Using Expert Projections (1 Viewer)

[Hurricane]

There's about three expert opinions I usually rely upon for my ff projections. Question: if I were to combine them into a consensus - average their projections - would that help or hurt the overall effort? Thanks in advance.

 

[Mungo Burrows]

There's about three expert opinions I usually rely upon for my ff projections. Question: if I were to combine them into a consensus - average their projections - would that help or hurt the overall effort? Thanks in advance.

Averaging them will give you a unique set of projections to work from but that probably doesn't help much as you'll just end up close to what everybody else has. the extremes of each projection will cancel each other out unless they're all in the tank for the same person.A different approach might be to look at where each of those projections diverge from the norm, try to understand why that person you seemingly trust differs from the sheep. If you agree with that reasoning, keep the person highly rated if not, discount it.

 

[bjabrad]

Cheatsheets from one site may end up being more accurate for one position like QB's than they are for RB's so try to remember which ones helped you by position if you can. Just like in your own league some guys are better at rating defenses or kickers than other owners.

I think there is value in averaging cheatsheets so that you can get rid of the fringe of the bell curve or the anomolies or the guys who are going out on a limb. If you only get your info from one source and treat it as gospel you may think you're in the safe zone and may not know that you're out on a limb with your sole source.

That's why the Shark Pool is so helpful. You can get analysis from all across the country to dial in your own opinion or gut for what to do each week. FBG's does a great job with their cheatsheets but I'd rather use them as one of many tools rather than to blindly trust what is served up.

The great thing is that if you are reading this here, then you are already a leg up on most of your other owners.

 

[The Jerk]

There's about three expert opinions I usually rely upon for my ff projections. Question: if I were to combine them into a consensus - average their projections - would that help or hurt the overall effort? Thanks in advance.

Averaging them will give you a unique set of projections to work from but that probably doesn't help much as you'll just end up close to what everybody else has. the extremes of each projection will cancel each other out unless they're all in the tank for the same person.A different approach might be to look at where each of those projections diverge from the norm, try to understand why that person you seemingly trust differs from the sheep. If you agree with that reasoning, keep the person highly rated if not, discount it.

This is a very insightful post. A pure averaging of projections will end up bringing players closer together rather than separating the talent to make it easier to figure out who to select and when.First, I think you should limit the number of projection sources you are using.Second, Rather than average players that have a large discrepancy, choose to remove the outlier or pick the projection that you think makes the most sense.

 

[barndog]

A rather large set of experimental data indicates groups do better at answering questions/predicting outcomes than individuals. Of course, what you are talking about here is not a group prediction but indiv predictions merged. Perhaps what we should do is ask FBG to get their rankers all around the table together once a week.

 

[The Jerk]

A rather large set of experimental data indicates groups do better at answering questions/predicting outcomes than individuals. Of course, what you are talking about here is not a group prediction but indiv predictions merged. Perhaps what we should do is ask FBG to get their rankers all around the table together once a week.

I don't dispute this, yet averaging the numbers mathematically is not equivalent to group decision-making, at least as far as I see it. All it takes is a few extremely high or low rankings to distort the consensus. In statistical terms, the median would be more useful than the mean. Perhaps averaging after throwing out the highest and lowest rankings/projections would be more effective in targeting the "group" opinion.In concrete terms... if 16 out of 20 experts rank Player A above Player B, but the other four experts significantly devalue Player A, they could end up with similar averages. Yet I believe the group consensus noticeably favors Player A.Multiply this situation dozens of times and you can see my problem with averaging. I have had better success sticking with one set of rankings/projections that I trust the most, then adjusting that by looking at discrepancies with other sources. Ultimately, my personal rankings supercede all of this, but they are certainly affected most by my #1 source and to a lesser extent by the other sources.

 

[Musesboy]

I think it comes down to how good you think you are at ranking players, or how good you think your experts of choice are.

In theory, if every person drafted from exactly the same rankings, most of the drafts would look the same. Unless one owner decides on three kickers or something strange, any potential draft advantage would be negated as the order would be so similar. Luck would come into play a lot more than it already does.

So for owners that look at the draft as a way to gain an advantage, I would suggest going with the picks that you agree with strongly. If you or your expert rank Rudi Johnson 5th and you draft 5th, take the guy. If you don't trust those rankings so much, go with the average. I just see taking that approach as a wasted opportunity though. FF is fun, and even more fun if you win. But I prefer to win or lose on my own terms by taking a stance on players, trades, lineups and any other decision.

To me, going with the consensus ranking is bringing you back to the average level. Going against the trend, even by a tiny amount, is the difference between a good or bad player and everyone else. It may not always work out, but it will separate you from the pack, and any wins you get will be your own.

Use the consensus rankings along with the ADP as a guide for how others will draft, and then make your own alterations where you disagree and put your own stamp on your team.

 

[barndog]

A rather large set of experimental data indicates groups do better at answering questions/predicting outcomes than individuals. Of course, what you are talking about here is not a group prediction but indiv predictions merged. Perhaps what we should do is ask FBG to get their rankers all around the table together once a week.

I don't dispute this, yet averaging the numbers mathematically is not equivalent to group decision-making, at least as far as I see it.

And as I see it

 

[Post Corner]

Well, it all depends on the quality of your expert. Averaging can be good as long as all three are very good and you are not dilluting your rankings. After all, why average a F with a A- and A to bring your overall to a C+??

I wouldn't really just average all three blindly though. Personally, I think a "floating" average with your own input works best, but again need to start with 3 high quality rankings. I like to look at each ranking individually... it helps show "sleepers" and lower round value better IMO. I then average based on my own impression and information.

As for sources, that is pretty easy. You simply keep the rankings/projections from the first year and compare to year end results relative to other sites. Also, rankings + projections are much harder than just rankings. Projections hold rankings in check by opening those sites to further criticism by the masses... so they had better darn well have their ducks in a row before releasing projections. If Tom Brady has 40 TDs but his WRs, RBs, and TEs have a total of 20.... not good quality. Again, masses will scream. If just do rankings, you can just pull crud out of your butt with just rankings and there really isn't any screaming from the masses outside of picks that are out of the norm.

SO, having said that. I play in high stakes leagues and do keep track of this stuff -- three sites have earned my trust (not a coincidence all include rankings + projections).

http://www.footballguys.com

http://www.footballdocs.com

http://www.fftoday.com

Again, the above with your own tweaking with weighted averages is the best route to go. My two cents.

 

[bostonfred]

Look at the guys that each has well ahead of ADP. Check those players out a little and see if you agree. Do the same for the guys you like better than their ADP. Try to get as many of them as you can.

 

[FFFIEND]

I have to agree with Musesboy and Post Corner.

Solid advice from the Forums here at fbg. (you just have to know how to filter the advice you get as well.)

It is all about good judgement and your gut.

 

[ComfortablyNumb]

This is an excellent thread because he who has the the most accurate projections clearly has the edge during the draft. FBGs has done an great job of developing tools (VBD, DD, Projections Dominator), but I'd like to see them place a little more emphasis on current and accurate projections. Maybe I'm just ignorant of the FBG process; I'm sure David Dodds isn't just sitting in his basement pulling numbers out of his ####. However, the importance of good projections cannot be overstated - they are the foundation of EVERYTHING (VBD, rankings, cheatsheets, etc.). I guess I'd just like to see a little more emphasis in (or maybe just a clear explanation of) the FBG projections process.

 

[Doug Drinen]

I'm sure David Dodds isn't just sitting in his basement pulling numbers out of his ####. However, the importance of good projections cannot be overstated - they are the foundation of EVERYTHING (VBD, rankings, cheatsheets, etc.). I guess I'd just like to see a little more emphasis in (or maybe just a clear explanation of) the FBG projections process.

Here are a couple of quotes from Dodds in similar threads:

I can assure you that I take my projections very seriously. My first cut projections take nearly 40 hours and match to NFL norms in virtually every category (ie add up all the RB yards and YPC is with 99% of league wide numbers, etc).Here is my process:- From the team view, run the last three years of data as a baseline.- Analyze differences in coaching philosophy, personnel, off-season moves, ages of players, etc to determine run/pass team prediction.- Make sure all teams added up fit with in NFL norms. Normalize the data to make this happen if needed.- Begin team by team to fit numbers into this team view keeping historical ypc, catches per year, role in offense, etc consistent with team numbers generated above.- When all teams are done, I then convert to rankings.- I also do checks for 1,000 yard rushers, 1,000 yard passers, and comparison to last three year's AVT numbers to ensure the data matches up from a historical perspective.

Because I try and analyze everything going on with a team. What changed from 2005 to 2006 (coaching, free agents, rookies, schemes, etc)? Will a team run or pass more? How will these touches be allocated? And I attempt to do it against what a player has shown in his history for yards per carry, etc. Additionally, I like to look at team history to get a feel for tendencies / opportunities / production success.and then after I do that for every team, I make sure all of the team data adds up. For every pass there is a reception. Some are not listed in these numbers (about 40 rushing yards and up to 100 receiving yards are lumped as other, but I definitely have accounted for every touch).I then look at the rank by position and see if it matches up closely with three year AVT data. I don't look for a perfect match, but I want the curves to approximate reality especially around picks 12-30 for a position.I then make sure that the aggregate list of data matches up with three year norms for a wide variety of stats.Here are those aggregatesass Attempts: We project 16,478 for this season.NFL data - 16,493 (2003), 16,354 (2004), 16,465 (2005)Pass Completions/Receptions: We project 9,793NFL data - 9,695 (2003), 9,772 (2004), 9,790 (2005)Passing Percentage: We project 59.43%NFL data - 58.8% (2003), 59.8% (2004), 59.5% (2005)Passing/Receiving Yards: We project 112,084NFL data - 109,467 (2003), 115,338 (2004), 11,721 (2005)Passing Average: We project 6.80 yards per attemptNFL data - 6.64 (2003), 7.05 (2004), 6.79 (2005)Passing/Receiving TDs: We project 678NFL data - 654 (2003), 732 (2004), 644 (2005)Interceptions: We project 516NFL data - 538 (2003), 524 (2004), 507 (2005)Rush Attempts: We project 14,461NFL data - 14,508 (2003), 14,428 (2004), 14,375 (2005)Rush Yards: We project 58,74160,341 (2003), 59,709 (2004), 57,583 (2005)Rushing Average: We project 4.04 yards per carryNFL data - 4.16 (2003), 4.14 (2004), 4.01 (2005)Rushing TDs: We project 423427 (2003), 416 (2004), 431 (2005)

 

[switz]

I actually respect alot of people's rankings, but favor some more than others. So I average the rankings, but actually weight those I favor higher, so they have more of an influence on the output.

Then I compare those rankings with my own statistical projections, and see where we disagree by alot. If we disagree by 2-3 spots, I don't adjust mine at all. But if we disagree by alot, I examine my reasoning for ranking a player at that spot. Then I may adjust my projections if I feel they were way off.

 

[wickster33]

I use expert rankings as a base, but I would never just use someone elses opinions. To quote Tom Hanks in Big, "Whats fun about that"

Its all about sleepers and special situations to show you're smarter than everyone else in the league.

 

[The Jerk]

I use expert rankings as a base, but I would never just use someone elses opinions. To quote Tom Hanks in Big, "Whats fun about that"Its all about sleepers and special situations to show you're smarter than everyone else in the league.

I'm fine with that. However, isn't draft preparation, discussing with other FF afficionados, researching expert opinions part of being "smarter than everyone else in the league", too. I want to put my own stamp on my team, yet my ego's not big enough to believe I'm smarter than everyone else not just in my league but in the entire FF universe.