to be learned from this exercise. In particular, I wish I had done this 6 weeks ago. It would have been very useful for ferreting out teams that had serious bye week problems in certain weeks. This also led me to discover my favorite sleeper team and some others that might have a hard time in week 10 but potentially look really good for the long haul.
Let me explain what the output is. Here is a team picked at random from the middle of the pack:
---- survival % ---- wk10 wk11 wk12 wk13 top10 top50-------------------------------------------------------- 1179. Team 107781 | 804 520 248 113 1 6I ran 1000 runs of the rest of the season. 804 times (80.4%), this team survived the week 10 cut. [This team's number was 82% on the weekly sim. I'm going to assume/hope the difference is due to random variation. Unless there's a bug, the algorithm is identical for week 10.]
52% of the time, this team made it past the week 11 cut.
24.8% of the time, they made it past the week 12 cut.
11.3% of the time, they made it to the final 250.
Once out of 1000 runs, this team's overall finish was in the overall top 10. Six additional times, it was in the top 50. So a total of 7 times, this team finished in the overall top 50.
I guess it's the teacher in me, but I don't want anyone clicking the link unless they understand what they're looking at, so I have buried the link to the full result set inside this sentence. It's the comma after the word 'at'. EDIT: In the light of morning, this schtick seemed just a little over the top.
Full result set
Enjoy.
Last edited by a moderator:
Dirty Weasel
Footballguy
I'm team #69 in your simulation. I think that's a sign, don't you?
ctriopelle
Footballguy
Ouch....down to a coin flip for survival. 47.5%. I thought I may be in a but of a pinch, but less than 50% isn't making me feel very good.
Looks like the predicted cut from the simulation is roughly 150 if I understand things correctly. Hang on tight girls, this week is gonna be a bumpy ride!
Good luck to everyone this week!
Looks like the predicted cut from the simulation is roughly 150 if I understand things correctly. Hang on tight girls, this week is gonna be a bumpy ride!
Good luck to everyone this week!
Das Boot
Footballguy
Wow, I love that team.Only a couple of minor misses on players, several very nice hits (so far), and great uniqueness.
ctriopelle
Footballguy
Just scrolled down to the big end game prediction. I'm curious to the sort methodology, as it isn't completely obvious to me. I'm in slot # 63, but I have poor advancement probabilities relative to others around me. It looks like it is somewhat tied to finishes in the top 10 (and I like my number), but it doesn't look to be strictly sorted either by that or the top 10's plus top 50's either. Any insight?
Last edited by a moderator:
ctriopelle
Footballguy
in the play out simulation, is it safe to say that the # of advances in week n+1 relative to the number of advances in week n would be your expected survival % rate for that week? Basically, since my numbers for this week are pretty bad, it holds my team's odds of making it to the finals much lower? If that is effectively the case, then I really like what I see, because I cash top 10 32 times out of the 161 times I get to the finals, which is the second highest % in the data (only lower than the guy at the very top).
If I am interpreting things correctly, I am REALLY feeling the pressure to advance this week, as it looks like my make or break point.
If I am interpreting things correctly, I am REALLY feeling the pressure to advance this week, as it looks like my make or break point.
dbc925
Footballguy
Not surprisingly, this is the top team in the simulation to take it all.
kroyrunner89
Footballguy
I found the link to see the one week only records, but can i have a link to view the rest of the stuff like week by week survival? i can't seem to find it...
never mind, its amazing the things you find when you read carefully...
never mind, its amazing the things you find when you read carefully...
Last edited by a moderator:
Das Boot
Footballguy
The more I look at it, the more amazing it is.Torain, Rice, Charles, and Avery are just now starting to add their contributions to his weekly scores.
EBF
Footballguy
Doug's numbers say I only have a 36.6% chance of surviving this week, so they're clearly not very accurate. 
Actually, I'm up a creek this week with Ricky and MJD being my only active RBs. However, if I can somehow survive this week then I might be one of the few teams remaining with Bush and Shockey.
Actually, I'm up a creek this week with Ricky and MJD being my only active RBs. However, if I can somehow survive this week then I might be one of the few teams remaining with Bush and Shockey.
Fantasy Feud
Footballguy
Doug or others - How does one access the future weeks simulation? I looked for a link within Doug's email but could not find one.
geoff8695
Footballguy
Try re-reading Doug's post VERY carefully. It's in there...
Yeah, that appears to be some buggy sort order, doesn't it. I was intended to do top10, then top50, then final250, but that's not the way it turned out. I'll look into it.
Yep. Dividing would give you your conditional probabilities. For instance, this guy again:
Code:
---- survival % ---- wk10 wk11 wk12 wk13 top10 top50--------------------------------------------------------1179. Team 107781 | 804 520 248 113 1 6BuckeyeArt
Footballguy
Very nice, Doug. But, you're going to be bored with nothing left to do on this. This will be a cool thing to see in the 1st week next year.Sort by final 250 1st. I think that's everyone's immediate goal.
Last edited by a moderator:
This entry is still alive.
1 2 3 4 5 6 7 8 9
--------------------------------------------------------------------------------------------
Kurt Warner $5 15.45 36.05 20.60 32.60 25.50 22.70 0.00 30.05 28.90
Matt Ryan $4 13.85 6.90 17.10 9.00 22.30 21.15 0.00 24.25 23.10
Kyle Orton $4 8.50 7.35 25.50 25.95 29.70 20.30 25.95 0.00 12.10
Steven Jackson $50 8.90 12.50 15.30 27.30 0.00 12.60 36.60 0.00 1.70
Clinton Portis $33 8.40 22.90 15.70 14.50 22.80 26.80 24.80 13.70 15.90
Ricky Williams $8 5.90 3.40 12.90 0.00 4.40 12.00 1.60 12.90 4.40
Ray Rice $4 9.80 0.00 2.10 0.00 0.00 12.20 6.90 11.60 19.10
Steve Slaton $1 6.40 0.00 20.40 21.60 22.10 6.60 16.40 13.60 15.80
Randy Moss $47 23.60 4.20 6.50 0.00 22.30 5.60 23.90 17.20 12.50
Calvin Johnson $25 17.70 30.90 8.00 0.00 3.60 18.50 23.20 16.40 22.80
Bernard Berrian $13 6.80 0.00 10.90 12.80 23.00 24.40 20.10 0.00 18.40
Donte Stallworth $12 0.00 0.00 0.00 0.00 0.00 3.90 4.40 10.30 0.00
David Patten $5 10.90 10.50 3.20 3.10 0.00 0.00 0.00 0.00 0.00
Derek Hagan $4 0.00 8.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00
L.J. Smith $12 17.40 2.50 0.00 0.00 7.10 7.70 0.00 5.90 0.00
Ben Utecht $5 4.00 0.00 0.00 4.00 1.80 9.40 0.00 0.00 5.10
Dustin Keller $2 0.00 3.40 16.10 9.90 0.00 0.00 2.10 9.80 4.90
John Kasay $2 20.00 11.00 6.00 8.00 13.00 3.00 15.00 12.00 0.00
Mike Nugent $1 2.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Minnesota Vikings $8 0.00 6.00 15.00 2.00 15.00 10.00 4.00 0.00 11.00
Carolina Panthers $3 9.00 5.00 5.00 2.00 9.00 0.00 5.00 6.00 0.00
Kansas City Chiefs $2 6.00 4.00 2.00 9.00 4.00 0.00 0.00 13.00 8.00
--------------------------------------------------------------------------------------------
TOTAL 141.15 145.55 139.40 138.80 163.00 145.20 193.05 148.95 149.50
CUTOFF 112.85 132.60 133.60 138.70 122.20 139.65 121.30 138.55 135.10
This may be my last week. I think I had the optimal QB strategy ($13 and highly productive every week), but had too many misses at WR and TE. With Portis off and Steven Jackson questionable, it will be very difficult to make it through. But it was great to last this long having had several very close calls.
1 2 3 4 5 6 7 8 9
--------------------------------------------------------------------------------------------
Kurt Warner $5 15.45 36.05 20.60 32.60 25.50 22.70 0.00 30.05 28.90
Matt Ryan $4 13.85 6.90 17.10 9.00 22.30 21.15 0.00 24.25 23.10
Kyle Orton $4 8.50 7.35 25.50 25.95 29.70 20.30 25.95 0.00 12.10
Steven Jackson $50 8.90 12.50 15.30 27.30 0.00 12.60 36.60 0.00 1.70
Clinton Portis $33 8.40 22.90 15.70 14.50 22.80 26.80 24.80 13.70 15.90
Ricky Williams $8 5.90 3.40 12.90 0.00 4.40 12.00 1.60 12.90 4.40
Ray Rice $4 9.80 0.00 2.10 0.00 0.00 12.20 6.90 11.60 19.10
Steve Slaton $1 6.40 0.00 20.40 21.60 22.10 6.60 16.40 13.60 15.80
Randy Moss $47 23.60 4.20 6.50 0.00 22.30 5.60 23.90 17.20 12.50
Calvin Johnson $25 17.70 30.90 8.00 0.00 3.60 18.50 23.20 16.40 22.80
Bernard Berrian $13 6.80 0.00 10.90 12.80 23.00 24.40 20.10 0.00 18.40
Donte Stallworth $12 0.00 0.00 0.00 0.00 0.00 3.90 4.40 10.30 0.00
David Patten $5 10.90 10.50 3.20 3.10 0.00 0.00 0.00 0.00 0.00
Derek Hagan $4 0.00 8.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00
L.J. Smith $12 17.40 2.50 0.00 0.00 7.10 7.70 0.00 5.90 0.00
Ben Utecht $5 4.00 0.00 0.00 4.00 1.80 9.40 0.00 0.00 5.10
Dustin Keller $2 0.00 3.40 16.10 9.90 0.00 0.00 2.10 9.80 4.90
John Kasay $2 20.00 11.00 6.00 8.00 13.00 3.00 15.00 12.00 0.00
Mike Nugent $1 2.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Minnesota Vikings $8 0.00 6.00 15.00 2.00 15.00 10.00 4.00 0.00 11.00
Carolina Panthers $3 9.00 5.00 5.00 2.00 9.00 0.00 5.00 6.00 0.00
Kansas City Chiefs $2 6.00 4.00 2.00 9.00 4.00 0.00 0.00 13.00 8.00
--------------------------------------------------------------------------------------------
TOTAL 141.15 145.55 139.40 138.80 163.00 145.20 193.05 148.95 149.50
CUTOFF 112.85 132.60 133.60 138.70 122.20 139.65 121.30 138.55 135.10
This may be my last week. I think I had the optimal QB strategy ($13 and highly productive every week), but had too many misses at WR and TE. With Portis off and Steven Jackson questionable, it will be very difficult to make it through. But it was great to last this long having had several very close calls.
Kuz
Footballguy
It's not really that surpising if you look closer at his RB's.. other than Barber he went for all rookies probably hoping a few will pan out. It now looks like quite a few will pan out.. Very tough team down the stretch...
Sheriff66
Footballguy
Can someone run my team and post the results for me, won't allow me to do it at work...104782
TIA
pennystock
Footballguy
75.2%, ranked 1,057Sheriff66 said:
Nick Vermeil
Footballguy
Yikes, 24%. I'm down to 1 QB with the benching of Anderson, and with Barber and Cooley on a bye, this could be it. Need a huge game from Garrard!
apalmer
Footballguy
Could be some crying if this guy doesn't make the cut.
ctriopelle
Footballguy
KW II is my warrior hero tonight! If I make it past this week, I will most definitely say it was due in large part to his contribution. Also pretty happy that Cutler is having a nice night for me too.
ctriopelle
Footballguy
117.25 from Cutler, KW II, Marshall, and Dawson. PLEASE let the rest of my guys put up some respectable numbers this weekend and get me past my nasty bye week! Hoping to shame the Turk with his 53% chance of my demise! 
PS, Warner can take this week off as long as Fitz and Breaston can still put up some points!
PS, Warner can take this week off as long as Fitz and Breaston can still put up some points!
Last edited by a moderator:
The Kerminator
Footballguy
With Winslow, Marshall and Cutler having HUGE games it's already looking like a high cutoff. 
You still had better odds to move on than to be eliminated. I say he got this one right. Now if he had you at 48% then117.25 from Cutler, KW II, Marshall, and Dawson. PLEASE let the rest of my guys put up some respectable numbers this weekend and get me past my nasty bye week! Hoping to shame the Turk with his 53% chance of my demise!
PS, Warner can take this week off as long as Fitz and Breaston can still put up some points!
Ministry of Pain
Footballguy
"You insensitive #######...now I'll move my car, but I want you to help him up."79.0% this week, 53.5% the next, 27.5%, 12.9%...and then a big fat zero...thanks for taking all the mystery out of this
ctriopelle
Footballguy
I had a 47% chance of moving on, hence a 53% chance of my demise. So he gave me a less than 50/50 shot...You still had better odds to move on than to be eliminated. I say he got this one right. Now if he had you at 48% then117.25 from Cutler, KW II, Marshall, and Dawson. PLEASE let the rest of my guys put up some respectable numbers this weekend and get me past my nasty bye week! Hoping to shame the Turk with his 53% chance of my demise!
PS, Warner can take this week off as long as Fitz and Breaston can still put up some points!
ctriopelle
Footballguy
Winslow is only on a tiny % of rosters. He won't impact the cut much by himself. Cutler and Marshall though are setting a bunch of teams up nicely. I only have Warner and Cutler as my QB's this week, so if I had to pick a really high scoring QB, I wanted it to be Cutler. Unfortunately my success is also tightly tied to Fitz/Breaston, so I will need Warner to have a decent day too (just hoping he throws a lot of short passes to those 2 to advance them faster than himself). My team is awfully thin this week with injuries and byes, so I can't really afford to have any poor scores.With Winslow, Marshall and Cutler having HUGE games it's already looking like a high cutoff.
Dirty Weasel
Footballguy
699 live teams (30.4%) with Cutler getting a nice start with 42.25 pts
385 live teams (16.7%) with Cutler/Marshall getting a nice start with 63.15 pts
94 live teams (4.1%) with Winslow getting a nice start with 38.10 pts
35 live teams (1.5%) with Cutler/Winslow getting a nice start with 80.35 pts
23 live teams (1%) with Cutler/Marshall/Winslow getting a nice start with 101.25 pts
4 live teams (.2%) with Cutler/Marshall/Winslow/Prater getting a nice start with 113.25 pts
2 live teams (.1%) with Cutler/Marshall/Winslow/Dawson getting a nice start with 117.25 pts
These numbers tell me that Winslow's monster game won't really raise the cutoff by much, because he is quite rare. I fall in the Cutler/Marshall category, along with one-sixth of the remaining teams. I didn't dig through all 385 teams, but of those 385 teams, here are the numbers for other high-priced players (RB's $20+, WR's $15+, TE's $10+) who are off...
- 150 own Clinton Portis
- 104 own Marion Barber
- 54 own Chris Cooley
- 52 own Santana Moss (I know he's only $14, but just wanted to add him)
- 50 own Willie Parker (looking iffy to play)
- 39 own Patrick Crayton (I know he's only $13, but just wanted to add him too)
- 33 own Jonathan Stewart (bad heel, missing practice)
- 32 own Nate Burleson
- 31 own Selvin Young (low score this week)
- 25 own Jason Witten
- 25 own Earnest Graham
- 15 own Heath Miller (doubtful)
- 13 own Terrell Owens
- 12 own Reggie Bush
- 11 own Darren McFadden (has missed last 2 games)
- 8 own Joey Galloway
- 7 own Steven Jackson (looking iffy to play)
- 5 own Braylon Edwards (low score this week)
- 3 own Randy McMichael
- 2 own Roy Williams
- 1 owns Larry Johnson
- 0 own Laurence Maroney
The only player I own on this list is Clinton Portis. Hoping for a huge cut of the 777 Portis owners (as long as I'm not one of them).
385 live teams (16.7%) with Cutler/Marshall getting a nice start with 63.15 pts
94 live teams (4.1%) with Winslow getting a nice start with 38.10 pts
35 live teams (1.5%) with Cutler/Winslow getting a nice start with 80.35 pts
23 live teams (1%) with Cutler/Marshall/Winslow getting a nice start with 101.25 pts
4 live teams (.2%) with Cutler/Marshall/Winslow/Prater getting a nice start with 113.25 pts
2 live teams (.1%) with Cutler/Marshall/Winslow/Dawson getting a nice start with 117.25 pts
These numbers tell me that Winslow's monster game won't really raise the cutoff by much, because he is quite rare. I fall in the Cutler/Marshall category, along with one-sixth of the remaining teams. I didn't dig through all 385 teams, but of those 385 teams, here are the numbers for other high-priced players (RB's $20+, WR's $15+, TE's $10+) who are off...
- 150 own Clinton Portis
- 104 own Marion Barber
- 54 own Chris Cooley
- 52 own Santana Moss (I know he's only $14, but just wanted to add him)
- 50 own Willie Parker (looking iffy to play)
- 39 own Patrick Crayton (I know he's only $13, but just wanted to add him too)
- 33 own Jonathan Stewart (bad heel, missing practice)
- 32 own Nate Burleson
- 31 own Selvin Young (low score this week)
- 25 own Jason Witten
- 25 own Earnest Graham
- 15 own Heath Miller (doubtful)
- 13 own Terrell Owens
- 12 own Reggie Bush
- 11 own Darren McFadden (has missed last 2 games)
- 8 own Joey Galloway
- 7 own Steven Jackson (looking iffy to play)
- 5 own Braylon Edwards (low score this week)
- 3 own Randy McMichael
- 2 own Roy Williams
- 1 owns Larry Johnson
- 0 own Laurence Maroney
The only player I own on this list is Clinton Portis. Hoping for a huge cut of the 777 Portis owners (as long as I'm not one of them).
thanks for finally showing up a week late Winslow :(
Yes, I'm crying
Still bitter about that Gaffney drop last week, I missed out by 1.5 pts
(Ok, I'm better now)
Now back to your scheduled programming of people actually STILL in the contest! Good luck to you guys
Yes, I'm crying
Still bitter about that Gaffney drop last week, I missed out by 1.5 pts
(Ok, I'm better now)Now back to your scheduled programming of people actually STILL in the contest! Good luck to you guys
Last edited by a moderator:
Jayrod
Footballguy
Last week was the second time that the loss of any one player would have done me in.
Had I not had Warner's 28.9 pts, I would have had to rely on Schaub's 5.95....a drop of 22.95. I only made the cut by 17.9.
The only other time was in week 2, when losing Barber's 25.4 would have left me with 3.4 pts from Keller at the flex (difference of 22.0 pts). I made the cut by only 20.95 that week.
Had I not had Warner's 28.9 pts, I would have had to rely on Schaub's 5.95....a drop of 22.95. I only made the cut by 17.9.
The only other time was in week 2, when losing Barber's 25.4 would have left me with 3.4 pts from Keller at the flex (difference of 22.0 pts). I made the cut by only 20.95 that week.
Grace Under Pressure
Footballguy
2213
26.4%
136
Oof. Not looking good.
26.4%
136
Oof. Not looking good.
Dirty Weasel
Footballguy
Doug,
When you get a moment, can you please post the current contest stats? You know, the one with the % owned and % live. I don't think it ever got posted after week 9. TIA
When you get a moment, can you please post the current contest stats? You know, the one with the % owned and % live. I don't think it ever got posted after week 9. TIA
Here you go, DW.
Dirty Weasel
Footballguy
Muchos gracias
Sheriff66
Footballguy
Sol D
Footballguy
Yes Mr. Warner feel free to take the week off... hand it to your new starting RB 35 times! With Schaub down Cutler picked a fine week to wake up. Thanks Jay!117.25 from Cutler, KW II, Marshall, and Dawson. PLEASE let the rest of my guys put up some respectable numbers this weekend and get me past my nasty bye week! Hoping to shame the Turk with his 53% chance of my demise!
Took 20 of the lower-ranked teams from the power rankings published before week 9 on October 29th (from #2500 to 2519). Went backwards and forwards. Here are the results.#2500 - finished #1953 in week 9 - previously had a power ranking of #3115 (before week 8 on Oct. 22nd)
#2501 - finished #1027 in week 9 - previously #2945
#2502 - finished #2890 in week 9 - previously #2553
#2503 - finished #1471 in week 9 - previously #2604
#2504 - finished #2779 in week 9 - previously #2964
#2505 - finished #2851 in week 9 - previously #3282
#2506 - finished #1391 in week 9 - previously #2455
#2507 - finished #519 in week 9 - previously #3544
#2508 - finished #1514 in week 9 - previously #3533
#2509 - finished #395 in week 9 - previously #2734
#2510 - finished #443 in week 9 - previously #2682
#2511 - finished #2151 in week 9 - previously #3492
#2512 - finished #2748 in week 9 - previously #3101
#2513 - finished #2839 in week 9 - previously #2292
#2514 - finished #2954 in week 9 - previously #3565
#2515 - finished #219 in week 9 - previously #2928
#2516 - finished #152 in week 9 - previously #2831
#2517 - finished #418 in week 9 - previously #2451
#2518 - finished #2288 in week 9 - previously #3063
#2519 - finished #3022 in week 9 - previously #2164
It appears that 13 of the 20 low-ranked teams (according to the latest power rankings) made the cut from 3106 teams to 2305 teams after week 9. The overall survival rate was 74.2% (2305/3106) among all teams. It was only a little lower - 65% - among the low-ranked teams. Kind of surprising that 6 of these teams (30%) finished in the top-1000 in week 9 - and 11 teams finished in the top-2000 (55%).
These 20 teams were also ranked low in the previous week's power rankings (from #2164 to #3565).
Of the 13 surviving teams, here are the latest predicted survival rank from the simulation for week 10:
#2500 - predicted survival #2166 (will not make cut to 1600 teams after week 10) - MADE WEEK 10 CUT (#1115) - INCORRECT PREDICTION
#2501 - predicted survival #521 (will make cut) - MADE CUT (#137) - CORRECT
#2503 - predicted survival #894 (will make cut) - DID NOT MAKE CUT (#1784) - INCORRECT
#2506 - predicted survival #1602 (will not make cut) - MADE CUT(#482) - INCORRECT
#2507 - predicted survival #1867 (will not make cut) - DID NOT MAKE CUT (#1903) - CORRECT
#2508 - predicted survival #2020 (will not make cut) - DID NOT MAKE CUT (#1638) - CORRECT
#2509 - predicted survival #601 (will make cut) - MADE CUT (#475) - CORRECT
#2510 - predicted survival #780 (will make cut) - MADE CUT (#602) - CORRECT
#2511 - predicted survival #1973 (will not make cut) - MADE CUT (#77) - INCORRECT
#2515 - predicted survival #2014 (will not make cut) - DID NOT MAKE CUT (#2263) - CORRECT
#2516 - predicted survival #1641 (will not make cut) - MADE CUT (#1258) - INCORRECT
#2517 - predicted survival #1529 (will make cut) - MADE CUT (#261) - CORRECT
#2518 - predicted survival #2302 (will not make cut) - DID NOT MAKE CUT (#2293) - CORRECT
So the first team was ranked #3115 in the week 8 power rankings, #2500 in the week 9 power rankings, actually finished #1953 in week 9 and made the cut. This team is ranked #2166 by the simulation for survival after week 10, and the prediction is that it will not make the cut to 1600 teams.
Thus, only 5 of the 13 surviving teams are predicted to make the cut this week - which is consistent with these teams being low-ranked for the past several weeks.
ETA (11/9/08): additional low-ranked teams (according to power rankings going into week 9) that survived week 9, with their survival rank according to the simulation for week 10:
#2524 - predicted survival #1652 (will not make cut) - MADE CUT (#821) - INCORRECT
#2526 - predicted survival #713 (will make cut) - MADE CUT (#1106) - CORRECT
#2530 - predicted survival #1354 (will make cut) - MADE CUT (#132) - CORRECT
#2531 - predicted survival #2218 (will not make cut) - DID NOT MAKE CUT (#2095) - CORRECT
#2533 - predicted survival #1857 (will not make cut) - MADE CUT (#1265) - INCORRECT
#2534 - predicted survival #2026 (will not make cut) - MADE CUT (#675) - INCORRECT
#2536 - predicted survival #1960 (will not make cut) - MADE CUT (#1230) - INCORRECT
#2537 - predicted survival #1118 (will make cut) - DID NOT MAKE CUT (#2023) - INCORRECT
#2538 - predicted survival #1014 (will make cut) - MADE CUT (#1235) - CORRECT
#2540 - predicted survival #1537 (will make cut) - MADE CUT (#239) - CORRECT
#2542 - predicted survival #1979 (will not make cut) - MADE CUT (#1134) - INCORRECT
#2543 - predicted survival #2294 (will not make cut) - MADE CUT (#1037) - INCORRECT
#2544 - predicted survival #1330 (will make cut) - MADE CUT (#447) - CORRECT
#2548 - predicted survival #1953 (will not make cut) - MADE CUT (#1483) - INCORRECT
#2549 - predicted survival #1591 (will make cut) - MADE CUT (#1509) - CORRECT
#2553 - predicted survival #2260 (will not make cut) - DID NOT MAKE CUT (#2074) - CORRECT
#2554 - predicted survival #1977 (will not make cut) - DID NOT MAKE CUT (#2150) - CORRECT
#2556 - predicted survival #1631 (will not make cut) - MADE CUT (#1396) - INCORRECT
#2559 - predicted survival #1995 (will not make cut) - MADE CUT (#1194) - INCORRECT
#2560 - predicted survival #1985 (will not make cut) - DID NOT MAKE CUT (#2003) - CORRECT
#2562 - predicted survival #775 (will make cut) - DID NOT MAKE CUT (#2198) - INCORRECT
#2563 - predicted survival #1570 (will make cut) - DID NOT MAKE CUT (#1570) - INCORRECT
#2567 - predicted survival #1061 (will make cut) - MADE CUT (#875) - CORRECT
#2568 - predicted survival #2087 (will not make cut) - MADE CUT (#1071) - INCORRECT
#2569 - predicted survival #2300 (will not make cut) - DID NOT MAKE CUT (#2246) - CORRECT
#2570 - predicted survival #1149 (will make cut) - MADE CUT (#1468) - CORRECT
#2571 - predicted survival #1058 (will make cut) - MADE CUT (#1587) - CORRECT
Of these 40 low-ranked teams (that survived through week 9), 17 are predicted to make the cut for week 10 (42.5%), where the overall survival rate for all remaining teams is 70%. If you had a purely random ranking of teams (#1, #2, ... , #2305), then you would expect about 70% of the high-ranked teams, 70% of the middle-ranked teams, and 70% of the low-ranked teams to survive the cut at week 10. OTOH if you had perfect predictive power, then every one of the first 1600 teams would survive, and none of the teams ranked #1601 to #2305 would survive. Although the old power rankings didn't appear to be much better than a random ranking for the low-ranked teams (for survival on a week-to-week basis), the week 10 simulation seems to be much more focused and specific to week 10 performance, and more accurate results would be expected. Considering the huge amount of variation in player performance from week to week, even the most accurate projections for individual players will differ substantially from actual performance, and simulations based on projected points will contain a similar level of variation (in predicting survival of individual teams). For example, the survival rate for all high-ranked teams would be expected to be significantly above 70% (maybe 85%) and the survival rate for low-ranked teams (above #1600) to be less than 40%. Someone could probably test whether the difference in week 10 survival rates (between the 17 teams predicted to survive and the 23 teams predicted to be cut) is statistically significant.
ETA (11/10/08): 29 more low-ranked teams (according to power rankings going into week 9) that survived week 9, with their survival rank according to the simulation for week 10:
#2454 - predicted survival #2271 (will not make cut) - DID NOT MAKE CUT (#2233) - CORRECT
#2455 - predicted survival #1851 (will not make cut) - MADE CUT (#773) - INCORRECT
#2456 - predicted survival #372 (will make cut) - DID NOT MAKE CUT (#1888) - INCORRECT
#2458 - predicted survival #1623 (will not make cut) - MADE CUT (#805) - INCORRECT
#2359 - predicted survival #2301 (will not make cut) - DID NOT MAKE CUT (#2296) - INCORRECT
#2463 - predicted survival #821 (will make cut) - MADE CUT (#438) - CORRECT
#2467 - predicted survival #2282 (will not make cut) - DID NOT MAKE CUT (#2291) - CORRECT
#2470 - predicted survival #708 (will make cut) - MADE CUT (#1004) - CORRECT
#2472 - predicted survival #1939 (will not make cut) - DID NOT MAKE CUT (#1996) - CORRECT
#2473 - predicted survival #2298 (will not make cut) - DID NOT MAKE CUT (#2271) - CORRECT
#2475 - predicted survival #1769 (will not make cut) - DID NOT MAKE CUT (#1963) - CORRECT
#2476 - predicted survival #1031 (will make cut) - DID NOT MAKE CUT (#1691) - INCORRECT
#2477 - predicted survival #1523 (will make cut) - MADE CUT (#623) - CORRECT
#2478 - predicted survival #1364 (will make cut) - MADE CUT (#964) - CORRECT
#2480 - predicted survival #1334 (will make cut) - MADE CUT (#735) - CORRECT
#2481 - predicted survival #1037 (will make cut) - MADE CUT (#1057) - CORRECT
#2482 - predicted survival #960 (will make cut) - MADE CUT (#1181) - CORRECT
#2484 - predicted survival #1706 (will not make cut) - MADE CUT (#273) - INCORRECT
#2486 - predicted survival #2213 (will not make cut) - DID NOT MAKE CUT (#2231) - CORRECT
#2487 - predicted survival #1246 (will make cut) - MADE CUT (#271) - CORRECT
#2488 - predicted survival #1530 (will make cut) - MADE CUT (#1490) - CORRECT
#2489 - predicted survival #2137 (will not make cut) - DID NOT MAKE CUT (#1939) - CORRECT
#2490 - predicted survival #1689 (will not make cut) - MADE CUT (#911) - INCORRECT
#2492 - predicted survival #2037 (will not make cut) - DID NOT MAKE CUT (#2000) - CORRECT
#2493 - predicted survival #750 (will make cut) - MADE CUT (#1405) - CORRECT
#2494 - predicted survival #2178 (will not make cut) - DID NOT MAKE CUT (#1981) - CORRECT
#2495 - predicted survival #1366 (will make cut) - DID NOT MAKE CUT (#1607) - INCORRECT
#2496 - predicted survival #2296 (will not make cut) - DID NOT MAKE CUT (#2161) - CORRECT
#2498 - predicted survival #2082 (will not make cut) - DID NOT MAKE CUT (#1791) - CORRECT
Of these 69 surviving teams (to week 9), 30 (43.5%) are predicted to make the cut by the week 10 simulation. If the survival rate for the group of 30 teams predicted by the simulation to make the cut is significantly higher than 70% and the survival rate for the group of 39 teams predicted not to make the cut is less than 25%, then IMO the simulation model is excellent, given the high level of variance in week-to-week player performance.
I'd venture the following guesses (1) the overall survival rate for all 69 teams will be less than 70% but significantly higher than 43.5%, (2) the survival rate for the group of teams predicted to make the cut will be significantly less than 70% but substantially higher than the other group, but not by more than 35% higher, and (3) the survival rate for the group not predicted to make the cut will be significantly greater than zero and probably higher than 25% (one in four teams not predicted to make the cut will actually make the cut).
ETA (11/11/08): results in caps above.
Of the 30 teams predicted to make the week 10 cut, 23 teams (76.7%) actually made the cut. Of the 39 teams predicted not to make the cut, 18 teams (46.2%) actually made the cut. For all 69 teams, 41 teams (59.4%) actually made the cut, compared to a prediction of 43.5%.
Regarding the above guesses (1) the overall survival rate was 59.4%, less than 70% but significantly higher than 43.5%, (2) the survival rate for the group of teams predicted to make the cut was 76.7%, higher than 70% and 30.5% higher than the other group, and (3) the survival rate for the group not predicted to make the cut was 46.2%, significantly greater than 25%.
For the 30 teams predicted to make the cut, the actual survival rate of 76.7% was somewhat higher than the average survival rate of 70% for all 2305 teams - which indicates a pretty good prediction model IMO - especially considering how these teams were initially selected (from among the low-ranked teams in the week 9 power rankings). OTOH, if you picked 30 teams at random from the 2305 teams alive last week, you would expect a survival rate of 70%. Highly-ranked teams predicted by the simulation model to finish #372, 775, 894, 1031, and 1118 did not make the cut - consistent with the high level of variability in week-to-week player performance - so having a high ranking by the model doesn't automatically mean that you'll survive.
For the 39 teams predicted not to make the cut, a highly accurate model would have a survival rate less than 25%. The actual survival rate for these teams was 46.2% - which IMO indicates that the major weakness of the model is in predicting which teams will not make the cut. Basically, if the model predicts that your team won't make the cut, you have about a 50% chance of actually making the cut. A team predicted to finish #2026 actually finished #675, and teams predicted not to make the cut actually finished #77, 273, 482, 675, 773, 805, 821, 911, 1027, 1071, 1115, 1134, 1194, 1230, 1258, 1265, 1396, and 1483.
Last edited by a moderator:
Das Boot
Footballguy
Just a little though... Torain, Rice, Charles, and Avery were all cheap speculative flyers that are just now starting to pay off - but his team was already very solid without the flyers.
Robbie Cooper
Footballguy
MJD and Ricky Williams could prove to be a decent combo this week with MBIII out.
Challenge Everything
Footballguy
Slaton and Rice not doing anything... yet. I might be done unless my WR's can pull me out.
EBF
Footballguy
Doug's numbers say I only have a 36.6% chance of surviving this week, so they're clearly not very accurate.
Actually, I'm up a creek this week with Ricky and MJD being my only active RBs. However, if I can somehow survive this week then I might be one of the few teams remaining with Bush and Shockey.
I'm probably jinxing myself by gloating so early, but it just goes to show you that the luck factor is huge. I absolutely needed Ricky and MJD to have two of their best weeks of the season and they certainly seem headed in that direction.I still have to dodge a lot of bullets to survive this week's cut, but I can't complain about my start.
QB - Peyton Manning - 27
QB - Kurt Warner - 5
RB - Maurice Jones-Drew - 34
RB - Reggie Bush - 33
RB - Clinton Portis - 33
RB - Ricky Williams - 8
WR - Calvin Johnson - 25
WR - Santana Moss - 14
WR - Ted Ginn - 13
WR - Isaac Bruce - 10
WR - Reggie Williams - 8
WR - Antwaan Randle El - 4
WR - DeSean Jackson - 4
WR - Demetrius Williams - 3
WR - Antonio Bryant - 2
TE - Jeremy Shockey - 18
TE - Dustin Keller - 2
PK - Mike Nugent - 1
PK - Rian Lindell - 1
TD - Houston Texans - 2
TD - New Orleans Saints - 2
TD - St. Louis Rams - 1
