Tillmanisahero
Footballguy
Historically (Especially this year) who is more accurate? Has anyone kept track?
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!
Who is more accurate on projections? (1 Viewer)
- Thread starter Tillmanisahero
- Start date
Tillmanisahero
Footballguy
apparently no one is keeping track 
dagwood
Footballguy
Bottomfeeder Sports
Footballguy
How do you determine accuracy? Once upon a time I tried to determine how various web pages did with their preseason projections (back in the day when there was a dozen free sites) but I never figured out a good way to do it. That is how many players do you compare? Just the top few? How do you handle when the fantasy points are pretty accurate, but how they got there was way off? So basically what I am saying is that I never figured out how to do such comparisons. Though I did offer some ideas in this thread last season.
Ultimately though, for week to week projections why do I care if the someone gets 90% of the prjections correct, if those that are wrong all concern my team? Isn't this how it seems to work anyway?
Ultimately though, for week to week projections why do I care if the someone gets 90% of the prjections correct, if those that are wrong all concern my team? Isn't this how it seems to work anyway?
Tillmanisahero
Footballguy
Yeah....it is a bit of a like futures but clearly some are better at that than others.
If it is unreasonable to wonder if one person is better at this than another....then why look at rankings at all or why is FFguys better than others ( eg the terrible Fantasy Football Sharks that is attached to MFL)
How would you compare them? Look at week "X" rankings and compare it to rankings for the week in your league after the games are completed. Track that over the a season and it might show a statistically significant difference. Not a stats major but I think that would give you an idea...would it not?
If it is unreasonable to wonder if one person is better at this than another....then why look at rankings at all or why is FFguys better than others ( eg the terrible Fantasy Football Sharks that is attached to MFL)
How would you compare them? Look at week "X" rankings and compare it to rankings for the week in your league after the games are completed. Track that over the a season and it might show a statistically significant difference. Not a stats major but I think that would give you an idea...would it not?
Avery
Footballguy
No many but why would some choose to invest with certain portfolio managers over others? Just because it is a highly unpredictable endeavor, does not mean that one person could be better at predicting overall than another.
dagwood
Footballguy
Maybe you should check out this guys website....he seems to have what you are looking for...
Iwannabeacowboybaby!
Footballguy
I think the question is a valid one. I've looked at their rankings all season long and I'm not sure who's had a better year. What's weird is I can tell who each of them has been a little higher on throughout the season. I could tell before looking at their projections that one would have "this player" a little higher than the other guy. I don't know how you would evaluate, maybe who had more land in their top 10 or top 20 or if you could break it down a little more specific like who landed in their top 10, how many of the predicted from 10 to 20 landed in those actual spots, becuase that's almost just as important.I mean, if you predict Frank Gore to be No. 1 for a week and he lands at No. 19 for the week, to me that's not very good forecasting. If you pick Antonio Bryant to be No. 14 and he finishes 18, that's pretty darn good.
Da Guru
Fair & Balanced
That is true. I have been playing FF for a long time and what I have come to realize is that people really do not listen to projections but want their own thoughts about a player confirmed by an "expert"
Drop
Footballguy
Is this topic a bit too touchy to be answered by the guys in question? If they spend so much time and effort on projections they probably have the math background to check the results, but are they bold enough to show up or expose themselves and/or their co-workers? On top of that, they will have to explain the results and concepts used to determine them to us so we can make some solid conclusions as to how successful they actually are.
PizzaDeliveryGuy
extra sausage
I always go with Bloom, plus he was about 7 of 8 on last minute Twitters.
Tillmanisahero
Footballguy
I asked a similar question last year and it was ignored.
fridayfrenzy
Footballguy
I have often wondered why FBG does not release some sort of hindsight ranking on their weekly or preseason projections. On a site that prides themselves on quality information, it seems odd that they have not done a self evaluation. I realize that this site is not solely based on the projections and rankings, but it is still something which could be looked at.
Perhaps they have done a self evaluation of their rankings and they were not positive, so they were not released.
Perhaps they have done a self evaluation of their rankings and they were not positive, so they were not released.
Chaka
Footballguy
IMO you shouldn't Rankings and projections are useful only if you want to blame someone else for your fantasy team failing (people will seldom credit them when they are correct).
Chaka
Footballguy
There are many reasons to choose FBGs, rankings and projections isn't one of them*.*Every website's rankings and projections are equally useless.
Tillmanisahero
Footballguy
I disagree, some sites do much better. FFguys does a good job. I never meant to imply anything different. I would just be curious which FFguy predictor does the best.
squistion
Footballguy
Exactly. And wouldn't injuries also have to be figured in the calculations? If a player is injured and can't play the entire game, wouldn't he have to be taken out of any study of the accuracy of projections? Or would you leave him in if he played most of the game but was hurt or rested in the last quarter? And what about a QB injured in the course of a game - wouldn't that impact the projections for the corresponding WRs of the team? I will also agree with your later point that even if accuracy could be determined, it can't be extrapolated to an individual team. You are correct that being 90% accurate is worthless to me if the 10% of the projections that were wrong involved players on my team.
To me, projections are worthwhile only for the purposes of cross checking the starting selections I have already made. If there is a big disparity between where I have a player ranked and the projection, then I can take a closer look at things to see if I have ranked the player too high or too low.
Chaka
Footballguy
I didn't say one was better than another just that they are all useless.I imagine that if you were to find a way to quantify accuracy of projections it would show over time that all of them are within the same ballpark overall. That's because it is ultimately guesswork.
Ruffrodys05
IBL Representative
My guess (if any viable results could be mathematically determined with a universally accepted, extremely complex, formula for grading prognosticators accuracy) would be Bloom.
Projections are not useless to me. I'm no expert, but I generally have a very good idea who I will start in each league every week. Having projections from experts to compare my own thoughts to will either confirm my feelings or give me pause. If an experts projections give me pause, I will then take the time to do some more research on my own to try to determine why there is such a discrepancy between my thoughts and the experts. I've already done some research prior to those projections becoming available, but maybe those projections will indicate that I need to do more.
I am pretty confident that most people who use experts projections do not just blindly follow along and make lineup decisions based on those numbers.
Projections are not useless to me. I'm no expert, but I generally have a very good idea who I will start in each league every week. Having projections from experts to compare my own thoughts to will either confirm my feelings or give me pause. If an experts projections give me pause, I will then take the time to do some more research on my own to try to determine why there is such a discrepancy between my thoughts and the experts. I've already done some research prior to those projections becoming available, but maybe those projections will indicate that I need to do more.
I am pretty confident that most people who use experts projections do not just blindly follow along and make lineup decisions based on those numbers.
dagwood
Footballguy
Why don't you take on this task and report back to us.....I think you might learn something from the endeavor.....
Tillmanisahero
Footballguy
I might give it a try....anyone have constructive and realistic suggestions on methodology?
The Comedian
Footballguy
Answer is Bloom.
dollarbill13
Footballguy
I think the Lowball Fantasy Sports model might be what we're looking for. Keeps track of the difference between your projected ranking and their actual ranking... I don't know anything more about it than that.
I also would like to see these results. I'll give some suggestions below on the methodology. I'd be willing to help if someone can round up the raw data for the analysis, and publish the results back on this forum.While past performance doesn't guarantee future results, it is a useful piece of information, whether you are choosing a player or a person making predictions. If somebody has consistently made better projections in the past, odds are they have a better chance of being more accurate in the future.
My initial thought is that you would want to evaluate how much the projections "varied" from the actual results. The statistical term for this is "standard deviation". We would evaluate this at each position QB,RB,WR etc.
My thought is we'd want to evaluate the raw projections (yards, TD's, fumbles, ints, etc.) versus the actual production. This could then be plugged into various point scoring systems to see how it translates to your actual pool (eg. Bloom is normally within 2 points for QB's each week while Dodds is within 2.8 points, etc.) We could also see if one was better in 2007, 2008, or 2009 etc.
First the raw data for projections for each week would be needed for each season. Second the actual performance numbers would be needed. If these were in comma separated value CSV files, they could be imported to Excel or another program to do the analysis.
Hope this helps
Elbows
Last edited by a moderator:
Tillmanisahero
Footballguy
Good post...thanks
Musesboy
Footballguy
There are all kinds of ways to do this.
One option would be to establish an acceptable margin of error (10%, 15% or a set number of points per position) and see how many predictions work on that basis.
But think about what projections are used for. Should I start Player A or Player B? The most important thing is to rank the players in the correct order, whether in the preseason or weekly. If Player A is expected to score 15.2 points and Player B 14.9, the order is the most important factor. If Player A goes on to score 32 and Player B manages 8.5, it doesn't matter that the projections rated them close together (unless comparing players at different positions). It matters that you started the right player.
The same is true in the preseason. If you had Forte ranked 6th, he would likely be drafted ahead of you in every league (or a player you ranked higher fell to you). It doesn't matter that he finished 19th - you still wouldn't have drafted him.
Similarly, if you ranked Addai in the Top 20, you likely ended up with him despite his Top 10 finish.
I'm all for accountability, but I understand why prognosticators choose not to report on their accuracy.
1) It's a lot of work
2) Many subscribers may not understand that a low accuracy rate can still be better than anyone else in the business
I still use my own rankings because I prefer to win or lose, or be right or wrong, on my own merits.
Gut instinct has a lot going for it. Trade offers highlight this point (for me). When I receive an offer, I don't project the remaining games of the players involved; I just know which side of the trade I prefer. I prefer to rely on those instincts during the draft and when setting my lineup.
One option would be to establish an acceptable margin of error (10%, 15% or a set number of points per position) and see how many predictions work on that basis.
But think about what projections are used for. Should I start Player A or Player B? The most important thing is to rank the players in the correct order, whether in the preseason or weekly. If Player A is expected to score 15.2 points and Player B 14.9, the order is the most important factor. If Player A goes on to score 32 and Player B manages 8.5, it doesn't matter that the projections rated them close together (unless comparing players at different positions). It matters that you started the right player.
The same is true in the preseason. If you had Forte ranked 6th, he would likely be drafted ahead of you in every league (or a player you ranked higher fell to you). It doesn't matter that he finished 19th - you still wouldn't have drafted him.
Similarly, if you ranked Addai in the Top 20, you likely ended up with him despite his Top 10 finish.
I'm all for accountability, but I understand why prognosticators choose not to report on their accuracy.
1) It's a lot of work
2) Many subscribers may not understand that a low accuracy rate can still be better than anyone else in the business
I still use my own rankings because I prefer to win or lose, or be right or wrong, on my own merits.
Gut instinct has a lot going for it. Trade offers highlight this point (for me). When I receive an offer, I don't project the remaining games of the players involved; I just know which side of the trade I prefer. I prefer to rely on those instincts during the draft and when setting my lineup.
Bottomfeeder Sports
Footballguy
Follow the link I already gave for a discussion. My link is to my post so you'll want to back track for more insights.
Similar threads
