Your 2016 Giants: ZiPS Projections

Fangraphs, as it has been doing in prior seasons, has been publishing Dan Szymborski's ZiPS projections on their website.  Here is the article that discusses some of the information.  Here is their depth chart graphic:

ogc thoughts

Per the depth chart, there are no surprises regarding who are the main roster components on the team:

• SP:  Bumgarner, Cueto, Samardzija, Peavy, Cain
• RP:  Casilla, Romo, Strickland, Osich, Kontos
• Lineup:  Posey, Belt, Panik, Duffy, Crawfor, Pagan, Span, Pence (plus Susac as backup C)

Lopez is also clearly one of the RP and Blanco is also clearly one of the bench players.  Heston, Tomlinson, and Adrianza are most likely the long reliever and backup MI, respectively, barring any breakout performance in spring training by any of the other players invited to spring training.  I think Kyle Blanks is the guy to beat for the last OF bench spot, though Williamson and Parker will get an equal shot at taking the spot.  Blanks just has had good performances in the majors before, hard to beat, but if Parker could keep up what he was doing last season, he could take the spot.

Together, per the depth chart above, they add up to 41 zWAR (unfortunately, Szymborski warns that one should not add them all up and that he'll karate chop anyone who does so, but still, that is what they add up to, so I'm throwing that out there.)

Interesting ZiPS Datapoints

Last year I was shocked and fascinated that Cody Hall had a very high ZiPS projected zWAR, so I thought I would look over the list this season and see if there are anything interesting to note:

• Position Players
• ISO:  Belt leads with 185 (why I still see people hating on him, I can't understand), followed by Pence with 165, but then Blanks is next with 165 (then Crawford 159, Posey 157, Parker 154;  Duffy is at 128).
• OPS+:  Posey leads by a lot with 133, followed by Belt with 122 (again...), Pence with 110, then Blanks with 105.  Duffy is 104, Crawford and Span are 103, Panik 102.  The rest are under 100.
• Def (only names known to Giants fans, and still in Giants system, and at least some chance of being in MLB):  Duffy and Crawford lead with 7, Parker is at 6, Lollis at 4, tied with Belt, Adrianza at 3, Posey at 2, Panik at 1 (really?)
• zWAR:  Posey leads with 6.3 (among leaders in majors), Duffy! at 3.7, Crawford at 3.5, Panik at 2.7, Belt at 2.5, Span at 2.4, Pence at 1.9, Susac at 1.3, Parker at 1.1, Adrianza at 0.9 (and only 331 PA, so he's actually projected to be an average 2 WAR player), Blanco at 0.9, Blanks at 0.5 (only 185 PA, so close to average player), Pagan 0.4, Tomlinson 0.4 (but he's not getting 518 PA).  
• SP
• ERA:  Bumgarner 2.70, Cueto 2.87, Samardzija 3.31, Peavy 3.66, Blackburn 3.81, Heston 4.00, Cain 4.25
• K/BB:  Bumgarner 5.02, Samardzija 4.09, Cueto 3.56, Peavy 3.20, Blackburn 2.83, Cain 2.68,  Heston 2.09
• zWAR:  Bumgarner leads with 4.8 (31 starts), Cueto 4.1 (30 starts), Samardzija 3.0 (29 starts), Peavy 1.4 (24 starts), Blackburn 1.0 (127.2 IP, 21 starts; he's not getting that many unless there is an injury or severe under-performance), Heston 0.9 (26 starts), Cain 0.3 (18 starts)
• RP
• ERA:  Strickland 2.59, Rom 2.82, Casilla 2.86, Broadway 3.19, Kontos 3.39, Osich 3.43, Lopez 3.45, Law 3.58, Gearrin 3.68, Black 3.77, Okert 3.80, Jacob Smith 3.93
• K/BB:  Romo 5.36, Strickland 5.00, Broadway 3.67, Kontos 3.22, Okert 2.50, Casilla 2.47, Gearrin 2.43, Osich 2.35, Smith 2.19, Law 1.87, Black 1.86, Lopez 1.77
• zWAR:  Strickland 0.8, Romo 0.6, Casilla 0.5, Kontos 0.3, Osich 0.2, Broadway 0.2, Law 0.1, Gearrin 0.0

Caveat Regarding Projections

I regard projections as aggregated data that I can use to have an alternative view of the lineup and pitching staff.  I calculate Runs Scored and Runs Allowed averages using the projections and get an idea what each system projects for the Giants.  It provides a pivot point on which to have a conversation regarding what the Giants might do in 2016.  Then I would pivot where I think that any particular player might do better than projected.  For me, it gives me a data point.

As such, in my view, they are not the end all and be all that many seem to treat them.  Each has their pluses and minuses, each has their bias.  For example, Bill James tends to be the highest projections.  Every year, without fail.  Steamer, on the other hand, tends to be on the low side.  Plus, projections cannot handle players coming off injury affected years well - like Cain or Lincecum - because they are just mechanical processes taking a weighted and perhaps nuanced (like accounting for aging or ballpark factors) look at a player's prior performances and projecting forward.  ZiPS seems to go up and down, against my expectations.  

Also, these datapoints work in the aggregate much better than in any particular player.  That's because there will always be errors in the process, but over a large number of players, the ups and downs tend to balance out somewhat.   And that will hopefully get that forecast close to the "correct" answer, which is their actual performance.  And that is the thing that needs to be remembered, the projections are meant to be mechanical with the goal of getting the closest to actual based on the aggregate of all the forecasts.  Of course, they would like to have the actual numbers, but that would take more intelligence in the algorithms that isn't there yet.

When Projections Go Bad

Here is an example of badly a forecast can come out, Hunter Pence (I chose ZiPS because their data was easily available on a historic basis because of these articles, Pence because that is one player I usually have issues with, as you'll see why;  I was originally going to do it for everyone, and I still might, but for now, just Pence).  The first part is the ZiPS projection, the second is his actual performance (from Fangraphs): 

• 2012 (29YO):  ZiPS 654 PA, .290 wOBA;  Actual:  688 PA, 1.5 fWAR, .323 wOBA
• 2013 (30 YO):  ZiPS 665 PA, 2.1 zWAR, .316 wOBA;  Actual:  687, 5.5 fWAR, .356 wOBA
• 2014 (31 YO):  ZiPS 665 PA, 2.6 zWAR, .325 wOBA;  Actual:  708, 4.7 fWAR, .341 wOBA
• 2015 (32 YO):  ZiPS 674 PA, 2.5 zWAR, .327 wOBA;  Actual:  223, 1.6 fWAR, .347 wOBA (had he performed at this level all year, it would have roughly been 669 PA, 4.8 fWAR)
• 2016 (33 YO):  ZiPS 485 PA, 1.9 zWAR, .329 wOBA

Obviously, ZiPS has been far off from what Pence has actually produced per Fangraphs.  As one can see, the projections has trended higher as his performances has continued to stay strong.  ZiPS factors in aging into the projections (many at Fangraphs do it by subtracting 0.5 WAR per year to simulate this aging process), as well as parks and major league equivalents (which is how he got projections for many of the Giants top prospects), in addition to whatever weighted formulation it has for his prior performances.  And the projections (not just ZiPS, but across the board) has been hurt by the fact that Pence's 2012 season was a down year for him, for in 2011 he had 668 PA, 4.3 fWAR, and .377 wOBA.   

A closer projection is if it were based on his trend since 2012, so that would put him somewhere at 3.8-4.3 WAR, depending on how you want to interpret the last two seasons.  That is a far cry from the 1.9 projected for him, which puts him as a slightly below average (where 2.0 WAR is average) player.  Whereas if he is anywhere near 4.0 WAR, he's an All-Star once again.  

Just Remember This:  A Forecast is Just a Forecast, Which Plays the Odds

The thing to remember is that projection systems are playing the odds.  They are not meant to project what that player will do that year.  They are meant to project what that player will do when you make a number of assumptions regarding that player, including aging, park effects, and the like, so that over the large population of players that you are projecting, overall your projections will be off less than other projection systems.  That is, if you had a player just like him, on average (and each system has their way of determining that average), a player in his 33 YO season would decline by X%.  But he might go up N% while another might decline by M%, and so on, and when you average them all out, you will find that they, as a group, declined by X%.  

Still, it gives a starting point to start a discussion or analysis, from which one can pivot to the scenario that one thinks will happen in the future.  So I think it's a useful tool, just misused by many (including me :^).  It is better for looking at aggregate things, like for the team (or if you have a fantasy team), but for one particular player, it will probably be off.  

Giants Prospects

Still, it is interesting to see what the ZiPS methodology says about our players.  Let's take a look at our prospects.

Technically not a prospect, but Andrew Susac has the highest zWAR in the list among non-starters with 1.3 zWAR (only 311 PA, so projected to be almost 3 zWAR player) with a batting line of .224/.305/.347/.652.  Parker is the first one I would call a prospect, and he's listed at 1.1 zWAR (515 PA) with .214/.294/.368/.662.  Adrianza is at 0.9 zWAR (only 331 PA means roughly 2 WAR player, or the mark of an average player, and there are people who don't much of him as a player) and his batting line is .224/.295/.310/.605 (obviously, most of his value is defense).  Kyle Blanks is projected at 0.5 zWAR wtih only 185 PA, which is also roughly average, just short of 2 zWAR, but with a nice .248/.319/.412/.731 batting line.  

The first to have no MLB experience at all is next, Christian Arroyo with 0.4 zWAR (459 PA).  His batting line is projected to be .246/.279/.352/.631.  Tomlinson is next at 0.4 zWAR (518 PA) and only a .239/.294/.307/.601 batting line, showing that the ZiPS methodology does not think that he will be able to keep his BABIP up at that high a pace.  Then there is Mac Williamson 0.3 zWAR (414 PA) and only .233/.302/.361/.663.  Then there are a number of players with major league experience, which I'll skip until I get to Trevor Brown, who is projected at 0.0 zWAR (383 PA).  The rest of the prospects are all negative zWAR.

Among pitchers, Blackburn leads the way with 0.9 zWAR, in 127.2 IP, 3.81 zWAR.   That evefn mor th what Heston is projected at and with less IP than Heston as well.  Next, all the way down the list is Chase Johnson with only 0.2 zWAR and a 4.30 ERA in 115.0 IP.  Ray Black, Derez Law, and Ty Blach are the only other ones with positive zWAR, at 0.1.  Black is projected at 28.2 IP with 3.77 ERA (and huge 30.4% K%;  Strickland and Romo are next with 26.8% K%), Law at 27.2 IP with 3.58 ERA, and Blach at 140.2 IP (I don't really pay much attention to the ZiPS projected IP, but only note because you get a better sense of what the player zWAR production rate is, and where he might be under other circumstances) and 4.41 ERA.  Law is the last prospect projected to have an ERA that is under the average (and thus is above average, in the topsy-turvy world of pitchers).

Projected Team Stats

With a lineup of Span, Panik, Duffy, Posey, Belt, Pence, Crawford, Pagan, I get a Runs Scored average of 4.25 runs scored per game.

With a pitching rotation of Bumgarner, Cueto, Samardzija, Peavy, and Cain, and a bullpen of Casilla, Romo, Lopez, Kontos, Strickland, Osich, and Heston as the long reliever, I got a team ERA of 3.16 and a Runs Allowed of 3.31 (last season the Giants allowed 0.15 unearned runs; they have basically the same defensive team back, which should be improved in the OF with Span manning CF).  Based on the projected wins from ZiPS and the pitchers, there is a winning percentage of .580 which works out to almost 94 wins in a 162-game season.

Putting the RS and RA projections together, Pythagorean says a 4.25 RS and 3.31 RA would result in a 101 win season.  Obviously, I don't think that is a likely projection, but I think it shows the potential of the team at its best.  I think that it's safe to say that a 95+ win season is what I'm expecting.

That's roughly what I was expecting last season, and that's corroborated by how well the team produced when Pence was actually in the starting lineup, 34-17, which works out to a 108 win season.  Again, not likely, as it was small samples, and yet, that's almost a third of a season (which is 54 games) and not that small a sample.  They only had to go 61-50 in the rest of the season to reach 95 wins, which is an 89 win seasonal rate, good but not great.

In any case, at minimum, I have high hopes for the season (again, like last season) and hopefully the core guys can stay healthy enough to play together for more than a third of the season together.  If they can do that, I would feel really good about our chances.

Go Giants!  Team of the 2010's!