Introducing Oliver

This week, The Hardball Times will be releasing the 2010 version of the Oliver projections as part of THT Forecasts. You will be able to find batting, pitching, fielding, and (coming soon) base running projections and ratings for nearly 8,000 players.

Let’s go over some of the data which you will find listed for each player. First, we have raw statistics going back to 2007, separated by league and team, so that you can see each player’s actual performance over the past few seasons.

This is followed by a single major league equivalency (MLE) for each season, in which the raw statistics have been adjusted for ballparks and leagues. Last summer I wrote an article for Baseball Prospectus in which I compared various ways of doing MLEs, and found that Oliver’s method of directly comparing each player’s minor league to major league performance is significantly more accurate than the chaining (A to AA to AA to MLB) method used by several other projection systems.

Of course, the thing you’re probably most interested in is the projections themselves. Oliver uses a simple weighted mean of the previous three seasons, with aging factors and regression to the mean. Adding extra complication has not proven to add any accuracy—what really makes the system shine is the quality of its minor league translations.

For each player, we’ve listed six years worth of projections to help see how high a young player is expected to peak or when an older one might fade away. For example, the consensus number one prospect in baseball, Jason Heyward of the Atlanta Braves, currently projects to 3.3 Wins Above Replacement (WAR) with a .299/.363/.502 batting line in 2010. By 2015, Oliver expects him to be worth 6.1 WAR, while batting .327/.394/603. On the flipside, Bobby Abreu, who was once a consistent 4-5 WAR player, projects to be worth just 1.9 WAR in 2010, 0.9 in 2011, and 0.2 in 2012.

The WAR statistic for hitters, by the way, is based on both hitting and fielding. To evaluate the latter, I have developed a fielding system which is based on detailed play-by-play data (and yes, it will be updated weekly in-season). Fielding runs above average are also presented in the same Raw/MLE/Projection fashion. Here you can see that Mark Teixeira had an excellent 2008 but an average 2009 saving runs with his glove, but weighting the past three seasons places him behind only Albert Pujols and Casey Kotchman in projected FRAA among first basemen in 2010.

You may be interested to know that the input data set for Oliver includes all season batting and pitching statistics for the major and minor leagues, independent leagues, and Japan back to 1998 (the last MLB expansion), and major and minor league play-by-play back to 2005. Also included are season batting and pitching for most Division I and Division II college teams back to 2002. That’s a lot of data, and it means that Oliver will be able to give you a level of accuracy and detail you won’t find with other systems.

Over the next few weeks, as I finish up the programming and the new season comes nearer, I will be writing articles which will examine some of the inner working of Oliver, measure its accuracy, and show leaders (including minor leaguers) in categories that aren’t found elsewhere. The first thing I’ll do is test the accuracy of my projections for college players; that will be followed by a discussion of aging curves.

As well, we will be adding a lot of data to the Oliver projections over the coming weeks and months. Many of these numbers are already calculated, and we simply have to work out how to best fit them in on the player cards; others will require a little work before they’re ready. Anyway, here’s a list of things I expect to add to the Oliver projections very soon:

{exp:list_maker}Base running statistics are a top priority. They’re tough to program, but I’ve made a lot of progress, and once we have base running, we’ll have outfield arms numbers as well.
Though we currently display only basic fielding runs above average information, my defensive ratings actually have quite a bit of granularity to them. Defenders are given a series of skill ratings, ranging from Excellent to Bad, on outfield range, outfield arms (coming soon), infield range, infield hands, double plays started and double plays pivoted. For more information, you can see this article I wrote last year at Fangraphs, where I described my method of grading infielders for range (preventing groundballs from being hits to the outfield) and hands (preventing infield hits and batters reached on errors). For example, I found that Derek Jeter has very good hands, but only so-so range. This past week Max Machi wrote here at THT about how well infielders ranged to their right or left, and this is something Oliver can incorporate as well. Look for all this data to be added to THT Forecasts soon.
I will also soon be adding left/right splits projections for batters and pitchers, based on the method described in The Book.
Another thing that has already been calculated but is not yet available on the player cards is error bars for the projections. A projection is just our best guess at what a player’s talent and what he will do in the future, but no projection is ever perfect. For that reason, we’ve calculated error bars that tell you both how confident we are that a projection has correctly the player’s true talent and how confident we are that his actual statistics will match what we project (since even if we’re right that Derek Jeter is currently a .305 hitter, luck and random variance mean that he probably will not actually hit .305 in 2010). {/exp:list_maker}

Beyond that, there is a long list of things I’d like to incorporate into Oliver, and as the season progresses, I’m sure some of them will make it into THT Forecasts.

If this all sounds like something you’d be interested in seeing (and it should!), sign up for THT Forecasts today to get access to the Oliver projections and much more.

In addition to writing for The Hardball Times, Brian has written for FanGraphs, consulted for a Major League Baseball team and invented the Oliver projection system. Follow him on Twitter @blcartwright.
Newest Most Voted
Inline Feedbacks
View all comments
14 years ago

Love these, Brian, and I should probably write more words to express that.

But Heyward at 6.1 in 2015?  Certainly quite possible, but as a mean or median projection, that seems really high.  Convince me I’m wrong to doubt you.

Which MLE method do the Davenport Translations use?

Brian Cartwright
14 years ago

I talked to Clay last year at BP’s book signing (which is coming up again in 5 days) and IIRC he said he chained, while PECOTA makes heavy use of comps. We spent quite a bit of time talking about MLEs until the store closed and they had to kick us and Jay Jaffe out.

Heyward’s in just about everyone’s top 3 or 5 prospects. He should hit for average and power, while making excellent contact.

14 years ago

I do have to say that the projections given for the prospects that I have looked at seem a bit… rosey?  Jesus Montero is shown as having a .386 wOBA this year, rising to .450 by 2014.  So basically it is projecting Montero to produce like Barry Bonds in the 90’s as soon as he enters the league?  (And much better than Barry Bonds in the 80’s?)

With that said, the most amusing thing is the following in the forecasts:

“Forecast Updated: 5/22/2010”

Well, the forecasts must be right because… they’re from the future!  Aaaaah!  With that said, a future that only knows an extra couple months of regular season play- so I still think we should be wary of these messages from the future.

Brian Cartwright
14 years ago

We did find an error in the five year forecasts in double counting of aging factors. I have now age normalized the league conversion factors, and will be able to produce a clearly defined ‘current value’ which is how the player would be expected to perform in the majors leagues in 2010, and also a ‘peak value’ which will be most useful to identify the prospects who still aren’t good enough to play in MLB this year.