2007 Mathematical Definition Of ‘No Chance In Hell’

The good folks over at Replacement Level Yankees Weblog have finished their annual mathematical projections of the 2007 season. In short they run 1000 simulations for the 2007 season using player projections from four different sources including Dan Szymborski’s ZiPS, Baseball Prospectus’s PECOTA, Diamond Mind’s projections and Sean Smith’s CHONE system.

Last season only the first three projections were used and when combined, the Rays were projected to win 71 games and made the playoffs 17 out 3000 seasons or slightly more than 0.5% of the time. The Rays ended up losing 148 games and made the playoffs 0% of the time.

Let’s take a look at the Rays 2007 projections…

• CHONE: The Rays are projected to finish last in the AL East with an average of 71.3 wins, only 0.6 wins behind the Baltimore Orioles. The projections have the Rays winning the division two times and the wild card three times. According to CHONE the Rays will have a big jump in runs scored (689 to 823) but also a slight bump in runs allowed (856 to 915).
• Diamond Mind: The Rays are projected to finish last in the AL East with an average of 69.3 wins, 5.6 wins behind the Baltimore Orioles. The projections may be the most realistic as the Rays failed to reach the playoffs in any of the 1000 seasons. The only other tam with that distinction is Kansas City. According to Diamond Mind the Rays will have a similar runs scored and runs allowed as projected by CHONE (812/930).
• PECOTA: PECOTA has the most ambitious projections for the Rays with a 4th place finish and 77.5 wins, 2.9 games ahead of the Orioles and in the playoffs 36 times (8 division titles and 28 wild card births) or 3.6% of the 1000 seasons. PECOTA predicts a similar bump in runs scored (809) but differs from the others by predicting slightly better runs allowed (839).
• ZIPS: Zips is more similar to the first two projections with 67.6 wins, 11.1 games behind the O’s and in the playoffs four times (1 division and 3 wild cards). Zips predicts the lowest offensive output for the Rays with only a slight bump in runs scored to 758. Like the first two the pitching is expected to surrender more runs (879).

If the four datasets are averaged together, the Rays are projected to finish approximately 71-91, with 11 division titles and 34 wild card appearances in 4000 seasons (1.1%). The Rays are also projected to score 799 runs. As a team the Rays scored 689 runs in 2006, lowest in the majors. The Rays are projected to give up 891 runs. In 2006, the Rays pitchers surrendered 856 runs. Only two teams allowed more.

While it seems realistic that the Rays will score a lot more runs this season, we are not sure why three of the projections have the pitching surrendering more runs in 2007. The staff could not have pitched worse in 2006, yet somehow the projections think they will in 2007. Oddsmakers have the Rays over/under win total at 67. 71 wins is more in line with what we think is realistic for this team in its current form.

Pitcher Projections

Yesterday we presented the Baseball Prospectus PECOTA projections for the Rays position players. Today we take a look at the weighted mean PECOTA projections for the pitchers.

Please see yesterday’s post for an explanation of PECOTA.

Player	W	L	ERA	IP	H	BB	SO
Scott Kazmir	10	7	4.02	151.0	140	62	150
Jae Seo	6	6	5.31	113.1	134	37	65
Casey Fossum	6	7	5.01	112.1	123	46	77
James Shields	10	9	4.49	168.2	183	47	127
J. P. Howell	8	8	4.62	131.0	139	54	100
Edwin Jackson	4	7	5.83	94.1	105	52	65
Jae Kuk Ryu	7	8	4.90	123.1	133	47	85
Seth McClung	3	4	4.54	52.0	48	27	41
Jason Hammel	8	9	4.91	140.2	150	53	100
Travis Harper	2	2	4.66	47.0	53	14	32
Rudy Lugo	3	2	4.34	59.0	58	27	39
Dan Miceli	2	2	4.49	37.2	38	18	28
Jeff Niemann	7	8	5.02	126.1	129	62	100
Chad Orvella	3	2	4.05	50.1	48	20	43
Juan Salas	3	3	5.25	56	58	31	47

A few thoughts on these numbers…

1. Right off the bat we see Scott Kazmir’s numbers and we have to raise our eyebrow. Remember, these numbers are a weighted mean. In other words, PECOTA is saying that there is about a 50% chance of Kid K having a better season than these numbers and about a 50% chance of having a worse season. Kid K’s success in 2006 at such a young age is a rare achievement. A lot of players that have success and throw a lot of innings at a young age have broken down in the following season. So these numbers are probably more a reflection of risk of injury than of talent.
2. Now compare Kazmir’s projection to the Rays other young starter, James Shields. Shields is actually projected to pitch more innings in 2007. That shows that there is more injury risk factored into Kazmir’s numbers. Why? For one, Shields is two years older (25) than Kid K (23), and injuries are less common in older pitchers. The projections are also likely to consider that Kazmir is a power pitcher and throws more pitches than Shields making Kaz more at risk for injury. Again, this doesn’t necessarily say Kaz is going to break down this season. However, injuries and off-seasons are common in players that Kazmir compares favorably to from past years. If he make 28-30 starts this season, his numbers will be better than the weighted mean projection.
3. Jeff Niemann probably does not have enough of a resume to make any realistic projections.
4. Some of these projections are hard to read because PECOTA can’t predict who will be in the rotation and hence a lot of these projections include starter innings and reliever innings. For example, the projection for Jae-Kuk Ryu includes 39 appearances and 17 starts. PECOTA is not predicting that that Ryu will spend half the year in the rotation. Think of it as a simulation and the weighted mean is an average of many simulations. If you simulated the 2007 season 1000 times, some times Ryu is going to be a starter and some he is going to be in the pen. The weighted mean is averaging together both scenarios.
5. The bullpen? Awful. Not a single pitcher with a sub-4.00 ERA projection. That will mean a lot of blown leads…again. What are the symptoms for PTSD?

Hitter Projections

Those lost souls that actually follow the Rays closely know that there is talent on this team and several players have a chance to break out this season. We here at RI try to stay level-headed and not get too optimistic with expectations for individual players. We have an idea of what each player is capable of, and hope that those players at the very least make progress towards those levels. We know that a few players will achieve to those levels, a few players will maintain past levels and probably a few players will regress, whether it be due to injury, past overachievments, steroid testing or baby mamma issues.

That being said we would still like to have a realistic idea of what can be expected from our Rays this season. Enter Baseball Prospectus and their PECOTA playing cards. In short, Nate Silver of BP has developed an algorithm using Sabermetrics to predict future performance of individual players. We don’t pretend to know the math behind PECOTA, but our best understanding is that it uses past performances, age and predicted playing time and compares those numbers to similar players from past years. In doing so, they develop a probability distribution for each player, with predictive stat lines from the 10th percentile to the 90th.

Below we have presented weighted means for players that we expect to receive regular playing time with the 2007 Rays. The weighted mean “incorporates all of the player’s potential outcomes into a single average, weighted based on projected playing time.” For example, Carl Crawford has a 10th percentile stat line that looks like .268-8-41-22. In other words, he has a 90% chance of having a season at least as good as that stat line. His 90th percentile line is .343-20-78-54, but he only has a 10% chance of having a season that good or better. The percentile does not apply to the individual stats, but to the entire line. As Rays fans we realize that CC has a better than 10% chance of stealing 54 bases, but a .343 batting average is probably not in his repertoire yet. We have only included the most common stats. You can find more stats in addition to more players cards at Baseball Prospectus…Tomorrow we will present the Rays pitching staff.

Player	PA	AVG	HR	RBI	SB	OBP	SLG
Carl Crawford	633	.309	16	66	43	.351	.476
Rocco Baldelli	520	.299	20	70	13	.342	.501
Akinori Iwamura	560	.275	17	64	7	.353	.449
Delmon Young	588	.297	18	75	22	.334	.473
B. J. Upton	604	.264	13	54	40	.347	.412
Jorge Cantu	548	.267	21	79	1	.308	.452
Elijah Dukes	532	.283	16	62	13	.349	.454
Jonny Gomes	530	.238	27	75	6	.343	.480
Brendan Harris	500	.259	11	56	3	.314	.390
Dioner Navarro	349	.269	8	39	3	.344	.410
Greg Norton	301	.260	11	42	1	.339	.442
Ty Wigginton	460	.269	19	65	4	.336	.467
Ben Zobrist	526	.279	5	46	9	.348	.393

A few thoughts on these numbers…

1. In 2006, the Rays had the worst team OBP in baseball (.314). It looks like PECOTA predicts the team to take a jump forward in 2007, with 10 players in the .330-.360 range. The premise being that players become more patient and adept at reaching base as they get older. That would go along way to improving this team dramatically in 2007.
   
2. Carl Crawford showed at times last year that he has a power game and we will probably see more home runs this season. Also, if CC bats third as predicted, look for more RBI.
3. Rocco Baldelli’s numbers are about what we are expecting, although we think he will probably be closer to 20-25 stolen bases.
4. B. J. Upton’s numbers look to be inflated a bit as he probably will not see that many plate appearances. Still, it is nice to see that despite his recent struggles, PECOTA still predicts him to emerge as a solid offensive player. He is young, and talented and there is still plenty of time to develop.
5. Jorge Cantu and Jonny Gomes…you would think they were a Hollywood couple as often as these two have been mentioned together this Spring. We might need a Brangelina-type nickname. We kinda like Jornny Gomtu. No? Anyway, Cantu’s and Gomes’ numbers are probably the most difficult to predict because half there careers to this point have been affected by injury. So we suggest taking their numbers lightly.
6. Delmon Young and Elijah Dukes (if he gets enough ABs) look to be exactly what most people would realistically expect.
7. If Ben Zobrist can give the Rays a consistent glove and .350 OBP, Joe Maddon will be very happy.