THE PROTRADE JOURNAL: An Earnings Projection Method for Hitters

And here’s Part 1 of Bosox Bob’s work...using Dan’s ZiPS projections.

In my first pass using this approach, I pro-rated the ZiPS earnings projections for the remaining May-September ABs. The result? The average error dropped to 23.2% - less than either data set’s projections taken separately. This approach is flawed however, as it would overweight the ZiPS projection and underweight any performance level changes as the season progressed.

To counter this, I computed a combined earnings rates from the April and ZiPS data, weighting by ABs to date and projected ABs respectively (aprox. ? April data to ? ZiPS data). This rate was used to project the remaining earnings for the year. With this method, the average error went down to 21.5%. As a check of this projection method, I repeated the process using stats through the end of May, which yielded an average error of 19.2% - more accurate than a month prior.

2008 Hitter Projections

Now that we’ve established a projection method, let’s see what players it currently identifies as good buys. Using 2008 ZiPS and player stats to date for the season, I generated projections for all hitters (note: expected ABs were based on player’s average ABs per team games played, minus games missed on DL). The following table contains all hitters whose pace and projection are at least 25% greater than their current price:

Mets Geek: Peterson: Interview: Tom Tango

New findings from the Tango Investigation Group…

Do you think it’s important that baseball researchers share findings and discuss process? Or can interesting and worthwhile things happen when someone like Bill James does his own thing and isn’t really aware of what others are doing?

Sure, interesting things can get done, but also a whole lot of redundant and terrible things too. Bill James has a few pieces that are simply things that I have already done. Why he would waste his time doing that, I don’t know. But that’s the way he is, by his own admission. Sometimes, you get a pearl out of it. So be it. Researchers at The Hardball Times is the model, as far as the web is concerned. I get immense satisfaction of reading sabermetric works. You’d think all researchers would think like that.

You always seem to be working on the next great statistical project. Tell us what you’re working on right now.

I’ve got a list I posted on my site somewhere called Tango On Demand. Next up for me is creating a Fielding Database. Normally, I would be tacking the PITCHf/x data, but (a) there’s a lot of bright people already there, so I’m happy to be an interested observer at the moment, and (b) the data is still in flux, and I’d rather work on stable datasets.

Comcast: Salfino: Phils Pitching by the Numbers

Live from the Bowl-O-Rama-Slamma-Lamma-Ding-Dong Lanes...the latest from Salfino!

Cole Hamels

ZiPS says, 14 wins, 192 innings, 185 K’s and 50 walks. He’s great. 4:1 K:BB ratio. Only 2:1 this year, but it’s early. At age 24, he should get better. The homer rate has been high, but there are park factors at play. Hamels threw more changeups (34 percent of all pitches) than any starter in the league last year other than Tom Glavine. A number of scouts I’ve spoken to this year consider the changeup to be the most effective complementary pitch, all things being equal. Hamels is clearly one of the top 10 pitches in the sport and reasonably can be grouped in the top half dozen.

Kyle Kendrick

ZiPS only gives him 60-something innings. So we’re going to have to fly solo here, too. I hate to keep beating the same horse here, but when you strike out less than 4.0 per nine innings like Kendrick in ‘07, you are swimming upstream against a swift current. He’ll survive that only if he continues to flash the great control he showed in 2007: (just 2.0 walks per 9 innings). But this year, he’s walked eight guys and K’ed one. Hitters are also crushing liners off him (well above average rates of 21.1 percent in ‘07 and 27.3 percent this year). Oh, and his K-rates were terrible in the minors, too.

Vegas Watch: Somebody’s Gonna Be Wrong

I know, I know...it sounds like a lost title from the Penn-Oldham catalog, but it’s buggy good stuff from the VW boys.

Seattle Mariners
Average: 86.2
High: Steve Phillips (ESPN), 92
Low: PECOTA, 75

Do you think Phillips can spell “Pythagorean”? I don’t.

That being said, 75 is really low. The PECOTA projection does come with something of a disclaimer, since it has Ichiro hitting .303/.346/.384. In 4774 career ABs, Suzuki has hit .333/.379/.437; this is his age 34 season, but thats a huge drop, and Ichiro has outperformed his PECOTA pretty much every year.

Still, even if we bump PECOTA’s projection up to 77, that’s a 15 win difference. And this is far from an isolated incident. The four ESPN guys (Stark, Kurkjian, Olney, and Phillips) and the three Yahoo! guys (Henson, Brown, and Passan) have the winning an average of 90 games. Sheehan, Law, and various computer projections predict an average of 79 victories.

They won 88 games last year, while being outscored by 19 runs. The high predictions employ the “88 wins + Bedard” logic. The others are starting with a baseline of 79, and giving them a boost for Bedard but factoring in some regression for their aging lineup. I don’t think it’s particularly hard to figure out who to side with here.

Comcast: Salfino: A Look at the Phils by the Numbers

From the city that gave us Jim Murgie’s legendary 900 Series...Salfino checks out the hitting zone conditions in Philadelphia.

Chase Utley, 2B
James: .910 (26 homers, 10 steals); PECOTA: .900 (27, 10); ZiPS: .896 (26, 12).

You identify greatness by having such a high and narrow range of projection. I think he’ll steal at least 15 bags; last year’s total dipped only because of time missed due to injury. There is more concern about the dip in power as measured by percentage of fly balls that cleared the wall. His rate of 10.4 percent is now only average and down from about 13 or 14 percent – likely a random variance. Utley hit only eight bombs on the road, but still sported an .886 OPS in those games. The .332 overall average last year was largely a function of great luck on batted balls (not including homers). He converted 36 percent of them into hits last year, about 34 percent in 2006 and 31 percent in 2005. Average is 30 percent. Utley’s range represents a 40-point average swing. Expect him to trade batting average gains for more homers in 2008.

Ryan Howard, 1B
James: 1.068 (53 homers); PECOTA: .957 (44 bombs); ZiPS: 1.028 (50 jacks).

Howard’s homer range is connected directly to his ability to convert fly balls. 2006 likely was a career year because his rate soared to 38 percent. Last year’s 28 percent is still outstanding (almost three times average). I think PECOTA is significantly light here because Howard was the fifth most extreme fly ball hitter in the majors, improving his rate significantly from 2006. The next guy was third most fly ball extreme.

The 2008 Diamond Mind Projection Blowout

Opening Day is almost here, so it’s time to present my annual Diamond Mind Projection blowout. The idea behind this is to take several projection systems and run the 2008 season through Diamond Mind Baseball, which I consider to be the most statistically accurate baseball simulator out there.
[...]
As you can see if you look at the prior runs, the results can be hit and miss, but that’s certainly understandable. This year, I’m using six different projection systems, and I’ve run each one 1000 times for a total of 6000 iterations.

There’s also a part 2 with awesome pie charts. Check out the NL West chart!

SNY: Salfino: Projecting the Yanks’ staff

More colorful than waking up in Ernie Schlegel’s closet...it’s the latest from Mike Salfino!

Andy Pettitte - James: 205/156/3.79; PECOTA: 176/117/4.22; ZiPS: 213/141/4.06. Is Baseball Prospectus right in projecting such a decline in K rate? Well, as I often write, the circus eventually leaves town for everyone. It’s not going to be clowns and cotton candy for Pettitte much longer (especially without Brian McNamee working the snack bar). Pettitte did go down to about 5.6 K’s per nine last year and will be 36 in June. But the K’s spiked to 7.0 per nine innings in the second half of ‘07. Even with those extra whiffs, the ERA and ratio were significantly worse because teams went from hitting a league-average .300 on balls in play to .350 (” ... just past a diving Jeter"). If Pettitte pitches 200 innings, I think he comes closest to James. But I think there is a very good chance he won’t. When guys get hurt there usually are about 30 or 40 terrible innings before they fess up and surrender to the MRI.

Phil Hughes - James: too young to project; PECOTA: 152/129/4.42; ZiPS: 141/106/3.70. Remember when Hughes was the best pitching prospect in baseball? Now most scouts I talk to say his ceiling is “No. 2 starter.” That’s not exactly an insult in scouting parlance, let’s be clear. But that is a clear step down from the 2007 scouting forecasts. What happened? He hurt his hamstring and then his ankle and never ironed out his mechanics to get back some lost velocity. Again, scouts tell me that mid-90s fastball talk was hyperbolic jive. But I see solid-plus velocity and the ability to spot the fastball. And his curve is top notch, too, though the change-up seems to have replaced it as the complementary pitch of choice. Hughes’ change is in the developmental stage, but it shows promise.

SNY: Salfino: Projecting the Mets’ staff

March: on Dick Selma Met staff projections.

Johan Santana: James: 216 innings, 228 K’s, 3.00 ERA; PECOTA: 230 innings, 244 K’s, 3.01 ERA; ZiPS: 234/244/3.04. Note James’ forecast was published when Santana was a Twin. I don’t even bother listing wins because that’s more of a team category, but they all give him 16 to 18 wins. Remember, one pitcher last season, Josh Beckett, had 20 wins. That all the systems give him such good numbers within such a narrow range shows you why Santana is now a very wealthy man. I expect something a little more spectacular: 260 K’s given the times he’ll face a pitcher versus the DH in the AL and an ERA closer to 2.00 than 3.00 given his home park and defense (and the weaker offensive league, though it’s not as weak after this offseason). James says there’s a 17 percent chance Santana pitches a no-hitter, but that’s a career stat. I don’t want to slip into silliness here, but no Mets ever pitching a no-no (is that a double negative?) is a streak that soon should end.

Billy Wagner: James: 69/82/2.74; PECOTA: 58/61/2.77; ZiPS: 73/84/2.71. Looking at modern power relievers, it’s probably safe to assume that Wagner won’t fall off the cliff until age 38 (he’s 36 now).

SNY: Salfino: Projecting the Yanks’ lineup

From James’ Goldmine to Carmine, Salvino...uhh, Salfino cranks out his latest.

Derek Jeter, SS: James: .828 (15 homers, 17 steals); PECOTA: .772 (8, 13); ZiPS: .821 (13, 15). Why does PECOTA hate Jeter so? BP says he turned into a singles hitter in the second half who grounded into too many double plays and, as James demonstrated, lost his baserunning edge. He’s such an extreme ground ball hitter (third-highest rate in baseball) that he has little chance to smack even 20 bombs. The 21 double plays aren’t surprising at all in that context. And even when he hits a fly ball, a below-average percentage cleared the wall. Given the declining speed, batting average projections should be closer to .300. The decline in walk rate probably just means pitchers aren’t afraid to throw him strikes. PECOTA looks right to me.

Cano, 2B: James: .872; PECOTA: .795; ZiPS: .837. Was the second-half power for real? No. He’s still an extreme ground ball hitter. So, as it always was with Jeter for the same reason, the prospect of future power is a mirage. The second-half fly-ball rate was the same as in the first half. But he went from hitting homers on 4 percent of fly balls to 18 percent. Maybe Pettitte and Clemens gave him Brian McNamee’s cell number. Cano did narrow the gap between strikeouts and walks in the second half. But it all screams fluke to me. Bet on PECOTA. Note he’s less than zero on the bases: James has him at minus-10 bags in ‘07.

SNY: Salfino: Projecting the Mets’ lineup

Bah!...Bill “ZiPS” PECOTA used to bat anywhere in the Mets lineup!

Still, I greatly respect the hard work and willingness to be so precise in laying it out there in black and white so that we can debate their relative merits. I’m using three projection systems. The father of the sabermetric movement, Bill James, is represented as published in The Bill James Handbook. We use the PECOTA system of Nate Silver as published in the Baseball Prospectus annual. That system was partially inspired by James’ “player similarity scores” but is more rigidly modeled. And Dan Szymborski’s ZiPS projections come courtesy of BaseballThinkFactory.com, which consistently lives up to its name. ZiPS looks at similar skills more than players in calculating projections.

...So, let’s go right down the projected starting lineup, leading off with shortstop Jose Reyes.

James gives Reyes a .789 OPS with 14 homers and 69 steals. Amazingly, PECOTA tabs him at .788 with 14 homers and 60 steals. ZiPS says 800 with 15 homers and 70 steals.

When did Reyes turn into such a projectable player? None of the three projected Reyes to hit .300 again. I do because his barely average .302 average on balls in play is very low considering his speed. Blame the elevated fly ball rate. That was a bad deal for Reyes because his rate of homers on fly balls was cut in half. His now (just about) 1-to-1 strikeout-to-walk ratio is what you look for from a guy who can contend for a batting title even when they can’t motor like Jose.

Sickels: Not a Rookie: Lastings Milledge

More knock points against the Church of Schneid…

For the future? My opinion about him really hasn’t changed. He has immense potential. He still has weaknesses that his natural talent covers, at least against minor league competition. He is still very young, turning 23 in April. He has flashed the skills to go with those tools, and my personal opinion is that he has a better chance to flesh out those skills than a lot of tools guys. The main thing he needs is to stay healthy.

...On pure stats my expectation is in line with what other people project. However, for some non-objective reason, I think Milledge could break out in a big way this year, with a big batting average spike. It’s just an instinct, one of those impossible to explain gut feelings. Anyway, I put the Gut Feeling projection up there along with the technical projection.

In the long run, I think Milledge will have several seasons where he is one of the best players in the league, but will also have several seasons where he is somewhat disappointing, perhaps due to injuries. Another “gut” thing, I think these seasons will be mixed up rather than a nice glide path...one bad one, one good one, a bad one, a bad one, then two good ones, that sort of thing. In short, I think he will be an occasionally outstanding, but somewhat frustrating player, who 20 years from now will be able to look back on an overall very fine career.

RLYB: Looking Ahead to 2008 - Derek Jeter

Defense

I always hate this part of my Jeter preview, because I always get grief for hating Jeter. So I’m doing something different this year.

If you take the blue pill, you can read about three-time Gold Glove winner Derek Jeter. If you take the red pill you can read about Derek Jeter and the statistics that say he’s not a good defender. If you take the red pill, you forfeit all rights to #####, you were warned.

Blue Pill: Derek Jeter has won three Gold Gloves. You have to watch him play every day to appreciate just how good he is. Stat nerds should take their heads out of their spreadsheets and watch some games. The end.

Red Pill: Derek Jeter’s defense continues to be a sore spot. It’s hard to argue against that when every single play by play defensive metric worth considering says the same basic thing. 2007 was especially bad. According to zone rating, it was the worst defensive season of his career, as Jeter made 27 fewer plays than an average SS, which translates to being 20 runs worse than average.

Which means...I’ll be back on Black Beauties and Yellow Jackets by seasons end!

TSN: Pinto: Buchholz can ease pain of losing Schilling

Or merely the changing of the old schutz?

Assuming Schilling is out and Buchholz throws 180, the remaining 105 would fall, in a best-case scenario, to swingman Julian Tavarez. With Tavarez’s ERA projected to be 4.89, those 105 innings would cost the Red Sox 4.6 earned runs over a league-average starter.

...The gains by Buchholz over Schilling are more than negated by Tavarez, although a six-run difference probably isn’t enough to cost the Red Sox a win. However, at this stage of his career, Buchholz’s ERA projections are going to be very conservative. For every tenth of a point Buchholz’s ERA is lower than 4.11, he picks up two earned runs saved. A 3.81 ERA would negate Tavarez’s deficit. A 3.31 ERA would save 10 more earned runs, enough for an extra team win.

Jim Palmer says Buchholz reminds him of himself. When Palmer was 23, back in 1969, he threw 181 innings (averaging 7 2/3 innings per start) and posted a 2.34 ERA. If Buchholz comes anywhere close to matching that season, the Red Sox won’t miss Schilling at all.

USS Mariner: A’s to win the AL West?

The division favorite was not the Angels but the torn-down Athletics, 47% to 42%, and Texas won the division almost as often as the M’s. The A’s-Angels thing is as much a shock as anything. General analyst-on-TV-or-radio seems to be that it’s all about the M’s-Angels, but Oakland fields the best pitching/defense combination in the AL and their offense is decent too.

Also, check out the follow-up article:

The 93-win Team

I mentioned in passing yesterday that I ran a DMB sim where the M’s won 93 games. I went through and ran them game-by-game and was happily surprised, but as you’ll see, it’s got a lot to do with the advantages of running a season myself compared to the computer managers (I started the year 6-1, for instance). I’ve used the team in the straight sims and it generally sucks. Still, I thought it’d be interesting to talk about how I put that together. Essentially, if you imagine a manager who’s memorized Earl Weaver on Strategy, operating without any kind of conscience or restraint, given control of a team with no consequences, that was the season.

Vegas Watch: Bill Bavasi And Pythagoras Are Not Friends

Although it is believed that Buzzie Bavasi once played pankration with Pythagoras.

The addition of Bedard obviously helps, but losing Jones cancels out a significant portion of that boost. With Ichiro already in center, Seattle goes from having Jones as their right fielder (projected VORP of 21.4, “Very Good” in RF) to having Brad Wilkerson (projected VORP of 4.1, “Average” in RF). So unlike the Santana trade, the Mariners are giving up a significant amount of 2008 value in order to acquire an ace. They’re also losing something in giving up Sherrill, who has held lefties to a .167/.227/.291 line in his career.

But still, Seattle is improving their 2008 roster. And since they won 88 games last year, and play in a relatively week (and small) division, there is a lot of value in a marginal improvement of 2-3 wins.

If they really were an 88-win team in ‘07, this would be a decent argument. But (and I’m obviously not breaking any new ground here, as Law, Sheehan, USSM, and many others have already discussed this) they weren’t. They were outscored by 19 runs- their Pythag record was 79-83. If you’re going from 88 to 91 wins, and significantly increasing your chances of finishing ahead of the Angels and reaching the postseason, this is a defensible trade*. If, as is the case here, you’re going from something like 82 wins to 85, and only marginally upping your odds of playing in October, it’s not.

The predictive value of a team’s RS/RA is not a complicated concept. Seattle scored 19 fewer runs than they allowed in 2007; they were a very average team. It seems absurd for a GM to lack a fundamental understanding of how good his current roster is, but that is the reason this trade went down.

YESNetwork: Keri: PECOTA and the 2008 Yankees

Joba Chamberlain
2007 Stats: 24 IP, 34 K, 6 BB, 0.38 ERA
2008 Projection: 145.2 IP, 162 K, 55 BB, 3.39 ERA

Before you get riled up over this projected drop, consider this: A 3.39 ERA would place Chamberlain 7th among all MLB starting pitchers. For a 22-year-old in his first full big-league season, that’s pretty impressive too. PECOTA, like every other projection system, tends to be conservative at times. In this case, it’s forecasting 15 starts and 50 relief appearances, a nod to the possibility that the Yankees might break him in slowly. Still, the system clearly sees something special in a pitcher whose No. 4 comp is Roger Clemens.

Chien-Ming Wang
2007 Stats: 199.1 IP, 104 K, 59 BB, 3.70 ERA
2008 Projection: 178.1 IP, 95 K, 59 BB, 4.37 ERA

PECOTA generally likes groundball pitchers, in that they usually do a better job of keeping the ball in the park then their flyball-chucking peers. But a strikeout rate of less than five every nine innings doesn’t sit well with pretty much all the projection systems out there — giving up that many balls in play means you need to rely heavily on your defense to help you out. Any time luck and defense start to enter into the equation, you can get outcome swings, like this one, to an ERA well over 4.

Thanks to Was Watching…

Sickels: Not A Rookie: Delmon Young

Another look at Dlemon (oops, another Strat-O-Mat-O scribble mix up) Young.

We don’t have PECOTAs for 2007 to look at yet, but his Sim Scores are fascinating: The top five are Tris Speaker, Rocco Baldelli, Roberto Clemente, Joe Kelley, and Mike Tiernan. When three of your top five Sim Scores are Hall of Famers, you have something going for you. Tiernan was an excellent player in the late 19th century. Baldelli has been disappointing, of course, due to injuries. PECOTA is a better system than Sim Scores though, and I’ll be very interested to see what it comes up with. Clemente seems like the best comp all things considered, in terms of what Young’s ultimate ceiling could be.

The two things to worry about with Young are his strike zone judgment and his attitude. I suspect that he will gradually improve his plate discipline, though this is not something that the Twins really emphasize. It’s not like he has to become a walk machine, but even marginal improvements in his selectivity should help him reach his ultimate potential.

...As for projecting performance, Bill James has Delmon projected to hit .301/.331/.446 this year. ZIPS has him at .292/.323/.424. My numbers are .287/.321/.437. We are all in the same ballpark basically.

Viva El Birdos: lboros: Projecting Scotty

Or...Just how cortisone shot is Scott Rolen?

Under other circumstances, we’d consider that post-cortisone line to be a slump for scotty; as things stand today, it’s cause for mild optimism. if the surgery enables him to sustain that level of performance over 500 at-bats, he’d hit 40 doubles and 17 homers, and contribute about 80 runs and 80 rbi. put that together with superlative defense, and he’s worth the $12 million easy.

but we’re dealing with a small and untrustworthy sample size. to fatten it up slightly, i went back and looked at a couple slices of rolen’s 2006 season. you’ll recall that his shoulder got steadily weaker over the course of the year; by september he needed a few days off to rest it, and by the end of the nlds he required a cortisone injection (which, like the 2007 shot, loosened the joint and immediately improved his performance). i decided to isolate two periods: first, his post-all-star line; second, the period between the all-star break and september 1, when his shoulder became so weak he could barely play. again, i’m groping for a rough approximation of what his physical capabilities might be post-surgery.

Amazin’ Avenue: Simon: 2008 Mets ZiPS Projections

A ZiPPYS Roadside Tour with Eric Simon.

The Bad

ZiPS doesn’t think a whole lot of Ruben Gotay, forecasting a .245/.304/.361 line. Luis Castillo has a .294/.361/.359 projection, continuing his career trend of out-OBP-ing his slugging percentage.

Scott Schoeneweis, Jorge Sosa and Guillermo Mota are all expected to perform worse than an average reliever, though few of us (Willie Randolph and/or Omar Minaya, I’m *not* looking at you) would have predicted otherwise. These guys were the bullpen equivalent of water to a grease fire and are likely to remain as such for the foreseeable future. We don’t need a fancy projection system to know that.

BP : 2007 Pitcher Projection Roundup

The rookie CHONE stacks up nicely against PECOTA on the pitching side…

The best you could have done last year is to bundle PECOTA and CHONE in about a 4:3 ratio. This would have increased your correlation coefficient from .451 using PECOTA alone to .461 with the hybrid version. The other systems wouldn’t really have contributed positively to your results. Taking an average of all eight systems, for example, leaves you with a correlation of .429, which is worse than either PECOTA or CHONE taken alone.

Donnie Sadler gets 50 game drug suspension

NEW YORK—Arizona Diamondbacks infielder Donnie Sadler was suspended for 50 games Saturday after testing positive for a drug of abuse under baseball’s minor league program.

The Book Blog: Tango: How reliable are PECOTA Forecast Percentiles

The latest from Tango…

I suspect that PECOTA first establishes the comparable pitcher list, and once that is established, the unreliability of the comps is thrown out the window.  That is, with Carpenter, we have a solid track record as to how good he is.  We’re pretty sure of it.  The “0.83” of Marcel.  With Wainright, not so much (Marcel reliability of “0.46").  But, once the comp list is created, the uncertainty of those comps is likely no longer considered.

Now, PECOTA does know more about Wainright than Marcel, since Marcel is intentionally oblivious to minor league performance, and PECOTA is not.  But, how much higher can that reliability estimate go?  0.55?  0.60?  Marcel has 135 pitchers with a reliability estimate of at least 0.70.  If we consider all minor league performances as well, can Wainright jump into that pool of pitchers?  I don’t see how he could.