I get some form of the following question every day.

‘Ray, where can I find your 2016 fantasy baseball projections?’

My response is – I don’t have any to offer.

We have rankings, more than 600 players are ranked in the Draft Guide for mixed leagues, AL-only and NL-only leagues for 10, 12 and 15-team leagues. Those rankings, by the way, are updated every day if needed.

So why no series of number “projections?” Let me count the ways.

1 – They are often done in a less than scientific manner. By that I mean we “experts” have the players ranked the way we think they should be. However, often times the numbers spit out by the computer end up getting “massaged” to match the rankings we have in our head before we see those numbers. In effect, the human element changes the projections to fit preconceived notions.

2 – Some people that give projections just make them up. Seriously. I’ve worked at more companies than I would like to admit over my 15 years as a full-time fantasy reporter. I’ve worked at companies where there is no program, no analysis, used in the projections. The writer simply sits down and says “.265-20-70-68-4 sounds right.” I’m 100 percent serious about this. It totally happens.

3 – A three year rolling major league average which is weighted to the most recent season, the “Marcel model” if you look it up, is about as accurate as any system out there. A study by the Hardball Times a few years ago basically pointed out that many of the main projection systems end up with similar results, there was no real difference between them, with ZiPS being the best option.  

4 – Projection systems, the best of them, are about 65-70 percent accurate.

5 – Even the most accurate projection system cannot be accurate for one simple reason (there are of course a myriad of reasons why) – we cannot predict injury/playing time. Heading into last season we thought that a safe bet would have been 25 starts from Drew Smyly, building in that he’s always hurt. Let’s say we predicted he would go 10-4 with a 3.25 ERA, 1.20 WHIP and a K-rate of eight per nine innings. That projection couldn’t have been more wrong since he made just 12 starts. But lookey what happens if we merely double his actual start total last season and give him the same rate of production over those starts.

2015 Projection: 10-4, 3.25 ERA, 1.20 WHIP, 8.0 K/9
2016 doubled: 10-4, 3.11 ERA, 1.17 WHIP, 10.40 K/9

On a per start basis our “projection” for Smyly would be ranked a “9.5” on the 1-10 scale. However, in reality, since he made only 12 start our projection was waaaaay off. Was our projection wrong? Nope, it was a home run if you consider the rate of production. Was it actually accurate though? Well, that depends how you understand the term now doesn’t it?

Don’t forget to pick up your copy of the 2016 MLB Draft Guide which includes more than 600 Players Ranked, auction values, rookie reports, hitting and pitching targets as well as all the info you need if you’re ready to start playing DFS.

6 – Let’s talk some actual projections to further illustrate why they just aren’t overly helpful.

EXAMPLE I

Let’s say Player A is predicted to hit 15 homers.
If he reaches that number you’re happy.
If he hits 20 homers you are giddy.
If he hits 10 homers you are ticked off.

Let’s review.

A 15 homer man hitting 10 or 20 homers is the same thing in terms of predictions. Both numbers are five off our projection – one is up five, one is down five. Your reaction to 10 or 20 homers would be massively different, but that projection is as accurate in either direction.

EXAMPLE II

Let’s say we predict Player A will steal 10 bases.
If he reaches that number you’re happy.
If he steals 15 bases you are giddy.
If he steals five bases you are ticked off.

A 10 steal man reaching 15 steals, or falling to five steals, is just as accurate as both are five off the baseline of 10.

EXAMPLE III

Let’s combine the first two examples.

Let’s say we predict Player A will go 15/10.

As we have seen there can be a 20 or 10 in the homer, and 15 or five in the steals column and be just as accurate. So…

Player A is a 15/10 guy according to our projection.

He could be anywhere from a 20 homer 15 steal guy to a 10 homer five steal guy.

How does that help you out at all?

EXAMPLE IV

Let’s take this in another direction.

Let’s say we projected Joc Pederson to hit .210 with 26 homers, 54 RBI, 67 runs scored and 20 steals last season.

The first four numbers, AVG/HR/RBI/Runs, would have all been 100 percent accurate since those were the numbers Pederson posted in ‘15. However, our projection of 20 steals was a massive failure given that he only stole four bases. Since steals are so important to the fantasy value of a player if our projection was off by 16 steals, even if we were 100 percent right with every other mark, that projection would result in a massive fall off in fantasy value for Pederson. Remember this. The best systems are 65-70 percent accurate. In Example IV we blew way past that rate and were 80 percent accurate, even though we bombed Joc’s steal total which meant he was about seven rounds worse than our expectations would have had him pegged, even though the projection was very accurate.

We could do projections at Fantasy Alarm. We don’t. The reasons, enumerated above, hopefully explain why that is. Projections systems just aren’t accurate enough. We all enjoy looking at numbers/projections, I know I always did, but the fact is that they are nothing more than someone’s guess and don’t speak about the season to come with much accuracy.
 

An audio cut of me talking about projections.

Ray Flowers can be heard Monday through Friday, 7 PM EDT and Friday on SiriusXM Fantasy Sports Radio (Sirius 210, XM 87). You can also hear Ray Sunday nights at 6 PM on the channel talking fantasy sports. Follow Ray’s work at Fantasy Alarm and on Twitter (@baseballguys).