Replacement Level Yankees Weblog

Disclaimer: If you think this is the official website of the New York Yankees, you're an idiot. Go away.

More Diamond Mind Sims - What do two different projection systems say?
by SG

This post is bumping two of Fabian's prospect posts down, so make sure you scroll down and read his writeups on Sean Henn and Estee Harris.

A couple of months ago, I posted the results of 100 Diamond Mind simulations that I ran with Dan Szymborski's ZiPS projection disk for Diamond Mind baseball, which can be found at Baseball Think Factory.

Since then, there have been some roster changes, so I set up my computer to re-run 1000 more with the latest build of ZiPS. Here were the *results for those 1000 seasons.


ZiPS
                                                       High  Low
Team                    W   L   RF   RA   DIV    WC     W     W
American League
East
Boston Red Sox          96  66  933  762  678.7  210.0  117   79
New York Yankees        91  71  867  772  275.7  379.5  115   71
Baltimore Orioles       82  80  838  829   26.3   90.0  103   59
Toronto Blue Jays       81  81  788  782   18.3   77.0   98   63
Tampa Bay Devil Rays    72  90  707  805    1.0    7.8   90   53

Central
Minnesota Twins         81  81  784  787  378.5    6.3   99   58
Chicago White Sox       79  83  783  809  257.8    6.0  101   60
Cleveland Indians       79  83  803  818  255.5    4.5  100   58
Detroit Tigers          75  87  788  835   99.7    2.8   94   54
Kansas City Royals      68  94  750  894    8.5    2.0   91   48

West
Oakland Athletics       88  74  829  749  450.5   77.3  107   67
Los Angeles Angels      88  74  831  774  417.5   82.0  106   68
Seattle Mariners        81  81  796  794  110.0   46.2  100   62
Texas Rangers           75  87  825  896   22.0    8.5   96   54

National League
East
Philadelphia Phillies   90  72  773  689  453.7  137.8  110   60
New York Mets           88  74  759  699  315.7  162.3  109   67
Atlanta Braves          82  80  758  749  118.8   88.3  105   62
Florida Marlins         81  81  713  703   65.5   75.8   99   62
Washington Nationals    80  82  689  697   46.3   34.8  101   58

Central
St. Louis Cardinals     98  64  786  634  929.5   34.5  116   81
Chicago Cubs            84  78  742  699   63.0  207.6  102   58
Milwaukee Brewers       76  86  705  751    2.5   31.2   96   58
Houston Astros          73  89  704  788    3.0   10.0   94   55
Cincinnati Reds         73  89  748  819    2.0    8.0   92   52
Pittsburgh Pirates      68  94  671  794    0.0    2.0   89   41

West
Los Angeles Dodgers     94  68  756  655  841.8   29.2  113   75
San Diego Padres        83  79  745  718  112.0  122.8  103   66
Colorado Rockies        77  85  819  861   27.0   28.8   97   56
San Francisco Giants    75  87  751  807   11.3   17.3   92   55
Arizona Diamondbacks    74  88  728  801    7.8    9.8   94   50

ZiPS likes Boston in the AL East, hates the whole AL Central, sees a dogfight in the AL West, and the AL East runner-up as the wild card. It also likes Philadelphia and the Mets in the NL East, LOVES the Cardinals(which it did last year as well), really likes the Dodgers, and sees the Cubs as the most likely wild card.

The Yankee projection was a bit troubling to me, as they missed the postseason 35% of the time. Of course, a healthy Jason Giambi could make most of that difference up.

In addition to running these with ZiPS, I ran another 1000 with Diamond Mind's own projection disk. Here's some background information about Diamond Mind's projection system. The *results for those sims were:


Diamond Mind
                                                       High  Low
Team                    W   L   RF   RA   DIV    WC     W     W
American League
East
New York Yankees        97  65  893  735  573.8  315.5  119   75
Boston Red Sox          95  67  920  765  407.3  451.0  118   77
Baltimore Orioles       82  80  834  829   17.3   68.7  103   62
Toronto Blue Jays       75  87  800  859    1.5   17.0   95   56
Tampa Bay Devil Rays    68  94  736  860    0.0    2.0   93   49

Central
Minnesota Twins         88  74  846  779  699.5    5.5  108   62
Chicago White Sox       79  83  811  836  116.0   13.5   97   61
Cleveland Indians       79  83  808  820  106.0   11.5   98   59
Detroit Tigers          76  86  813  859   67.0    9.5   94   58
Kansas City Royals      68  94  764  888   11.5    1.0   90   44

West
Oakland Athletics       86  76  867  809  425.5   36.8  108   66
Seattle Mariners        83  79  785  773  238.7   24.7  103   63
Los Angeles Angels      83  79  780  775  226.0   31.0  102   65
Texas Rangers           80  82  854  875  109.8   12.3   98   61

National League
East
Philadelphia Phillies   89  73  855  784  484.8   92.0  107   61
Atlanta Braves          86  76  785  734  276.8   97.6  107   64
Florida Marlins         81  81  752  747  103.0   54.6   98   60
New York Mets           81  81  777  773   97.7   47.6  102   61
Washington Nationals    78  84  747  784   37.7   24.8   96   59

Central
St. Louis Cardinals    103  59  879  670  981.0   10.0  121   83
Chicago Cubs            84  78  786  771   17.5  187.4  103   64
Houston Astros          77  85  760  813    0.0   25.8   94   55
Cincinnati Reds         72  90  776  853    1.5    6.3   91   53
Pittsburgh Pirates      71  91  720  804    0.0    5.0   92   49
Milwaukee Brewers       70  92  716  818    0.0    2.0   88   51

West
Los Angeles Dodgers     90  72  807  726  450.8  172.5  109   71
San Francisco Giants    90  72  863  772  445.8  173.8  109   71
San Diego Padres        83  79  786  762  101.3   97.0  109   59
Arizona Diamondbacks    69  93  792  909    1.0    2.0   91   50
Colorado Rockies        69  93  845  978    1.0    1.5   91   50

In the only league that really matters, the Yankees were about 6 wins better on average, and made the postseason 89% of the time. Minnesota projected as a stronger favorite in the Central, Oakland projected as better than Los Angeles of Anaheim of Orange County of California of the United States of North America, but worse overall. Again the AL East runner-up looks like the most likely wild card.

In the inferior league, Philadelphis still projected as the favorite, although Atlanta improved and the Mets plummeted by about eight wins. So Mets fans can take their ZiPS-induced delusions of grandeur with a big grain of salt. San Francisco improved the most of any team, and looks about even with Los Angeles according to Diamond Mind's projection disk. Again the Cubs look like the team with the best chance for the wild card.

*For both sets of runs, W = average wins, L = average losses, RF = average Runs For, RA = average Runs Allowed, Div = division titles won, WC = wild cards won, and High and low W are self-explanatory. I kept ties in there, giving partial credit, which is why those numbers are not all integers.

Upon noticing the divergent results for several teams, I compared the two sets of data.


                        ZiPS - DMB         Combined
                                           Avg   Avg  Total   Total   High   Low
Team                    W    RF    RA      W     L    DIV     WC      Wins   Wins
American League
East           
Baltimore Orioles       0     4     0      82    80     43.6  158.7   103    59
Boston Red Sox          1    13    -3      96    66   1086.0  661.0   118    77
New York Yankees       -6   -26    37      94    68    849.5  695.0   119    71
Tampa Bay Devil Rays    4   -29   -55      70    92      1.0    9.8    93    49
Toronto Blue Jays       6   -12   -77      78    84     19.8   94.0    98    56

Central
Chicago White Sox       0   -28   -27      79    83    373.8   19.5   101    60
Cleveland Indians       0    -5    -2      79    83    361.5   16.0   100    58
Detroit Tigers         -1   -25   -24      76    86    166.7   12.3    94    54
Kansas City Royals      0   -14     6      68    94     20.0    3.0    91    44
Minnesota Twins        -7   -62     8      84    78   1078.0   11.8   108    58

West
Los Angeles Angels      5    51    -1      85    77    643.5  113.0   106    65
Oakland Athletics       2   -38   -60      87    75    876.0  114.1   108    66
Seattle Mariners       -2    11    21      82    80    348.7   70.9   103    62
Texas Rangers          -5   -29    21      77    85    131.8   20.8    98    54

National League
East
Atlanta Braves         -4   -27    15      84    78    395.6   185.9  107    62
Florida Marlins         0   -39   -44      81    81    168.5   130.4   99    60
New York Mets           7   -18   -74      85    77    413.4   209.9  109    61
Philadelphia Phillies   1   -82   -95      89    73    938.5   229.8  110    60
Washington Nationals    2   -58   -87      79    83     84.0    59.6  101    58

Central
Chicago Cubs            0   -44   -72      84    78     80.5   395.0  103    58
Cincinnati Reds         1   -28   -34      73    89      3.5    14.3   92    52
Houston Astros         -4   -56   -25      75    87      3.0    35.8   94    55
Milwaukee Brewers       6   -11   -67      73    89      2.5    33.2   96    51
Pittsburgh Pirates     -3   -49   -10      69    93      0.0     7.0   92    41
St. Louis Cardinals    -5   -93   -36     100    62   1910.5    44.5  121    81

West
Arizona Diamondbacks    5   -64  -108      72    90      8.8    11.8   94    50
Colorado Rockies        8   -26  -117      73    89     28.0    30.3   97    50
Los Angeles Dodgers     4   -51   -71      92    70   1292.6   201.7  113    71
San Diego Padres        0   -41   -44      83    79    213.3   219.8  109    59
San Francisco Giants  -15  -112    35      83    79    457.1   191.1  109    55

Here, W, RF, and RA are the difference between what ZiPS projected and what Diamond Mind projected, The last six columns are the combined totals of the 2000 runs. Obviously, the most striking difference was in San Francisco's projection. I also noticed that Diamond Mind was far more favorable to the Yankees. If you look at the differences in RF and RA, that will give you some insight into why teams performed differently in the two runs.

For example, the Yankees projected to score 26 more runs in Diamond Mind, and allow 37 fewer, an overall run differential improvement of 63. Statistically, 10 runs is roughly equivalent to a win, so the Yankees 6 win improvement from ZiPS to Diamond Mind is due to Diamond Mind having a more favorable projection for both their offense and defense(including pitching).

One of the biggest differences that I am aware of between the two projection systems is that ZiPS uses Voros McCracken's controversial DIPS theory when projecting pitchers. DIPS basically focuses on a pitcher's strikeouts, walks, and home runs allowed, and assumes that their control of hits on balls in play(non-homer hits and outs) is minimal. Tom Tippet of Diamond Mind did his own research on this theory, and concluded that pitchers "have more influence over in-play hit rates than McCracken suggested", so he uses a pitcher's hits allowed totals in his projections.

ZiPS also uses comparisons with similar players in building its projections, whereas Diamond Mind uses a Marcel type projection system which only focuses on what a player himself has done. I am also pretty sure that ZiPS is harsher to older players than Diamond Mind, which would be another reason why San Francisco and the Yankees would project better in the system. I would tend to give Diamond Mind a little more faith because they know the intricacies of their game when they are building projections, not to take anything away from Dan Szymborski, because he does a great service by building his disk and providing it for free.

I always caution about reading too much into projections. Projection systems can only use available data to calculate what a player will do going forward. These are still athletes playing games, and their true talent level can always change. A pitcher can learn a new pitch, or just find themselves (look at Jason Schmidt prior to joining San Francisco for example). Also, injuries always turns out to be a big factor, and Diamond Mind uses a random injury system which would even that out over a large number of sims. I just like running these to get a baseline feel for how the season would play out if teams and players do what they're supposed to do, which never happens in baseball.

I think the more data you have the better, so I'd look at the combination of what the two systems project.

Anyway, in a couple of weeks, none of these things will really matter all that much. Of course, to me that's one of the many great things about the game. First pitch in two weeks.

--posted at 6:27 PM by SG / |

Larry Mahnken and SG's
	Replacement Level Yankees Weblog
"Hey, it's free!"