OK, I know it's waaay too early for this plus he only used 100 simulated seasons but still, gotta report it: Replacement Level Yankees Weblog posted his simulation of 100 seasons using the current team roster for every team (at that time) and inputing in the CHONE prediction for their performance because he wanted to see how the Yankees do, but he posted the full MLB stats and guess what?, the Giants had the highest average win per the 100 seasons, with 88.6!, a full six games ahead of San Diego. In that simulation run, they were in the playoffs, either by winning the division or gaining the wildcard, in almost exactly two-thirds of the simulations.
Of course, that's possibly because of Barry Bonds playing 162 games or the assumption that Klesko hits like he did in the past (his CHONE forecast was not in the released database), but it was Diamond Mind that was used to play the simulation, so perhaps they can set it to target an AB/PA goal for each season. However, the Giants lead the entire NL in runs scored with 836 runs average, with the Cubs a distant 2nd with 804, and Atlanta 3rd with 803, so there must have been some hocus-pocus done to our lineup, it does not seem capable of that many runs.
Here is the CHONE rate stats for our lineup hitters:
Bonds - .290/.458/.575/1.033 - 21 HR in 284 AB, try again next year...
Durham - .283/.357/.468/.825 - 17 HR in 491 AB
Molina - .296/.339/.447/.786 - 11 HR in 342 AB
Roberts - .282/.359/.400/.759 - 5 HR in 452 AB, 30 SB
Winn - .284/.345/.428/.773 - 13 HR in 575 AB, 13 SB
Linden - .249/.330/.414/.744 - 16 HR in 458 AB, 8 SB, way too high AB unless someone DLed
Aurilia - .272/.329/.415/.745 - 13 HR in 466 AB
Feliz - .263/.304/.449/.753 - 20 HR in 527 AB
Vizquel - .274/.338/.353/.691 - 3 HR in 523 AB, 17 SB
Klesko - as noted, no forecasted performance
CHONE for pitchers:
Zito - 3.88 ERA, 166 SO/79 BB in 206 IP, 190 H
Cain - 3.90 ERA, 177 SO/87 BB in 184 IP, 158 H
Sanchez - 3.95 ERA, 83 SO/43 BB in 87 IP, 77 H; must have him slotted as reliever?
Chulk - 3.96 ERA, 67 SO/31 BB in 77 IP, 70 H
Wilson - 3.98 ERA, 48 SO/31 BB in 59 IP, 54 H
Taschner - 4.06 ERA, 78 SO/35 BB in 78 IP, 70 H
Benitez - 4.13 ERA, 32 SO/17 BB in 41 IP, 39 H
Sadler - 4.15 ERA, 64 SO/41 BB in 65 IP, 55 H
Lowry - 4.18 ERA, 133 SO/62 BB in 183 IP, 182 H
Correia - 4.24 ERA, 56 SO/30 BB in 74 IP, 72 H
Kline - 4.31 ERA, 35 SO/23 BB in 54 IP, 54 H
Morris - 4.36 ERA, 112 SO/51 BB in 194 IP, 207 H
Hennessey - 4.55 ERA, 77 SO/59 BB in 146 IP, 152 H
As I noted for the Zito forecasts, who knows what exactly went into the forecasts. Sounds like he tried his best to account for things though.
13 Pitchers and the worst ERA is Hennessey at 4.55 and the best is Zito at 3.88? Only a .67 spread from best to worst? You can take it to the bank that this ain't happening!
ReplyDeleteIn case you didn't notice, Linden's ABs seem to have come from Bonds and Roberts as Bonds only had 284 ABs and Roberts only had 452 ABs. They both will walk some (and in Barry's case alot) but, this still leaves enough Plate Appearances for Linden to get 458 ABs.
The worst OPS is Vizquel at .691, the best is Barry at 1.033, and only Durham at .825 joins Barry above .800. The average of everyone not Barry is just above .750. This sounds about right to me. I guess he came as close to nailing the offense as one could expect.
In conclusion, he just doesn't have anyone really good or really bad on the pitching staff and that just isn't realistic. The mean for the pitchers may turn out to be right but the standard deviation sure won't.
Good points, Mark, I was just rushing to get this out and didn't think about looking at the spreads. Doh!
ReplyDeleteI agree with your points about the mean but not the standard deviation.
Looking at the pitchers, I think the projections look about right for the most part, when they are looked at individually. Zito's is in line with the past two years. Cain reflects modest improvement over his first season (very conservative, but I can understand that from someone doing a projection on a player he doesn't really know to a great detail, as a Giant's fan might). Sanchez too, if I remember right. Chulk is in line with career. Wilson is probably an MLE projection based on his success in the minors and tempered with his struggles in the majors. Taschner too.
Benitez probably reflects his recent stats plus loss of health and getting into the danger zone for performance drops due to age. Sadler, I assume MLE. Lowry is a safe guess between how good he was in 2005 vs. how bad he was in 2006; again, conservative, I think with health and development, he should be closer to 2005 than 2006. Correia seems like a safe estimate there too.
Kline probably is an average of his last two seasons. Morris probably the same too, plus if you look at his expected ERA, I think in Shandler's Baseball Forecaster, his eERA has been around 4.40 the past three seasons. Hennessey, this is probably around his career numbers.
Thus, individually, I think the estimate for each pitcher is legitimate enough for someone to go with.
Of course, as you note, when reality rears its ugly head, some ERA will balloon upward and some will shrink downward. Most of the time, they will counterbalance each other so that if you take a collective look at the Giants pitching staff, that should be around what we can expect to get from the pitching staff as a whole. However, in bad years, more stink than usual, and in good years, you get some career years mixed in and the team over-achieves and maybe make some noise in the playoffs. That's baseball, that's life. And with that reality, the standard deviation will balloon upward from the short spread of the projections.
I know that Linden should get the ABs from Bonds and Roberts but 458 AB just seems to be more than one should get. Adding up PA's for each, I guess there are more AB's to pass around than I had thought (though I should add that he included 482 AB for Schierholtz, 489 for Finley, 483 for Lewis, 300 for Ortmeier, and 363 for Horwitz. :^).
Thanks for pointing that out!
I emailed the guy and one major error is he had Durham getting 637 AB on average and totaling 705+ PA on average. In Durham's dreams!
ReplyDeleteAlso, Durham batted 2nd, Bonds 3rd, Winn (!!!) 4th, Vizquel 8th.
All the stats look good though, so that was accurate enough. And Bonds ended up with 32 HR on average in the sim, enough for the record and then some.
Hmmm, he also had Finley playing some, as well as Matheny and Worrell, and, what the heck, Benitez. :^) Plus a number of prospects I've never heard of (am I that out of the loop?).
Just take it with a gigantic grain of salt, but rejoice in the results.
Haha your very welcome for searching for this one buddy. Unless you found it on your own, in which case, thank you. Either way, I still like the look of the predictions. It gives us something exciting to look forward to. I'd bookmark this and see how true it turns out to be as a marker for Diamond Mind's and CHONE's accuracy.
ReplyDelete-Nick
Actually, I found it via Baseball Musing, I should have credited them, my bad.
ReplyDeleteDiamond Mind has been running simulations of the upcoming season just before the season starts, in order to get the latest planned roster, and comparing its results at the end of the season with other predictions and they have usually been among the best in terms of being closest to the mark among the predictions. Here is the latest comparison, for the 2006 season: http://www.diamond-mind.com/articles/tmpred06.htm. Out of the past 9 years, only the Las Vegas Over-under line has beat them, and just barely.
Obviously this Yankee simulation used CHONE projections, whereas Diamond Mind has its own projections it uses to do its predicted standing. "The Diamond Mind simulations entry is the one representing the average result of simulating the season 100 times. These simulations were done a couple of weeks before the season started."
Yes, it would be interesting to see how CHONE does against, say, Diamond Mind's standings.