"What is the evidence, and what does it mean?" Bill JamesThis article was mailed to the Blue Jays' mailing list Jan 30, 1997.
Date: Thu, 30 Jan 1997 22:25:52 -0500 From: Stephen Tomlinson To: jays@hivnet.ubc.ca Subject: Jays Project to 83-79 record
The Jays project to an 83-79 record in 1997. That's 5 wins short of a wildcard spot, based on last year.
The '97 Jays project to score 4.73 runs per game, which would have been 12th in the American League last year.
The '97 Jays project to allow 4.62 runs per game, which would have been 1st in the American League last year.
(4.73^2) / (4.73^2 + 4.62^2) = .512 = 83 wins, 79 losses
Whenever available, I used the 1997 player projections in the 1997 STATS Major League Handbook. I made a few adjustments when playing time wasn't available or for pitchers from the National League, all explained below. For players with no projections I typically used last year's stats.
For estimating Jays' offense, I assumed the same number of team plate appearances as last year (but corrected later for the possibility of this being a significantly wrong estimate). On average, spots in the lineup will have 18 more plate appearances than their following spot (162/9), so the lineup spots got this number of plate appearances:
753, 735, 717, 699, 681, 663, 645, 627, 609
Then I guessed Cito's starting lineup, and for the leftover plate appearances, guessed the most likely backup. Here are the projections:
Age
1. CF 38 Nixon 452 PA, 52 ERP, 72 R, 30 RBI, 4.37 RPG (STATS projection)
CF 23 Stewart 301 PA, 38 ERP, 41 R, 26 RBI, 4.83 RPG (61% of STATS)
2. 2B 29 Garcia 490 PA, 55 ERP, 61 R, 49 RBI, 4.10 RPG (STATS as Pirate)
2B 24 Crespo 245 PA, 30 ERP, 29 R, 24 RBI, 4.76 RPG (76% of STATS)
3. DH 25 Delgado 573 PA, 87 ERP, 72 R, 88 RBI, 6.13 RPG (STATS)
DH 32 Brumfield 144 PA, 18 ERP, 23 R, 13 RBI, 4.73 RPG (37% of STATS)
4. 1B 37 Carter 645 PA, 78 ERP, 77 R,103 RBI, 4.48 RPG (STATS)
1B 32 Brumfield 54 PA, 7 ERP, 9 R, 5 RBI, 4.73 RPG (14% of STATS)
5. 3B 29 Sprague 640 PA, 72 ERP, 76 R, 79 RBI, 4.18 RPG (STATS)
3B 23 TPerez 41 PA, 4 ERP, 4 R, 3 RBI, 3.65 RPG (10% of STATS)
6. LF 30 Merced 566 PA, 77 ERP, 74 R, 76 RBI, 5.46 RPG (STATS as Pirate)
LF 28 RPerez 97 PA, 11 ERP, 10 R, 10 RBI, 4.14 RPG (32% of STATS)
7. C 32 Santiago 452 PA, 52 ERP, 50 R, 56 RBI, 4.37 RPG (STATS as Phillie)
C 36 O'Brien 193 PA, 18 ERP, 16 R, 19 RBI, 3.30 RPG (63% of STATS)
8. RF 24 Green 492 PA, 70 ERP, 64 R, 57 RBI, 5.60 RPG (STATS)
RF 32 Brumfield 135 PA, 17 ERP, 21 R, 12 RBI, 4.73 RPG (35% of STATS)
9. SS 24 Gonzalez 609 PA, 72 ERP, 75 R, 66 RBI, 4.46 RPG (99% of STATS)
---
758 ERP, 774 R, 716 RBI, 4.73 RPG
As you can see, Brumfield got spread over 3 spots, but I still didn't use all of the plate appearances STATS projected for him. I did not adjust the projections for Garcia, Merced or Santiago, even though STATS didn't know they would be Jays -- I don't think any major adjustment is needed. There's no backup shortstop because 99% of Gonzalez's projection used up all of the #9 spot's plate appearances.
To correct for wrongly estimating the number of plate appearances, I took the weighted average of the RPG projections for each player (weighted by plate appearances, which isn't really proper since RPG is on a per out instead of per plate appearance basis, but when I looked at a Runs Per Plate Appearance measure, it didn't make a significant difference except for superstars, and the Jays' don't have any in their lineup).
Note that RPG is based on ERP (Estimated Runs Produced, from a formula I used on the projected stats). I did not use the projected Runs scored, which would lead to projecting an extra 1-2 wins, nor use the projected RBI, which, after inflating for the normal ratio of Runs to RBI, would project a fraction fewer wins.
I did not use any of the projections for Sandy Martinez (I project him for Syracuse), nor any of the projections for Juan Samuel (last I checked he wasn't on the 40-man roster).
For pitchers I assumed 9 innings per game, 1458 in total (perhaps a slight overestimate, actually, but I just use the ratio later, so it's not much of a concern). I used STATS' projections for the Jays' first 4 starters, then guessed Flener to win the 5th starter spot, if for no other reason than to balance 4 hard-throwing righties in the rotation. The projections for the top 5 starters had 10 starts left over, so I gave them to Andujar when called up from Triple-A. I left Janzen in Triple-A all year.
For pitchers with no projections, I normally used last year's TRA, i.e. I took their total runs allowed (including unearned runs) per 9 innings pitched. From TRA I estimated ERA as 91% of TRA. Cosmetically this gives different ERAs than last year, but overall pretty similar. STATS did not project runs allowed, just ERA and Innings Pitched, so I estimated TRA as ERA + 10%.
Age 28 Hentgen 252 IP, 4.11 ERA, 127 R, 35 GS (STATS) 34 Clemens 214 IP, 3.83 ERA, 100 R, 31 GS (STATS) 32 Hanson 213 IP, 4.77 ERA, 124 R, 34 GS (STATS) 30 Guzman 176 IP, 3.84 ERA, 83 R, 27 GS (STATS) 28 Flener 157 IP, 4.63 ERA, 89 R, 25 GS (Last year's IP, GS and ML TRA) 24 Andujar 55 IP, 5.08 ERA, 34 R, 10 GS (ML TRA, IP/start)
For Flener I added up his innings and games started in both majors and minors, but just used his '96 Major League TRA to estimate ERA and runs allowed. For Andujar, the games started was just the 10 left over, so I used his typical innings pitched per start and TRA over his major league career.
The Jays have lots of relievers. I think Timlin, Crabtree, Spoljaric, Risley and Plesac are virtual locks for a bullpen spot. That leaves at most one spot for Williams, Person and Quantrill. I'm picking Williams to make the team, Person back to Triple-A, and Quantrill to be traded or released:
35 Plesac 70 IP, 4.29 ERA, 37 R (STATS + 15% to ERA) 27 Crabtree 67 IP, 3.16 ERA, 26 R (Last year's runs and IP) 26 Spoljaric 65 IP, 3.66 ERA, 29 R (Last year's TRA + 27 IP (minors)) 30 Williams 59 IP, 4.58 ERA, 33 R (Last year's IP and TRA) 31 Timlin 54 IP, 3.83 ERA, 25 R (STATS) 30 Risley 50 IP, 3.93 ERA, 24 R (Last year's TRA + 8 IP(Minor Lg)) 27 Person 26 IP, 5.25 ERA, 17 R (Last year's TRA + 15%, leftover innings)
As indicated above, I added 15% to Plesac's projected ERA to correct for switching leagues. I estimated Spoljaric's innings by adding last year's Majors' plus Minors' innings, but just used his Major League TRA. For Risley I tacked on the 8 innings he worked in the minors, which gave a total close to his career norm. 26 innings were left over, so I guessed Person to be called up for those.
On the whole, the pitching looks awesome:
Starters: 1067 IP, 4.27 ERA, 557 R, 6.6 IP/start
Relievers: 391 IP, 4.00 ERA, 191 R
Total: 1458 IP, 4.20 ERA, 748 R, 4.62 RPG
Last year Cleveland's 4.34 ERA led the American League. The Jays project to do better this year, at 4.20.
Of course, it's impossible to accurately predict a teams' number of wins every year. I've seen studies that show even if you program in the final stats for the players and replay the season on a computer, you can get wildly different final standings (I think the '93 Jays did anywhere from 74 to 103 wins over hundreds of projected 1993 A.L. seasons, if memory of an article I read serves).
An optimist might figure most Jays' hitters to beat their projections because the team is excited about this year and will prepare more diligently. This might even be true.
If you want to adjust the above predictions, a handy guideline is that 10 more runs equals one win.
-- Stephen Tomlinson Ottawa, Ontario "What is the evidence, and what does it mean?" (Bill James)
Access count for this page:
Last Updated: 1997 Mar 31
Comments are welcome at comments@stephent.com.