Comparing CHONE and FANS projections

A while ago Tango wrote about the optimistic fan projections and I followed it up with a look at the fans’ playing time projections. I think the fan projections are very interesting and I wanted to look at another aspect before the season started. I wanted to see, broadly, how the fans’ projections compared with a computer-based system, like CHONE.

Here I plotted the FANS WAR projection versus CHONE WAR projection for each hitter with more than 15 fan projections. I use CHONE as the x-axis – not to say that I think it is the “independent” or “correct” variable, but just because one of them had to go on the x-axis. The red line is NOT the best fit line, but rather the y=x line. If the FANS and CHONE projected exactly the same values everything would fall along the line. Dots above the line are players whose FANS projection is higher and those below whose CHONE projection is higher. I added the names of a couple outlying points.

First off, there is a very strong positive relationship between the two projections: generally, players with a good FANS projection also have a good CHONE, and those with a poor projection of one have a poor one of the other. Not surprising, but reassuring to see.

Next, CHONE is more pessimistic than the FANS (or the FANS are optimistic). Most of the points fall above the line (FANS project higher WAR than CHONE). Interestingly, though, the FANS optimism (or CHONE pessimism) increases for better players. For below-average players (zero- to two-win players) there are a points below and above the line. But as you move up or right on the graph, almost all the players are above the line. The equation of best-fit line, FANS = 0.17 + 1.08*CHONE, bears this out. Since the slope is greater than one high-WAR players will have the greatest difference between FAN and CHONE projections. So the FANS optimism, relative to CHONE, is seen most in the best players.

Finally, I highlighted a couple outliers. The FANS really like Ian Desmond, Elvis Andrus and Evan Longoria compared to CHONE. The difference in each case is largely driven by a difference in defensive projection: the FANS think Andrus and Longoria will be elite defenders, CHONE thinks they will be just good; and the FANS think Desmond will be an average defensive shortstop, but CHONE thinks he will be quite poor. There were two players that CHONE liked considerably more than the FANS: CHONE likes Yuniesky Betancourt to be merely replacement level, while the FANS think he will be about a game worse; and CHONE sees Melky Cabrera as a three-win player while the FANS see him as a 1.5-win player. Finally I noted Troy Tulowitzki since he is one of the few superstars projected higher by CHONE than FANS. Again the difference is driven by defense. CHONE likes him to be an elite defender, while the FANS just a good defender.




Print This Post



Dave Allen's other baseball work can be found at Baseball Analysts.

23 Responses to “Comparing CHONE and FANS projections”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. B says:

    Just throwing out a possible theory here, but I’d guess fans are more likely to project a talented player at a level they’ve been recently, like if a player puts up 2-3 straight seasons of 7 WAR, the fans will project him at 7 WAR, while a computer system like CHONE will project him for less (at least they should and I think they do) because it accounts for the possibility of regression, injury, playing hurt, having a down year, and whatever else, while fans will just see he’s been a 7 WAR player and assume he’ll continue to be one. So that may be a factor contributing to the difference, especially when you start looking at guys at the top of the spectrum like Longoria. If he’s been a +15 and +18.5 defender, fans might assume he should be ~+17 next year, even if they really should expect at least some regression and account for the possibility of something going wrong that makes him play worse….

    Vote -1 Vote +1

    • Dave Allen says:

      This is a good point that I should have brought up in the article. I think you are right that the biggest reason for the difference is that CHONE regresses much more than the FANS do.

      Vote -1 Vote +1

  2. DavidCEisen says:

    CHONE seems to have some really odd UZR projections. It will be interesting to see how these end up comparing to reality. Though UZR is a bit noisy…

    Vote -1 Vote +1

    • Adam says:

      I agree with this. Because the UZR projections seem to be the main point of contention between CHONE and FANS, it would be interesting to see a similar graph with the wRAA projections of the two so we could see how the batting projections stack up against each other. I’m guessing the points would be much closer to the “y=x” line.

      Vote -1 Vote +1

    • tangotiger says:

      CHONE has its own fielding system called TotalZone. It’s not forecasting UZR.

      Vote -1 Vote +1

      • DavidCEisen says:

        ok, its still kind of odd that Ryan Zimmerman is projected to be zero runs above average at third.

        Vote -1 Vote +1

      • Bronnt says:

        That’s almost certainly the entire premise for defensive differences. Fans voting on this site will be most used to UZR, and when voting, recent UZR performances are shown for the player’s recent results as a reference. TotalZone and UZR have differences of opinion that are sometimes rather vast, so it’s unsurprising if those more used to UZR projection differently than a TotalZone projection.

        Vote -1 Vote +1

  3. Rick says:

    I assume this doesn’t control for the differences in playing time? Given that WAR is a counting stat, I would be interested to see how much of the difference is simply due to fans optimism about guys staying healthy…

    Vote -1 Vote +1

    • jfpbookworm says:

      I think if you accounted for projected playing time differences you’d get the slope back toward 1, but the outliers would still be outliers.

      Vote -1 Vote +1

      • RedsManRick says:

        That’s what I would expect and that’s why I would like to see it. Where do the fans see a truly different level of performance versus merely an higher (or lower) amount of playing time?

        Vote -1 Vote +1

  4. The Usual SusBeck says:

    Interesting that both the FANS and CHONE predict Mauer to be 7.3 WAR. I’d expect the fans to be a little higher on him than a projection system.

    Vote -1 Vote +1

  5. Llewdor says:

    The prevalence of Mariners fans here explains the Yuniesky Betancourt projection. Familiarity breeds contempt.

    Vote -1 Vote +1

  6. Jason B says:

    Who are the two dots at the top, that the fans are projecting for 8 WAR? Some combination of Pujols, Lincecum, Greinke, Mauer, and/or Utley?

    Vote -1 Vote +1

  7. Phil Birnbaum says:

    I bet the fans are closer to CHONE’s 45 degree line on hitting than on fielding …

    Vote -1 Vote +1

  8. Jared says:

    The three guys you have pointed out that the FANS like more than CHONE make perfect sense. These are three young guys who have received a fair share of hype. Think of it as the opposite effect that B makes in the first comment. FANS expect(s) (singular or plural verb???) these guys to reach their peak quickly due to the amount of media they have received while the projection systems understand it will take awhile. This is a great social experiment!

    Vote -1 Vote +1

  9. michael says:

    i also think that the fans are much more likely to go out on a limb than the projection systems. for example, a player like billy butler, who perhaps has not shown all of his potential power, will receive a modest chone projection while the fans are much more optimistic that he will live up to his hype as a power hitter.

    Vote -1 Vote +1

  10. joser says:

    As several folks have suggested, it would be interesting to break this down along the component factors that we think show the most divergence: one graph showing FANS vs CHONE for just playing time, another showing just offense, and another showing just defense.

    Vote -1 Vote +1

  11. NWS says:

    Plate Appearances (I actually used AB+BB+HBP for both since CHONE doesn’t have PA projections on their main Fangraphs page)

    http://i43.tinypic.com/vg66b5.jpg

    wOBA

    http://i43.tinypic.com/8xpgkw.jpg

    as you can see, the wOBA projections are extremely consistent with each other, while the Fans project higher PAs, especially at the upper end. None of this is surprising, It just goes to show where the difference in WAR projections are coming from, even before accounting for defense

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *