Projections (1 Viewer)

[BuckeyeArt]

Sorting through the staff projections and the projections in the spotlight series, it's clear that many are making projections based upon end of season stats versus per game stats. Why are they done this way? Wouldn't the projections be more useful if done on a per game basis?

Almost everyone plays in HTH leagues where you can submit new starting lineups each week. We don't have to take a zero when a player is injured. We get to start another player. So why do we include those injured weeks in the projections?

A good example is Steven Jackson. Dodds, Henry, and Woods all have him between 290-295 carries (around 18 per game). Tremblay has him at 334 (around 21 per game). Over the last 3 years, he has had an average of 279 per year but 21 per game. Does everyone but Tremblay expect him to drop from 21 to 18 carries per game?

Wouldn't it be far more useful to make projections on an average per game basis and let everyone else decide how many games they think the player will play?

 

[Dinsy Ejotuz]

Wouldn't it be far more useful to make projections on an average per game basis and let everyone else decide how many games they think the player will play?

Until someone shows that past injuries are correlated to future injuries, it would absolutely make more sense.

 

[Lott's Fingertip]

 

[Jason Wood]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

 

[Darko M]

I agree with the OP...it's also annoying that CBS won't give me per game averages from previous seasons...you pay $180 for a league and can't get that one stat?!

 

[davearm]

I haven't looked closely at this year's projections, but in the past anyway, Donovan McNabb was a great case study for this phenomenon.

For the last year or two, the FBG folks have typically projected season stats that placed him in the QB6-QB8 range (or worse), but those stats were based on 11 or 12 games played, because of his spotty injury history.

Look at the same forecast on a PPG basis, and suddenly he was like QB2. Roster a decent backup with him, and suddenly you're getting elite-level QB output for a bargain price.

 

[Lott's Fingertip]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

Picking a nit here, but how do we know how many games you are projecting for? If player A and player B are both being projected for around 1500 yards and 10 TDs, but not in the same number of games, don't you think this is pertinent information?

 

[Tigerbot Hesh]

 

[Deuce'sWild]

I don't like to assume injuries, even for guys who are injury prone. You just NEVER know. I base my rankings off a full 16 game schedule as if they will be there.

It's a good point, and I think trying to predict injuries has some merit, but is also somewhat of a guessing game.

 

[Jason Wood]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

Picking a nit here, but how do we know how many games you are projecting for? If player A and player B are both being projected for around 1500 yards and 10 TDs, but not in the same number of games, don't you think this is pertinent information?

Fair point. I know the GP data is available in the Dominator and VBD Excel spreadsheet, but it's not a column displayed on the site [even though the data is obviously there]. Not sure how non-trivial it would be to expose that data, probably something better handled by David or Keith or Doug.

 

[wraith5]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

Picking a nit here, but how do we know how many games you are projecting for? If player A and player B are both being projected for around 1500 yards and 10 TDs, but not in the same number of games, don't you think this is pertinent information?

Fair point. I know the GP data is available in the Dominator and VBD Excel spreadsheet, but it's not a column displayed on the site [even though the data is obviously there]. Not sure how non-trivial it would be to expose that data, probably something better handled by David or Keith or Doug.

I wonder if using the Projections Dominator to populate the DD data doesn't further complicate this issue... great discussion though. Being able to see a PPG column along with the season total could be very interesting and helpful. I know when I do my own projection work I have both (PPG and TOTAL), then I can sort by either or a blended ranking category. unt:

 

[BuckeyeArt]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

Picking a nit here, but how do we know how many games you are projecting for? If player A and player B are both being projected for around 1500 yards and 10 TDs, but not in the same number of games, don't you think this is pertinent information?

Fair point. I know the GP data is available in the Dominator and VBD Excel spreadsheet, but it's not a column displayed on the site [even though the data is obviously there]. Not sure how non-trivial it would be to expose that data, probably something better handled by David or Keith or Doug.

That would be great. It would also be nice to see it included as a requirement for the projections in the player spotlight series next year. There's a big difference between someone viewing a player in a RBBC, taking the FT job midseason, or missing a few games due to injury. They could all have the same end of season projections but have entirely different ppg projections.

 

[RCon]

We project for games played, too. So, for example, Lynch's numbers assume 13 games played. Maurile is typically more beholden to using average time missed in his projections, and will project down to the tenth decimal (for example, he'll project to 8.2 TDs) whereas I assume 16 games played as the baseline and adjust from there if a player is suspended, injured or has enough of an injury risk/concern that I feel a reduction is logical.

Picking a nit here, but how do we know how many games you are projecting for? If player A and player B are both being projected for around 1500 yards and 10 TDs, but not in the same number of games, don't you think this is pertinent information?

Fair point. I know the GP data is available in the Dominator and VBD Excel spreadsheet, but it's not a column displayed on the site [even though the data is obviously there]. Not sure how non-trivial it would be to expose that data, probably something better handled by David or Keith or Doug.

Where would we find the games played data in DD or the VBD spreadsheet?