How accurate are your projections? - Fantasy Baseball Cafe 2014 Fantasy Baseball Cafe
100% Deposit Bonus for Cafe Members!

Return to Baseball Leftovers

How accurate are your projections?

Moderator: Baseball Moderators

Postby Amazinz » Fri Oct 13, 2006 12:03 pm

mak1277 wrote:
TheYanks04 wrote:In a draft it is a similar thing. Can I reasobably expect Smoltz to get back to me if I bypass him her in round 6? What happens if someone else gets him and what is my fallback, etc. The sheets are guides to relative values of one player versus another and a guide that should show you major differences in values pretty clearly.


See, I think this is a crucial concept. If you are only using your own projections to base your draft on, you're going to end up overpaying for certain players and missing out on other because you aren't taking into account the relative value (based on public perception, for lack of a better term).

Just because you have a certain player projected to break out doesn't mean you should draft him based on his projected stats. Just because you think a player is a second rounder based on your projected stats doesn't mean you should draft him there. If everyone else on the planet thinks he should be a 5th rounder, you're overpaying by taking him in round 2.

I think ignoring the "expert" rankings is just as big a mistake as blindly following them.


I agree with this entirely but who uses their projections/rankings in a vacuum?
Image
Maine has a good swing for a pitcher but on anything that moves, he has no chance. And if it's a fastball, it has to be up in the zone. Basically, the pitcher has to hit his bat. - Mike Pelfrey
Amazinz
Mod in Retirement
Mod in Retirement

User avatar
CafeholicFantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeWeb SupporterPick 3 Weekly WinnerSweet 16 SurvivorLucky Ladders Weekly Winner
Posts: 18800
Joined: 16 Mar 2004
Home Cafe: Baseball
Location: in Canada, toughening up figure skaters

Postby cordscords » Fri Oct 13, 2006 12:04 pm

I'm curious to see how many people draft according to their projections, or their predictions? I've got 2 sets of rankings, from my head and from my gut.

For example:

After Beltre had his career year I thought that he would go back to his normal crappy stat lines, which he did. However I don't know for sure if this was a lucky year or if he has finally reached his potential. So for guys like Beltre I always shave off some of their amazing stats to just make them good stats.

So in '05 I had his projection down as: .280, 35 HR

By my prediction was: .250-.260, 20-25 HR

Who else does this?
cordscords
General Manager
General Manager

User avatar
Mock(ing) Drafter
Posts: 4343
(Past Year: 20)
Joined: 23 Jan 2005
Home Cafe: Baseball

Postby mak1277 » Fri Oct 13, 2006 12:18 pm

Amazinz wrote:It seems like this animosity toward people who enjoy creating their own rankings surfaces every year. It's funny that throughout the season the so-called expert sites get bashed regularly but when it comes to their projections they're experts.


My attitude is far from animosity. For anyone who actually does the detailed projections, good for them. I really am more curious to find out if actually going into that much detail really helps when it comes to the end result.


Amazinz wrote:Also, realize that those of you who enjoy doing rankings or fine-tuning someone elses projections are basically doing the same thing. We're all applying our opinions and experiences to a base data set. The major difference is that those who enjoy making their own are taking it several steps further and usually have a more structured approach.


Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that. If the people who do projections are just using gut feeling and throwing stats into a spreadsheet, I'm much less interested in that.
mak1277
General Manager
General Manager

User avatar
CafeholicFantasy Expert
Posts: 4569
(Past Year: 5)
Joined: 14 Nov 2005
Home Cafe: Baseball
Location: Northern Virginia

Postby J35J » Fri Oct 13, 2006 1:10 pm

Amazinz wrote:
mak1277 wrote:
TheYanks04 wrote:In a draft it is a similar thing. Can I reasobably expect Smoltz to get back to me if I bypass him her in round 6? What happens if someone else gets him and what is my fallback, etc. The sheets are guides to relative values of one player versus another and a guide that should show you major differences in values pretty clearly.


See, I think this is a crucial concept. If you are only using your own projections to base your draft on, you're going to end up overpaying for certain players and missing out on other because you aren't taking into account the relative value (based on public perception, for lack of a better term).

Just because you have a certain player projected to break out doesn't mean you should draft him based on his projected stats. Just because you think a player is a second rounder based on your projected stats doesn't mean you should draft him there. If everyone else on the planet thinks he should be a 5th rounder, you're overpaying by taking him in round 2.

I think ignoring the "expert" rankings is just as big a mistake as blindly following them.


I agree with this entirely but who uses their projections/rankings in a vacuum?



Yeah, I don't think anyone has said that they don't look at other rankings and such. I just don't use them to put mine together. After mine are all done I will go to other sources to see where I might be able to wait on some of my guys or know that I might have no chance at other guys because it looks like the experts are too high on other guys.



Jason
J35J
Hall of Fame Hero
Hall of Fame Hero

User avatar
CafeholicFantasy ExpertCafe WriterCafe RankerMock(ing) DrafterPick 3 Weekly Winner
Posts: 10467
(Past Year: 429)
Joined: 26 Jan 2004
Home Cafe: Baseball

Postby Amazinz » Fri Oct 13, 2006 1:11 pm

mak1277 wrote:Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that.


To get the base numbers you can start with the monkey system (50/33/17). This is most likely the base with which a site like RotoTimes begins with and considering how simple it is, it is incredibly accurate. Then I make adjustments for age based on aging trend analysis I have done thanks to the Lahman database. You can actually use Brock2 for both of these steps although I don't. I like my own system not because I think I'm great or better than Bill James but because it is a hobby of mine and enjoy it no matter how futile it might be in the long run.

After age adjustments it's onto ballpark adjustments, adjustments based on analysis of the indicators (BIPA, LD%, etc.) that suggest abnormal production, and analyzing the projections in a team context. I think analyzing a team as a whole is an important step that is totally overlooked by many of the projection services. After that it's all about the fudge.

I do go back and check my projections for accuracy. I almost always come out on top of a site like RotoTimes and comparing it to ESPN is always good for an ego boosting. PECOTA projections are always impressive and I doubt my system is anything compared to theirs but I am not far enough off the mark to abandon it. The most important thing is that I am almost always competitive using my projections. Not that makes me better than the guy using the newsstand rankings. It's all about what you get out of it and I get enjoyment out of that.

There are several threads discussing projections and how people go about them. It was a hot topic in 2004 when I first joined and I remember several good ones. You could probably search for them and find them if you actually interested in doing this yourself. As for giving exact steps and the specific formula that I use for age adjustments, for example, well I'd like to keep some things to myself.
Image
Maine has a good swing for a pitcher but on anything that moves, he has no chance. And if it's a fastball, it has to be up in the zone. Basically, the pitcher has to hit his bat. - Mike Pelfrey
Amazinz
Mod in Retirement
Mod in Retirement

User avatar
CafeholicFantasy ExpertCafe WriterCafe RankerMock(ing) DrafterEagle EyeWeb SupporterPick 3 Weekly WinnerSweet 16 SurvivorLucky Ladders Weekly Winner
Posts: 18800
Joined: 16 Mar 2004
Home Cafe: Baseball
Location: in Canada, toughening up figure skaters

Postby mak1277 » Fri Oct 13, 2006 1:55 pm

Amazinz wrote:
mak1277 wrote:Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that.


To get the base numbers you can start with the monkey system (50/33/17). This is most likely the base with which a site like RotoTimes begins with and considering how simple it is, it is incredibly accurate. Then I make adjustments for age based on aging trend analysis I have done thanks to the Lahman database. You can actually use Brock2 for both of these steps although I don't. I like my own system not because I think I'm great or better than Bill James but because it is a hobby of mine and enjoy it no matter how futile it might be in the long run.

After age adjustments it's onto ballpark adjustments, adjustments based on analysis of the indicators (BIPA, LD%, etc.) that suggest abnormal production, and analyzing the projections in a team context. I think analyzing a team as a whole is an important step that is totally overlooked by many of the projection services. After that it's all about the fudge.

I do go back and check my projections for accuracy. I almost always come out on top of a site like RotoTimes and comparing it to ESPN is always good for an ego boosting. PECOTA projections are always impressive and I doubt my system is anything compared to theirs but I am not far enough off the mark to abandon it. The most important thing is that I am almost always competitive using my projections. Not that makes me better than the guy using the newsstand rankings. It's all about what you get out of it and I get enjoyment out of that.

There are several threads discussing projections and how people go about them. It was a hot topic in 2004 when I first joined and I remember several good ones. You could probably search for them and find them if you actually interested in doing this yourself. As for giving exact steps and the specific formula that I use for age adjustments, for example, well I'd like to keep some things to myself.


Excellent. Thanks for that response...exactly the type of thing I hoped to find.
mak1277
General Manager
General Manager

User avatar
CafeholicFantasy Expert
Posts: 4569
(Past Year: 5)
Joined: 14 Nov 2005
Home Cafe: Baseball
Location: Northern Virginia

Postby J35J » Fri Oct 13, 2006 1:59 pm

Amazinz wrote:
mak1277 wrote:Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that.


To get the base numbers you can start with the monkey system (50/33/17). This is most likely the base with which a site like RotoTimes begins with and considering how simple it is, it is incredibly accurate. Then I make adjustments for age based on aging trend analysis I have done thanks to the Lahman database. You can actually use Brock2 for both of these steps although I don't. I like my own system not because I think I'm great or better than Bill James but because it is a hobby of mine and enjoy it no matter how futile it might be in the long run.

After age adjustments it's onto ballpark adjustments, adjustments based on analysis of the indicators (BIPA, LD%, etc.) that suggest abnormal production, and analyzing the projections in a team context. I think analyzing a team as a whole is an important step that is totally overlooked by many of the projection services. After that it's all about the fudge.

I do go back and check my projections for accuracy. I almost always come out on top of a site like RotoTimes and comparing it to ESPN is always good for an ego boosting. PECOTA projections are always impressive and I doubt my system is anything compared to theirs but I am not far enough off the mark to abandon it. The most important thing is that I am almost always competitive using my projections. Not that makes me better than the guy using the newsstand rankings. It's all about what you get out of it and I get enjoyment out of that.

There are several threads discussing projections and how people go about them. It was a hot topic in 2004 when I first joined and I remember several good ones. You could probably search for them and find them if you actually interested in doing this yourself. As for giving exact steps and the specific formula that I use for age adjustments, for example, well I'd like to keep some things to myself.


I couldn't have said it better myself!! I agree with everything Amazinz just said and what I've been trying to say the whole time but obviously I'm not good at explaining it.

what it comes down to is it is a hobby, we enjoy doing this. My respect for the guys who also take the time to go through it all actually goes up a little.

Jason
J35J
Hall of Fame Hero
Hall of Fame Hero

User avatar
CafeholicFantasy ExpertCafe WriterCafe RankerMock(ing) DrafterPick 3 Weekly Winner
Posts: 10467
(Past Year: 429)
Joined: 26 Jan 2004
Home Cafe: Baseball

Postby mak1277 » Fri Oct 13, 2006 2:43 pm

J35J wrote:
Amazinz wrote:
mak1277 wrote:Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that.


To get the base numbers you can start with the monkey system (50/33/17). This is most likely the base with which a site like RotoTimes begins with and considering how simple it is, it is incredibly accurate. Then I make adjustments for age based on aging trend analysis I have done thanks to the Lahman database. You can actually use Brock2 for both of these steps although I don't. I like my own system not because I think I'm great or better than Bill James but because it is a hobby of mine and enjoy it no matter how futile it might be in the long run.

After age adjustments it's onto ballpark adjustments, adjustments based on analysis of the indicators (BIPA, LD%, etc.) that suggest abnormal production, and analyzing the projections in a team context. I think analyzing a team as a whole is an important step that is totally overlooked by many of the projection services. After that it's all about the fudge.

I do go back and check my projections for accuracy. I almost always come out on top of a site like RotoTimes and comparing it to ESPN is always good for an ego boosting. PECOTA projections are always impressive and I doubt my system is anything compared to theirs but I am not far enough off the mark to abandon it. The most important thing is that I am almost always competitive using my projections. Not that makes me better than the guy using the newsstand rankings. It's all about what you get out of it and I get enjoyment out of that.

There are several threads discussing projections and how people go about them. It was a hot topic in 2004 when I first joined and I remember several good ones. You could probably search for them and find them if you actually interested in doing this yourself. As for giving exact steps and the specific formula that I use for age adjustments, for example, well I'd like to keep some things to myself.


I couldn't have said it better myself!! I agree with everything Amazinz just said and what I've been trying to say the whole time but obviously I'm not good at explaining it.

what it comes down to is it is a hobby, we enjoy doing this. My respect for the guys who also take the time to go through it all actually goes up a little.

Jason


I understand that it's something you like doing, and as far as that goes, good for you. I was hoping to have someone give me a reason (beyond pure enjoyment) for why I should consider doing projections. My only interest is winning my leagues, and I'm on the lookout for ways to do that more consistently. I could never really tell though if people were just "making up" their projections or if they had some type of formula behind them. The answer to that question makes an enormous amount of difference to me.
mak1277
General Manager
General Manager

User avatar
CafeholicFantasy Expert
Posts: 4569
(Past Year: 5)
Joined: 14 Nov 2005
Home Cafe: Baseball
Location: Northern Virginia

Postby TheYanks04 » Fri Oct 13, 2006 3:13 pm

mak1277 wrote:
J35J wrote:
Amazinz wrote:
mak1277 wrote:Another of my reasons for this thread was to figure out what that "structured approach" actually is. I mean, if people out there have time-tested formulas for predicting stats, I'm all for hearing about that.


To get the base numbers you can start with the monkey system (50/33/17). This is most likely the base with which a site like RotoTimes begins with and considering how simple it is, it is incredibly accurate. Then I make adjustments for age based on aging trend analysis I have done thanks to the Lahman database. You can actually use Brock2 for both of these steps although I don't. I like my own system not because I think I'm great or better than Bill James but because it is a hobby of mine and enjoy it no matter how futile it might be in the long run.

After age adjustments it's onto ballpark adjustments, adjustments based on analysis of the indicators (BIPA, LD%, etc.) that suggest abnormal production, and analyzing the projections in a team context. I think analyzing a team as a whole is an important step that is totally overlooked by many of the projection services. After that it's all about the fudge.

I do go back and check my projections for accuracy. I almost always come out on top of a site like RotoTimes and comparing it to ESPN is always good for an ego boosting. PECOTA projections are always impressive and I doubt my system is anything compared to theirs but I am not far enough off the mark to abandon it. The most important thing is that I am almost always competitive using my projections. Not that makes me better than the guy using the newsstand rankings. It's all about what you get out of it and I get enjoyment out of that.

There are several threads discussing projections and how people go about them. It was a hot topic in 2004 when I first joined and I remember several good ones. You could probably search for them and find them if you actually interested in doing this yourself. As for giving exact steps and the specific formula that I use for age adjustments, for example, well I'd like to keep some things to myself.


I couldn't have said it better myself!! I agree with everything Amazinz just said and what I've been trying to say the whole time but obviously I'm not good at explaining it.

what it comes down to is it is a hobby, we enjoy doing this. My respect for the guys who also take the time to go through it all actually goes up a little.

Jason


I understand that it's something you like doing, and as far as that goes, good for you. I was hoping to have someone give me a reason (beyond pure enjoyment) for why I should consider doing projections. My only interest is winning my leagues, and I'm on the lookout for ways to do that more consistently. I could never really tell though if people were just "making up" their projections or if they had some type of formula behind them. The answer to that question makes an enormous amount of difference to me.



There are formulas like the 3 year rule etc. It still is just educated guesswork imo. You can adjust the projections, use 3 year avgs, adjust for ball parks, leagues, etc. It is all statistical nonsense based on mathematical models that someone believes to be accurate.

The problem is that it is like economic forecast models or weather forecast models, you can throw all the Cray supercomputers you want at it and they still can't accurately predict what the economy is like a year from now or what the weather is like next month.

Are they "better" than gut feeling...probably but unlike some here, I do not have that sort of time to spend on something which even in the absolute best case is not proven to be any better than the method you and I use. I have yet to see a pre-season projection list that when the season was over was anything close to what you would consider accurate quantitatively. So again, I am with you.

Why spend inordinate amounts of time and effort to come up with numbers that in the end are no more likely to help you win your league than any of the other dozens of projection lists? Find one that you like pretty much and looks reasonable, tweak it and let someone else do 75% of the work for you.

If someone gets a thrill from running stats through computer formulas, great. I have never seen any evidence that correlates so and so expert's projections to a better winning percentage in fantasy leagues or better performances in fantasy leagues. I am sure if someone had it, they would be publishing it all over the place because there is some serious money to be made there if that were true.

The bottom line is that the projections end up being so far off at end of season as to be almost worthless quantitatively when all is said in done.
TheYanks04
Hall of Fame Hero
Hall of Fame Hero


Posts: 9531
Joined: 12 Dec 2003
Home Cafe: Baseball

Postby GotowarMissAgnes » Fri Oct 13, 2006 5:13 pm

TheYanks04 wrote:There are formulas like the 3 year rule etc. It still is just educated guesswork imo. You can adjust the projections, use 3 year avgs, adjust for ball parks, leagues, etc. It is all statistical nonsense based on mathematical models that someone believes to be accurate.


It's not simply belief. Several threads here have studied which models predict better. You can measure how accurate a system is.

TheYanks04 wrote:
The problem is that it is like economic forecast models or weather forecast models, you can throw all the Cray supercomputers you want at it and they still can't accurately predict what the economy is like a year from now or what the weather is like next month.

Are they "better" than gut feeling...probably but unlike some here, I do not have that sort of time to spend on something which even in the absolute best case is not proven to be any better than the method you and I use. I have yet to see a pre-season projection list that when the season was over was anything close to what you would consider accurate quantitatively. So again, I am with you.

Why spend inordinate amounts of time and effort to come up with numbers that in the end are no more likely to help you win your league than any of the other dozens of projection lists? Find one that you like pretty much and looks reasonable, tweak it and let someone else do 75% of the work for you.

If someone gets a thrill from running stats through computer formulas, great. I have never seen any evidence that correlates so and so expert's projections to a better winning percentage in fantasy leagues or better performances in fantasy leagues. I am sure if someone had it, they would be publishing it all over the place because there is some serious money to be made there if that were true.

The bottom line is that the projections end up being so far off at end of season as to be almost worthless quantitatively when all is said in done.


First of all, I agree with you that the data on accuracy clearly show that this is not really all that much difference right now between a simple projection, like the 3 year rule, and more sophisticated measures....certainly not enough difference to be a bid deal for your typical fantasy player.

But, I do think your perspective misses the point on the utility of projections. Every single projection system will have tons of errors on their individual player projections.

But, you don't draft one player. You typically draft 22-30 players. And the real utility of projections is that you will get a VERY accurate assessment of your team's projected totals if you rely on a decent projection. Because, while the systems are inaccurate for any individual player, the individual player errors tend to balance out. So, if you count up the projections for your whole team, you will tend to get, within a small range of error, a good sense of your team's likely totals.
GotowarMissAgnes
Hall of Fame Hero
Hall of Fame Hero

User avatar
Fantasy Expert
Posts: 5516
Joined: 12 Dec 2003
Home Cafe: Baseball
Location: Happy Valley

PreviousNext

Return to Baseball Leftovers

Who is online

Users browsing this forum: ensanimal and 8 guests

Forums Articles & Tips Sleepers Rankings Leagues


Today's Games
Thursday, Jul. 24
(All times are EST, weather icons show forecast for game time)

Boston at Toronto
(12:37 pm)
Texas at NY Yankees
(1:05 pm)
San Francisco at Philadelphia
(1:05 pm)
Houston at Oakland
(3:35 pm)
Miami at Atlanta
(7:10 pm)
San Diego at Chi Cubs
(8:05 pm)
Cleveland at Kansas City
(8:10 pm)
NY Mets at Milwaukee
(8:10 pm)
Chi White Sox at Minnesota
(8:10 pm)
Detroit at LA Angels
(10:05 pm)
Baltimore at Seattle
(10:10 pm)

  • Fantasy Baseball
  • Article Submissions
  • Privacy Statement
  • Site Survey 
  • Contact