"What is the evidence, and what does it mean?" Bill JamesThis article was posted to the alt.sports.baseball.tor-bluejays newsgroup March 27, 1999. I want to emphasize that even though I used almost all of the STATS projections that I could, I had to fill in a lot of my own projections (especially for pitchers), and I assumed in my EqA calculations that the league average runs per 9 innings would be the same as in 1998, which might not be the baseline STATS was using.
"Let me take you back to spring training in Dunedin --
I don't think *anyone* would have predicted this team
would be going for 13, 14 games above .500."
-- Brian Williams on CBC, Sept 27/98,
when the Jays were at 13 games above .500
and had one game left in the season.
Welcome to my 3rd-annual "Blue Jays Projected Record" article, where I take the best of the publicly available player projections and add them up to project a win-loss record for the Jays *before* the season starts.
Does this work? Last year it sure did, with the projected 88 wins becoming 88 real ones. (Hint to Mr. Williams: 88-74 is 14 games above .500.) But in general, the chance of the method producing exactly the right number of wins is probably less than 6% (explained below). Two years ago, 83 wins were projected and the team won 7 fewer, which is typical of the error to expect. (These projections are still available in full on the web at
http://www.stephent.com/jays/proj98.html http://www.stephent.com/jays/proj97.html
).
This year's projection is 88 wins once again. This is based on overriding STATS' projections for Delgado, Stewart and Hamilton; otherwise, the projection would have been just 85 wins. If Delgado, Stewart, Cruz, Green and Hamilton all perform at their 2nd-half level of last year, the projection would be 94 wins. A big year for the Jays is possible if most of the players can just pick up from where they left off last year.
When available, I normally used the player projections in the STATS Major League Handbook 1999, including their projections for playing time if possible. STATS' hitter projections last year were found to be the most accurate ones published (see http://www.baseballprospectus.com/news/19990201brooks.html ).
I started by estimating the number of plate appearances in each spot in the lineup as follows:
757, 739, 721, 703, 685, 667, 649, 631, 613
Each spot is 18 less than the previous (162/9). Of course, the actual number of plate appearances will depend on the team's on-base percentage, and that's accounted for later.
I guessed the Jays' starting lineup, and for the leftover plate appearances, guessed the most likely backups. Here are the projections:
{Note: PA is Plate Appearances, estimated from At Bats + Walks;
ERP is Estimated Runs Produced, calculated by me, not STATS
(ERP = 0.16*(3H+2D+4T+6HR+2BB+SB-0.605(AB-H+CS))
alternately, for the '99 Jays,
ERP = (((EqA/.260)^2.5)*((5.06/1.04)*.99)/25.8)*Outs)),
Outs is AB-H+CS,
EqA is Equivalent Average, .260 is average, .300 excellent,
calculated by me, not STATS}
Age
1. LF SStewart 25 606 PA, 91 ERP, 402 Outs, .281 EqA (STATS was .270)
JHurst 27 102 PA, 13 ERP, 68 Outs, .259 EqA (me)
WKirby 35 49 PA, 4 ERP, 36 Outs, .204 EqA (me)
2. 2B HBush 26 546 PA, 53 ERP, 397 Outs, .227 EqA (STATS, PA*4.1)
CGrebeck 34 193 PA, 18 ERP, 133 Outs, .231 EqA (STATS, PA*94%)
3. RF SGreen 26 669 PA, 105 ERP, 448 Outs, .285 EqA (STATS)
WGreene 27 52 PA, 8 ERP, 34 Outs, .281 EqA (STATS, PA*12%)
4. 1B CDelgado 27 624 PA, 112 ERP, 401 Outs, .305 EqA (STATS was .293)
TEvans 24 79 PA, 10 ERP, 52 Outs, .267 EqA (STATS, PA*17%)
5. DH WGreene 27 396 PA, 58 ERP, 256 Outs, .281 EqA (STATS, PA*88%)
CFielder 35 145 PA, 17 ERP, 98 Outs, .251 EqA (me)
GBerroa 34 144 PA, 19 ERP, 97 Outs, .264 EqA (STATS, PA*51%)
6. 3B Fernandez 37 407 PA, 49 ERP, 278 Outs, .255 EqA (STATS, PA*86%)
TEvans 24 260 PA, 34 ERP, 171 Outs, .267 EqA (STATS, PA*57%)
7. CF JCruz 25 600 PA, 87 ERP, 389 Outs, .280 EqA (STATS)
WKirby 35 49 PA, 4 ERP, 36 Outs, .204 EqA (me)
8. C DFletcher 32 413 PA, 48 ERP, 287 Outs, .248 EqA (STATS)
MMatheny 28 157 PA, 11 ERP, 117 Outs, .201 EqA (STATS, PA*44%)
KBrown 26 61 PA, 8 ERP, 41 Outs, .258 EqA (STATS, PA*27%)
9. SS AGonzalez 26 600 PA, 67 ERP, 428 Outs, .243 EqA (STATS)
CGrebeck 34 13 PA, 1 ERP, 9 Outs, .231 EqA (STATS, PA*6%)
--- --- ---
817 ERP 4178 Outs, but see below ==> 809 ERP
Some notes on the above:
809 runs (in 1443 innings) corrected for park factor projects to slightly above average and would have ranked 8th in the league last year. Last year, the Jays' offense was average and did rank 8th.
STATS doesn't run nearly as many pitcher projections so I have to fill in more on my own. In general, pitchers are harder to predict than hitters, so a lot of these projections will probably look silly at the end of the year. The hope is that the errors will largely cancel out.
Note that RA is Runs Allowed per 9 innings, i.e. like ERA but includes unearned runs. I project RA from STATS' ERA by multiplying by 1.088. I estimate R (runs allowed) from RA*IP/9.
I assumed a total of 1443 innings, about the same as the '98 A.L. average.
------------
'99 Starters
------------
Age
36 DWells 216 IP, 3.79 ERA, 4.12 RA, 99 R, 31 GS (STATS for NYY + 1%)
30 PHentgen 206 IP, 4.50 ERA, 4.90 RA, 112 R, 31 GS (STATS)
28 JHamilton 209 IP, 4.65 ERA, 5.06 RA, 118 R, 33 GS (STATS was 5.65 RA)
23 KEscobar 193 IP, 3.73 ERA, 4.06 RA, 87 R, 28 GS (me, '98 * ~2.4)
24 CCarpenter 211 IP, 4.19 ERA, 4.56 RA, 107 R, 32 GS (me, '98 ~2nd-half*2)
22 RHalladay 46 IP, 4.42 ERA, 4.81 RA, 25 R, 7 GS (me, 7 starts to fill)
----
1081 IP, 4.19 ERA, 4.56 RA, 548 R, 162 GS
Notes:
-------------
'99 Relievers
-------------
Age
37 DPlesac 50 IP, 3.60 ERA, 3.92 RA, 22 R (STATS)
30 PQuantrill 42 IP, 3.14 ERA, 3.42 RA, 16 R (STATS, IP*50%)
32 BRisley 38 IP, 4.50 ERA, 4.90 RA, 21 R (STATS)
32 GLloyd 41 IP, 3.77 ERA, 4.10 RA, 19 R (STATS for NYY + 1%)
29 RPerson 67 IP, 4.94 ERA, 5.38 RA, 40 R (me based on career ERC)
26 NRodriguez 40 IP, 4.89 ERA, 5.32 RA, 24 R (me based on career ERC)
22 RHalladay 84 IP, 4.42 ERA, 4.81 RA, 45 R (me, long relief)
---
362 IP, 4.27 ERA, 4.65 RA, 187 R
Notes:
Altogether the Jays project to allow 735 runs, which suggests they'll have the 4th-best pitching & defense in the league, same as last year.
If we apply the Pythagorean formula we get the following projection:
809^1.83 / ( 809^1.83 + 735^1.83 ) == .544 WPct ==> 88 wins, 74 losses
Last year, 88 wins was 4th in the A.L., tied with Texas, who went to the playoffs, but behind the wildcard team (the Red Sox with 92 wins last year).
If you use STATS' projections for Delgado, Stewart and Hamilton in place of mine, you get that the Jays will just win 85 games. But 88 wins is my official projection once again.
What if Delgado, Stewart, Cruz, Green and Hamilton all perform the same as they did in the 2nd-half of last year? Then the projection would jump to 94 wins, a .580 percentage, the same as the Jays' actual winning percentage in the 2nd-half of '98. So the Jays could have a big year if a lot of their players can just pick up where they left off last year.
You might think that if you know the "probability" of a team to win each game, you can project its final record very accurately. For example, if you "know" a team has 50% chance of winning each game, you might think that team is certain to win 81 games (out of 162).
But in fact such a team only has a 6% chance of winning exactly 81 games, and there's a 31% chance it will miss that total by 7 or more, i.e. win 74 or fewer, or 88 or more. 162 games is not big enough a sample for the win-loss totals to always even out.
So even if I projected every player accurately, and even if from that I predicted the team's probability of winning a typical game accurately, chances are my win-loss total would still be off by several wins. That's one reason last year's success is unlikely to be repeated.
But a team projected to win 88 games should still be a better bet to snag a playoff spot than one projected to win just 78. This team is a contender, just like last year's. Believe otherwise at your peril.
-- Stephen Tomlinson http://www.stephent.com/jays/ mailto:stephent@ottawa.com Ottawa, Ontario "What is the evidence, and what does it mean?" (Bill James)
Access count for this page:
Article Written 1999 Mar 27 (HTML formatting added 1999 Apr 4)
Comments are welcome at comments@stephent.com.