Projection Accuracy

Edwards Kings · Post by **Edwards Kings** » Wed Feb 07, 2007 12:47 am

A lot of the frequent posters here come from industry and amateur enthusiasts who prepare there own projections. I primarily use projections of others and am an ardent supporter of Baseball HQ (and thanks to Todd Zola's blogs and Perry's insightful postings will probably be joining their site as soon as the bonus comes in). I do use others projections to try and determine additional trends, but that is a whole other story.

All of that to say this...for those who prepare their own product/projections, is there a range of accuracy (i.e. actual results compared to beginning of the season projections) that would lead you to believe your projections were successful? Of course, 100% would be nice, but we all know that will not occur. Do you rank your success on the individual player or a class of position or what? We all know of the less scupulous sites with selective stat proclaimations ("My site is 45% more accurate than the other sites!"), but I was just wondering what level of accuracy (risk and probability) would make you say you had nailed it.

Gordon Gekko · Post by **Gordon Gekko** » Wed Feb 07, 2007 1:07 am

Wayne - if you use others projections, you are only as good as the source. To get the projections you use on draft day, you gotta do your own damn work

Tampa Munchers · Post by **Tampa Munchers** » Wed Feb 07, 2007 1:08 am

What do people think of Shandlers projections and Rotoworlds projections?

Gordon Gekko · Post by **Gordon Gekko** » Wed Feb 07, 2007 1:26 am

I heard a rumor that a successful nfbc owner simply uses the rotoworld cheatsheets to draft his team

Edwards Kings · Post by **Edwards Kings** » Wed Feb 07, 2007 2:12 am

Originally posted by Gordon Gekko:

Wayne - if you use others projections, you are only as good as the source. To get the projections you use on draft day, you gotta do your own damn work I hear you Mark. However, the sources I use I feel can do a better job than I can. I do not follow them blindly and utilize the projections to generate my own "value", but Baseball HQ projections are the basis.

Edwards Kings · Post by **Edwards Kings** » Wed Feb 07, 2007 2:15 am

Originally posted by Tampa Munchers:

What do people think of Shandlers projections and Rotoworlds projections? As I mentioned in several message board strings, I feel the Baseball HQ (Shandler) information is the best (all due respect to Todd and Perry) I have seen. They do not say that there projections are absolute, but primarily base their projections on a players underlying skills.

[ February 07, 2007, 09:47 AM: Message edited by: Edwards Kings ]

cindy · Post by **cindy** » Wed Feb 07, 2007 3:04 am

the problem is that 75% of all owners are using the exact same projections on draft day. so how are those people supposed to get an edge? the overall winner will have to be someone that does not use hq/zola/rotoworld projections. the only way to gain an advantage in this contest is to go off the beat and path because so many are using the SAME map!

devilznj · Post by **devilznj** » Wed Feb 07, 2007 3:32 am

Everyone is going to use one or more publication's projections as a tool. But even if everyone used the exact same projections it should have little effect on the dynamics of the draft. Everyone contructs their team individually placing different priorities on power, speed, strikeouts, etc.

ToddZ · Post by **ToddZ** » Wed Feb 07, 2007 3:34 am

I can say unequivocally that no one uses Zola's projections.

That's a lot of players, we can't agree on everybody

Edwards Kings · Post by **Edwards Kings** » Wed Feb 07, 2007 3:46 am

Originally posted by cindy:

the problem is that 75% of all owners are using the exact same projections on draft day. so how are those people supposed to get an edge? the overall winner will have to be someone that does not use hq/zola/rotoworld projections. the only way to gain an advantage in this contest is to go off the beat and path because so many are using the SAME map! You make a good point, Cindy, but I agree with the Prospectors. We may all have access to the same roadmap(s), but when we get to the end of our driveway, some of us turn left, some turn right. The options and opinions multiply exponentially from there.

JohnZ · Post by **JohnZ** » Wed Feb 07, 2007 5:39 am

Originally posted by cindy:

the problem is that 75% of all owners are using the exact same projections on draft day. so how are those people supposed to get an edge? the overall winner will have to be someone that does not use hq/zola/rotoworld projections. the only way to gain an advantage in this contest is to go off the beat and path because so many are using the SAME map! I would venture to say that number is closer to 50%, and for that reason, you MUST know what those projections are so they can be used to your advantage during the draft. That's why I always start with Ron's numbers and tweak them as I see fit.

I like the premise of how Ron gets his numbers, except for players that defy statistics at times like Vlad, Soriano, and now Howard.

bjoak · Post by **bjoak** » Wed Feb 07, 2007 10:32 am

I kind of agree with Cindy. As I've said before, you can do better with your own less accurate projections because no one else uses them. In other words, if you're wrong all the time, it won't work, but if you're more accurate just 33% of the time you have the advantage of getting a lot of values others aren't looking for.

I get a lot of value with my own pitching stats, not that they are less accurate, but I use sources to help, particularly when I am confused on how to project a peripheral number. ERA's and Wins are my own. K's, not so much. I am best at evaluating pitchers in season. Unfortunately, it is harder to find value on the NFBC FAAB list than in a smaller or trade offering league.

I agree with Edwards to an extent, though, at least half of projections is how you use them, and then if you are using a good combination of sources, that could help.

New York, New York hates it when I disagree with his method, but I feel that the danger in using too many sources equally is that you make all the projections bland. In other words, some systems are going to go out on a limb on players, and, particularly if it is a good system, it can find you value, but you will never get those with an average of five sources.

But I've gone too far today in trying to convince everyone that I take too much risk, so I'll stop now.

[ February 07, 2007, 04:36 PM: Message edited by: bjoak ]

bjoak · Post by **bjoak** » Wed Feb 07, 2007 10:35 am

Originally posted by UFS:

quote:Originally posted by cindy:

the problem is that 75% of all owners are using the exact same projections on draft day. so how are those people supposed to get an edge? the overall winner will have to be someone that does not use hq/zola/rotoworld projections. the only way to gain an advantage in this contest is to go off the beat and path because so many are using the SAME map! I would venture to say that number is closer to 50%, and for that reason, you MUST know what those projections are so they can be used to your advantage during the draft. That's why I always start with Ron's numbers and tweak them as I see fit.

I like the premise of how Ron gets his numbers, except for players that defy statistics at times like Vlad, Soriano, and now Howard. [/QUOTE]I had Howard projected as a first rounder. I don't think he defied anything. But I agree with you on Soriano. That's the human element. I agree that it is extremely helpful to know what everyone else is looking at.

nydownunder · Post by **nydownunder** » Wed Feb 07, 2007 12:09 pm

Originally posted by cindy:

the problem is that 75% of all owners are using the exact same projections on draft day. so how are those people supposed to get an edge? the overall winner will have to be someone that does not use hq/zola/rotoworld projections. the only way to gain an advantage in this contest is to go off the beat and path because so many are using the SAME map! "Exact same projections" - No Chance! I would agree that 75% have similar projections...as most publications have correlations in the low to mid 90's.

For one, anyone who is serious is going to use more than one publications and/or is going to make their own adjustments. And if you are real serious you are going to ensure you are using the better ones or a weighted portion of them. There also several programs out there that people use on draft day as well as some home grown programs (ie Bjoak). How these programs value each category also changes the value of player rankings.

bjoak · Post by **bjoak** » Wed Feb 07, 2007 12:16 pm

Oh, I said New York, New York by accident! Someone must be daydreaming about Vegas.

nydownunder · Post by **nydownunder** » Wed Feb 07, 2007 12:56 pm

Didn't know you meant me Bjoak. You must think I average all 7 sources. Not so at all. Some publications are better projectors than others, both overall and for Batting and/or Pictching, thus I use them acordingly. Having all of them available also allows me to do several other things that I find valuable for the draft. One of them is projecting draft order as closely as possible.

But even with that said, an equal weight of all 7 of the pubclications I have used have had better acuracy than any single publication over the last 2 years...with a lot less risk (variance).

The other thing that many need to be aware of is that several publications are overly conservative, thus under project on a whole. This is bad if you rely heavily on one of them, but good if you use them as to keep your overall projections in check (ie 5-10% weight).

I'd rather have projections closer to the mean, BUT NOT THE MEAN, because its important to build a team across all categories and if you miss one or two categories because you went out on a limb TOO OFTEN with some players' projections in certain categories, then you may struggle to recover in FAAB. The smaller you miss, they easier it is to patch things up. Bench/FAAB is 25-33% of your accumulated stats by year's end!

bjoak · Post by **bjoak** » Wed Feb 07, 2007 2:20 pm

Bench/FAAB is 25-33% of your accumulated stats by year's end! Careful with your pronouns there. FAAB made up 23% of my pitching and 22% of my hitting, but your discussion of the mean indicates that you are referring to guys one needs to replace through faab because of injuries or lack of playing time. You should take underperformance and guys that you just replaced with better players through faab out of the equation because that shouldn't affect your original mean or only affect it positively if your mean was correct. Guys that I *had* to replace for playing time concerns were closer to 7% for me. I account for more than that, but to say I need to account for 30% of replacement level is inaccurate to say the least.

[ February 07, 2007, 08:26 PM: Message edited by: bjoak ]

Crazy Like a Fox · Post by **Crazy Like a Fox** » Wed Feb 07, 2007 2:30 pm

Originally posted by Gordon Gekko:

Wayne - if you use others projections, you are only as good as the source. To get the projections you use on draft day, you gotta do your own damn work I agree 100%. Good post.

nydownunder · Post by **nydownunder** » Wed Feb 07, 2007 2:43 pm

Originally posted by bjoak:

quote: Bench/FAAB is 25-33% of your accumulated stats by year's end! Careful with your pronouns there. FAAB made up 23% of my pitching and 22% of my hitting, but your discussion of the mean indicates that you are referring to guys one needs to replace through faab because of injuries or lack of playing time. You should take underperformance and guys that you just replaced with better players through faab out of the equation because that shouldn't affect your original mean or only affect it positively if your mean was correct. Guys that I *had* to replace for playing time concerns were closer to 7% for me. I account for more than that, but to say I need to account for 30% of replacement level is inaccurate to say the least. [/QUOTE]Huh?????

I'm not talking about replacement level or the 'mean' player. What planet you on? When I used the word mean I am talking about the average value (my value system) of a single player using the 7 publications (ie Pujuls mean is 19). I DO NOT WANT TO VENTURE TOO FAR FROM HIS MEAN IN PROJECTING HIS VALUE. IF YOU HAVE THIS TENDENCY YOU CAN MAKE A MESS OF BUILDING YOUR CATEGORIES BECUASE YOU ARE OVER AND UNDER ACROSS A LARGE POPULATION OF PLAYERS. I would reread because you are off in left field right now in undertsanding what I posted.

As for the 25-33%, it was my way of saying projections aren't everything in this competition. That's it, nothing more nothing less.

bjoak · Post by **bjoak** » Wed Feb 07, 2007 3:08 pm

No need to get riled up; if I had trouble understanding your post there is no need to get excited because you don't understand mine. I thought you meant that your team has to exceed your mean team value to account for the faab players who will be on your team, i.e. you need a .286 average but you have to plan for a .290 average because you will have crappy faabers bringing your average 4 points below .286.

As for what you are saying, I take pains to ensure that I am not overprojecting any more than I am underprojecting. If you have to account one way or the other, your projections are, er, not good.