2007 PythagenMatt Summary (SABR Matt)
Before I go over the final "score" from the 2007 season, I'll remind you all what this here PythagenMatt thingy is. Quick review: PythagenPat is a modified version of the original Bill James discovery of the Pythagorean equation RS^2 / (RS^2 + RA^2) = W%. Patriot (as he calls himself) realized that the exponent was dependent on the run scoring environment for a given team over the course of the season and so wrote the 2s and Xs where X = ((RS+RA)/Games)^0.285. That is still the best seasonal pythag equation you will find in terms of lowest RMSE and highest correlation with real W%.
PythagenMatt, as I've dubbed it (following in the footsteps of Davenport's PythagenPort and Patriot's PythagenPat), is PythagenPat…but calculated one game at a time. What this does is forces the run scoring for each individual game to only count for at most one whole win. When the Orioles got beat 30-3 this year, it completely threw off the seasonal PythagenPat numbers for both them and the Rangers and this happens all the time with teams during blowout games. It also allows me to correctly reward teams for winning a lot of close games and get a better handle on who's hot and who's not over a given recent stretch.
Let's start with the big summary chart. It's similar to the chart I updated daily during the regular season until the Mariners' collapse took the wind out of my sabermetric sails.
| Team | G# | W | Matt | Pat | M-P | Closer | ME | PE |
| ANA | 162 | 94 | 92.4 | 90.0 | 2.4 | MATT | -1.6 | -4.0 |
| ARI | 162 | 90 | 84.4 | 78.9 | 5.5 | MATT | -5.6 | -11.1 |
| ATL | 162 | 84 | 84.5 | 88.7 | -4.1 | MATT | 0.5 | 4.7 |
| BAL | 162 | 69 | 75.7 | 70.3 | 5.4 | PAT | 6.7 | 1.3 |
| BOS | 162 | 96 | 99.2 | 101.8 | -2.6 | MATT | 3.2 | 5.8 |
| CHA | 162 | 72 | 69.0 | 66.5 | 2.6 | MATT | -3.0 | -5.5 |
| CHN | 162 | 85 | 88.4 | 87.5 | 1.0 | PAT | 3.4 | 2.5 |
| CIN | 162 | 72 | 74.4 | 74.3 | 0.0 | TIE | 2.4 | 2.3 |
| CLE | 162 | 96 | 93.0 | 91.8 | 1.2 | MATT | -3.0 | -4.2 |
| COL | 163 | 90 | 89.1 | 90.8 | -1.7 | TIE | -0.9 | 0.8 |
| DET | 162 | 88 | 87.5 | 89.4 | -1.9 | MATT | -0.5 | 1.4 |
| FLO | 162 | 71 | 69.5 | 71.6 | -2.1 | PAT | -1.5 | 0.6 |
| HOU | 162 | 73 | 72.4 | 72.0 | 0.4 | TIE | -0.6 | -1.0 |
| KCA | 162 | 69 | 70.1 | 73.6 | -3.5 | MATT | 1.1 | 4.6 |
| LAN | 162 | 82 | 78.7 | 81.8 | -3.1 | PAT | -3.3 | -0.2 |
| MIL | 162 | 83 | 84.1 | 83.5 | 0.6 | PAT | 1.1 | 0.5 |
| MIN | 162 | 79 | 78.2 | 80.3 | -2.0 | MATT | -0.8 | 1.3 |
| NYA | 162 | 94 | 98.2 | 98.3 | 0.0 | TIE | 4.2 | 4.3 |
| NYN | 162 | 88 | 90.4 | 86.4 | 4.1 | TIE | 2.4 | -1.6 |
| OAK | 162 | 76 | 75.7 | 79.3 | -3.5 | MATT | -0.3 | 3.3 |
| PHI | 162 | 89 | 90.4 | 87.6 | 2.8 | TIE | 1.4 | -1.4 |
| PIT | 162 | 68 | 67.6 | 69.0 | -1.4 | MATT | -0.4 | 1.0 |
| SDN | 163 | 89 | 92.3 | 89.0 | 3.3 | PAT | 3.3 | 0.0 |
| SEA | 162 | 88 | 84.9 | 79.2 | 5.7 | MATT | -3.1 | -8.8 |
| SFN | 162 | 71 | 74.2 | 77.1 | -2.8 | MATT | 3.2 | 6.1 |
| SLN | 162 | 78 | 74.1 | 70.7 | 3.3 | MATT | -3.9 | -7.3 |
| TBA | 162 | 66 | 63.9 | 66.2 | -2.3 | PAT | -2.1 | 0.2 |
| TEX | 162 | 75 | 72.5 | 78.3 | -5.9 | MATT | -2.5 | 3.3 |
| TOR | 162 | 83 | 86.9 | 86.6 | 0.3 | TIE | 3.9 | 3.6 |
| WAS | 162 | 73 | 68.1 | 69.6 | -1.5 | PAT | -4.9 | -3.4 |
| SYS | W | RMSE | ||||||
| MATT | 18.5 | 2.972 | ||||||
| PAT | 11.5 | 4.169 |
Notice, I was incorrect in my memory about how the Mariners actually finished in terms of PythagenMatt. I did continue to track a few teams of particular interest (all of the AL West, Arizona because of the Pythagorean implications of that team, the Cubs, and a couple of others that had my attention all year) so I had some end-of-seaso data even before I went and acquired the gamelog data for 2007 and did the full report. But I was remembering 86.8 PythagenMatt wins for the Mariners and it was actually 84.8. Sorry about that. :\
I referred to Root Mean Square Error (RMSE) when I talked about the strong-form of the statistical argument in favor of using PythagenMatt over seasonal Pythag earlier. Looking at data from 1900-2006, the PythagenPat error on a 162 game season is very close to the error represented above (about 4.2 Wins). Using the same data set, the RMSE on PythagenMatt is 3.4 (I was wrong about that too, sorry Doc…I thought it was 3.7…LOL), so we did a little better than average in 2007, but the point remains clear. Not only does PythagenMatt do about 4% better correlating with real W% (going from 0.9416 to 0.9805!)…it reduces error by 25%. That's LARGE, folks. The common wisdom before PythagenMatt was that we had maxed out our ability to estimate W% using RS data and that we would never do better than being off by 4 or 5 wins on a full season. This is the equivalent (in terms of statistical achievements) of the jump from OPS to Linear Weight Runs Created in terms of explaining team run scoring.
Now then…how about some pretty pictures?
If you want to see that in a larger size, click on the thumbnail. Here is a graph of the Seattle Mariners' season viewed through the eyes of PythagenMatt and PythagenPat. You can see that Matt had the drop on Pat from the very beginning…that the Mariners consistently outperformed Pat in good times and bad times…streaks and slumps. There was NEVER a point when this wasn't true. There is clearly something intrinsic about this team that makes it win games in a way that PythagenPat wouldn't see (and I've already discussed the two main culprits so I won't rehash here.
Here are the 2007 Diamondbacks viewed similarly:
Once again, in the final analysis, PythagenMatt did significantly better in predicting the performance of the team, but this time it took half the season for the difference to become apparent. Although to be fair, it also took half the season for the D'Backs to get really hot and start winning a bunch of close games since it took half the season for their bullpen to fully gel.
How about a couple of "underachievers" in PythagenPat wisdom? Let's start with the Oakland As, shall we?
The teams with large biases all look like this…PythagenMatt takes the correct side of the over/under with PythagenPat on projecting the rest of the season in almost every instance. This is why I was never worried about the As even when their Pythag was over .600…they were a bully team early on and then they collapsed completely.
If you folks want to see other teams, I can produce those graphics…just ask and you shall receive. Also…if you want to see other data displayed, feel free to chime in.












February 11th, 2008 at 11:06 am Quote
This is great stuff Matt.
One quick suggestion - without referencing all of the detail, you should remind readers the general difference between what the two items “PythagenMatt and PythagenPat” represent, each time you post a new page on it.. For the new visitor especially, who may not want to read all the sabrmetrics behind the calcs, a quick exec summary on what they are would add a boatload of context.
Still, though, this is great stuff. There’s a whale of a difference between an assumed 76 win team vs an 86 win team. This type of context makes a huge difference in terms of decision making. If the starting point for debate (i.e. DOV and other blogs) is completely different, of course the results would be different. Count me as even more of an optimist.
Thanks for doing the leg work.
February 11th, 2008 at 11:06 am Quote
As usual, this is top-notch Matt. Thanks a lot!
Just browsing through the “Closer” column, I noticed that majority of those closer to PAT are teams with a not so good bullpen. Also, majority are from the NL. This could just be because of small sample size though.
February 11th, 2008 at 11:26 am Quote
I admit I’m late to the party on the details of this stuff. Matt, have you run comparisons like this across multiple seasons? Is that in the THT article? Would like to see if the distributions are similar.
If you’ve mentioned this before, I apologize in advance as I don’t have as much time to invest in fishing out the details.
February 11th, 2008 at 12:20 pm Quote
Great stuff here.
So we were 6.9 wins behind the Angels before the offseason?
We added Bedard and Silva, they added Hunter and Garland. Are we really THAT far behind the Angels??? For me its a total tossup right now for the division.
We make another prospects for proven star move and we’re the team to beat.
February 11th, 2008 at 12:58 pm Quote
Yeah Matt. High fives man.
As it turns out, the winter’s debate about the 2007 M’s Pythagenport record has provided the deserved wind for your PythagenMatt sails and for your work generally.
……………….
We’ve never cared much what was on any other blog, but recently we’ve taken to posting on Geoff Baker’s blog once in a while. He had been writing fine articles and receiving Seattle P-I type of response; now it seems things are turning the corner there.
We have a mention of your findings up at Baker’s blog this morning, also. :daps: If so inclined, you might go over and make sure that it doesn’t get folded, spindled and mutilated by amigos who are a bit confused as to what you’re saying. :- )
………………..
Need to get that button in this year to link to PythagenMatt.
February 11th, 2008 at 1:06 pm Quote
Or, if Felix or Sexson perform as stars?
February 11th, 2008 at 2:59 pm Quote
Angels also lost Cabrera and got hug eoverperformances from Figgins, Willits and Anderson last year while the Ms got overperformances from Vidro and that’s IT…and more or less replaced Guillen with Wilkerson.
February 11th, 2008 at 3:19 pm Quote
Hey Padna…glad you stopped by…
Firstly…I’ve addressed your concern about newcomers with a little editing to this post. Added a two-paragraph summary of the process.
As for what’s in the THT article, it was just an obscure reference to it…it reached them because Tom Tango noticed PythagenMatt when I was talking about it over at Baseball-Fever.com and Studeman ran an analysis of PythagenPat and PythagenMatt used as projections of the second halves of season from the 81 game mark (IOW, he wanted to know how useful Matt and Pat were predicting games that have not yet been played). He found an 11% jump in predictive value between Matt and Pat…the R^2 values were something like .13 for PythagenMatt and .115 for PythagenPat…both very low as you’d expect, but PythagenMatt showing a distinct advantage.
Now…do you want to elaborate on what you mean for “across multiple seasons.” What sort of comparisons are you after…because I have the PythagenMatt records for every team in Major League history since I have the full gamelog set.
February 11th, 2008 at 3:23 pm Quote
Thanks for the support, Doc (:daps:)
I get tired of the mantra about Pythagorean records showing just how unworthy the Mariners were last year. These folks must not have been watching the games. We were quite obviously an above average team with two big black holes at the back of the rotation. With the black holes gone, there’s no reason to expect anything but continued improvement.
February 11th, 2008 at 3:40 pm Quote
Why isn’t this thread showing up in the Groks anymore?
February 11th, 2008 at 4:18 pm Quote
SABRMatt wrote:
Just curious how it relates in general. Sounds like you’ve done that legwork too. Again, if you’re referred that in previous pages, my comprehension isn’t what it used to be, so I may need to go back and reread your original posts.
Your chart shows ‘08. Would a the results be similar in ‘07 and ‘06 (or ‘97 and ‘96)?
The ramifications of this are not small. I think it absolutely debunks the pessimists’ side of the M’s situation by destroying its foundation. Aand once again may illustrate that sabrmetrics sometimes catches us back up to where those idiot GM’s already are.
You could completely blunder a perfectly good opportunity - and this past M’s season is the perfect example as to why punting this year shouldn’t get rewarded. It should get a GM fired.
February 11th, 2008 at 4:33 pm Quote
I haven’t done multi-year studies, but I remember posting an analysis of 2006 early last year…though I didn’t do anywhere near the amount of clarifying the data that I’ve done for ‘07.
I can pull any year you want out of my gamelog table and display it. It might be kinda cool to look at ‘95 (the surprise Ms) and ‘97 (the disappointing Ms) and see what PythagenMatt looked like.
February 11th, 2008 at 6:34 pm Quote
What is 2s Xs and X? No definition in you article.
February 11th, 2008 at 8:09 pm Quote
Tht was supposed to say “and so he replaced the 2s with Xs”…in other words..wherever you see a 2 in the original Pythagorean equation, there’s an X in the PythagenPat equation where X is defined as ((RS+RA)/Games)^0.285
February 11th, 2008 at 8:21 pm Quote
Matt,
This. Stuff. Rocks!
Lonnie
February 11th, 2008 at 9:43 pm Quote
Thanks Lonnie.
I do what I can.
I’m going to prepare a historical perspective…show lists of teams with high Matt-Pat differentials and try to make connections. That’s next up on the agenda.
February 11th, 2008 at 9:44 pm Quote
BTW…I’ll also look to see if those high M-P teams that overperformed pythag tended to retain their success (do better than the previous year’s pythag suggested).
February 12th, 2008 at 8:21 am Quote
Here’s a thought, Matt.
In regards to the “whys” for skewing wildly away from raw Pythag - I think you’re on the right track in regards to the back-end rotation guys were skewing outside the normal deviance threshold.
But, I’m wondering if it’s actually simpler to locate - in that both of these guys (Weaver/HoRam - and you could throw in Fearabend, too), were not only volatile - but they were both beyond the bounds of “replacement” player production.
My thought is that it is like in “most” cases, that anyone outside the bounds of replacement level production would only continue getting chances if they were, in fact, volatile — that they were above replacment level production often enough as to not lose their jobs.
I’m thinking it might be as “simple” as (especially on the pitching side), identifying ‘out-of-bounds’ players who still get significant innings/chances.
If there is a correlation, one might not need to actually see the direct volatility with game-by-game comparisons, but ‘intuit’ said trait indirectly.
February 12th, 2008 at 12:17 pm Quote
I’m coming at it from the opposite perspective. I’m going to look up large M-P differentials where the Matt was closer than the Pat and see which types of teams those are. I’ll bet we end up with a whole bunch of Mariner/D’Back type teams.
February 13th, 2008 at 10:53 pm Quote
They were half debating yours vs. normal pythag, and I mentioned that PyMatt was trying to minimize the distortion of blowout wins/losses, and in doing so was closer to the actualy wins by 07 M’s. I got destroyed because I don’t really know the predictive value of your calc. Even though I thought I saw you had, what, 91-94 somewhere. Not sure how you use it to do predictive calcs, but I assume you just play with the end result numbers.
I agree with you/Doc that when I watched games last year I didn’t see a 78 win team, and I prefer a metric that more closely matches what happened.
February 14th, 2008 at 12:16 am Quote
#18 — and that should be easiest to perceive if the out-of-bounds players were #4-5 starters, right?
Here you could perform a very clear study — teams that had extremely poor ERA’s in the 4-5 slots, and their records vs Pythag. Outrageously bad ERA’s might be associated with overperformance vs Pythag, and if so, you’d be pretty well on your way to discovering something very significant.
……………
Just noodling here, I remember the year that Fassero hit the wall… 1999? OK, he had an ERA of 7.38, and… yeah, his replacements (after 140 IP) were in the 8’s and 9’s.
That club was over Pythag by 2 … you’re talking only the 5 starter being bad, not the 4 and 5 together.
……………..
What was that year the Yankees brought in Al Leiter and a bunch of guys to try to stop the bleeding in the 5 slot …
OK, 2005. Kevin Brown 6.50 x 73ip … Jaret Wright 6.08 x 63 … Leiter 5.49 x 62 … Henn, May and Redding got trials in the 5 slot, with ERA’s of 11, 16, and 54, LOL.
Those Yanks were +5 over Pythag — it’s easy to imagine the #5 pitchers ruining the run differential.
………………..
Just conversation, of course. …if you had a historical database, it wouldn’t be hard to filter for (decent!) ballclubs that had the worst 4 and 5 SP’s, to see if those guys unduly ruined the RD.
February 14th, 2008 at 12:21 am Quote
Matt, could you add actual W% to those graphs?
February 14th, 2008 at 12:35 am Quote
Well I took the time to read Grahams’ analysis over at LL, and found it pretty well written. I didn’t see where a bunch of shots were taken at your program, Matt. Mostly, the call as I understand it, is for a predictive metric, as compared with a descriptive one. I think that (and my understanding is limited) you are on to something with your correction of the original formula.
So far, I am not satisfied with what I have seen from the predictive sort. Whether its pessismistic input of data, or something else that is being done to tweak the outcome, the M’s don’t score well. I just don’t buy this team being forcast to win less than 80 games. I think that a 90 win season should not be too tough with the roster in place.
My guess would be that a lot of over age 30 decline is insisted on, and a bunch of under 30 progression is not accounted for.
February 14th, 2008 at 4:55 am Quote
PythagenMatt isn’t designed to be predictive primarily. That said, in-season it does have more predictive value than either W% or PythagenPat. Though if I recall, regressed W% still does better (regressed W% = current W% regressed to the mean - .500 - to account for the probability of small sample error).
All of my efforts sabermetrically have consistently gone into understanding with as much clarity as possible how good each team is (and each player)…PythagenMatt does this. I can say with confidence that the 2007 Mariners were an 85 win team and that that is the base we’re building off of as we head into the off-season. Which means adding Bedard doesn’t merely take us from 78 wins to 83 or whatever…
From where I’m sitting, Bedard makes 90 wins a slam dunk for these Mariners.
February 14th, 2008 at 5:01 am Quote
.
Ya, well, a slam dunk in the sense it’s a slam dunk for the Angels… that on paper they *are* of that caliber heading into the season…
Nothing unusual though for a ballclub to greatly surprise or disappoint…
………………
You get a 112 ERA+ and a 102 OPS+ and you’ve got a 90+ win team, on paper. Hard to see how two 130 SP’s and three 100 SP’s, with a good pen, post less than a 112 ERA+.
But y’never know :- )
February 14th, 2008 at 5:13 am Quote
Well…of course…there’s always the “that’s why they play the games” caveat.
The Ms and Angels are of equal strength heading into 2008 if you axe me.
February 14th, 2008 at 6:46 am Quote
Isn’t it more predictive for the 2008 season to not worry about what either pythag said for 2007 and instead focus on the expected RS/RA once we know our roster and make a pythag off that?
February 14th, 2008 at 7:07 am Quote
Fett…not if there is reason to believe the team will defeat Pythag again. Which there is given the strength of our bullpen.
February 14th, 2008 at 7:15 am Quote
So you’re saying 2007 as a base with PythagMatt compensates for the bullpen effect better than assuming an extra 1-3 wins on a 2008 pythag based on expected RS/RA?
February 14th, 2008 at 9:24 am Quote
I’m saying that I believe philosophically that the best way to understand today’s problems is with yesterday’s data. In other words, you can “guess” 1-3 wins with the bullpen over pythag but it’s just a guess. I’d rather proceed from what we know…that the 2007 squad was an 85 win team and that we’ve probably added 7 or 8 wins to that.
February 14th, 2008 at 6:14 pm Quote
What do you have for expected RS/RA this year, Fett?
February 14th, 2008 at 7:23 pm Quote
I’m not going to pretend like I have any sort of prediction for expected RS/RA that isn’t just a guess or that has been calculated mathematically (like the one USSM just put up, for example). I’m just interested in looking at the different methodologies of others who are predicting our W/L record.
Much easier to comment from the peanut gallery then to throw your own projection system out there, to be sure… (not that such comments aren’t productive)
February 14th, 2008 at 9:14 pm Quote
Which is fine, Fett.
Since you asked about it, I took a 30-second glance at the USSM calculations. OK, here’s the thing with those calculations.
…………………..
They’ve got the 2008 M’s as scoring +735 runs, allowing -725, and being a game or two over .500. 110-115 ERA+ and a 90 OPS+ … that’s a great pitching team with a joke offense.
Everybody should at this point stop, draw a breath, and clarify. This argument stands or falls with the idea that the Mariners will have a terrible offense.
…………………….
The -725 is not unreasonable for RA: they allowed 813 last year, and just swapped out two very bad SP’s for two SP’s who are, as a pair, very good. (I see that they don’t have Felix improving by much, either, which is fine.)
But +735 runs scored? The Mariners scored 794 last year — with Lopez posting a 71 OPS+ and Sexson around 80.
+735 is exactly what the Kansas City Royals have averaged on offense the last two years (+731.5). The Mariners are paying their starting 9 hitters $60MM this year, in case we forgot. This is hardly the St. Louis Browns offense here.
………………
Fett — you being one of the ‘neutral observers’ as it were… fast question for you. Again, let’s retreat to the 2007 baseline idea — because 8 of the 9 hitters are the same guys. Do you see the 2008 Mariners scoring radically fewer runs than in 2007?
Please comment on the logic in the Ron Shandler article: Sexson and Lopez with huge room for improvement, Ibanez and Vidro with moderate room for decline, the rest of them very, very stable compared to other ML hitters on average.
……………….
If the Mariners score +735 and have a joke offense, then they’re going to have a hard time being much over .500, no argument!
But OTOH, if they allow 725 (per USSM) and score a mediocre + 794 as last year, they Pythag out to 88 wins per a simple Pythagorean Theorem (exp 1.81) — 88 wins, plus any leverage they get out of a JJ Putz bullpen.
A 794-run offense and 725-run defense obviously puts the 2008 Mariners in the vicinity of 90 wins — and in the pennant fight.
………………..
I guess I’d ask them if they concede that their “don’t expect much” argument …. stands or falls with the prediction that the 2008 offense will be far worse than it was in 2007?
February 14th, 2008 at 9:34 pm Quote
.
Note carefully that per USSM’s logic, the 2008 Mariners are in the range of 90 wins IFF their offense repeats 2007 exactly.
………………
That’s a MID projection! Last year’s offense, plus USSM’s pitching calculations.
Now, what if JuBet, or JLo, or Sexson, or Beltre, or Clement, Felix, surprise? Or two of them do?
That’s the UP projection: Sexson’s BABIP normalizes, and you get a breakout year or two from among the YOUNG players.
……………….
I think that 90% of us would agree that USSM’s prediction is the DOWN scenario: the 2007 offense loses 60 runs and the pitching staff does nothing special. That’s where they get their 82-83 wins: offense does as poorly as reasonably possible, and the pitching staff doesn’t have a Felix bustout or anything like that.
……………….
USSM is projecting the DOWN scenario. D-O-V is projecting the MID scenario (around 90 wins). But there is an UP scenario too: 95+ wins if we get a bustout from any two of [Felix, JLo, Beltre, JuBet, Clement, etc.]
February 14th, 2008 at 10:01 pm Quote
Even USSM’s projection is short. They call us an 82-83 win team based on simple Pythag…they forget the probability that our bullpen will get extra leverage again. I think the DOWN projection is 85 wins, MID is 91 and UP is 97.
February 14th, 2008 at 10:05 pm Quote
I would agree that 735 appears on the low end. Personally I’d like to see the numbers Dave plugged in to get the wOBAs he did… at least a BA/OBP/SLG prediction for each player.
No one seems to be in major disagreement that Vidro, Ibanez, Ichiro, and Wilkerson (compared to Guillen) are each more likely to somewhat underperform their respective OPS from last year than to maintain. I also think most people agree that Betancourt, Beltre, and YuBet will stay about the same. Not having those numbers, I do suspect the biggest difference between USSM and DOV is how much Sexson and Lopez can be expected to improve.
As a side note, assuming you’re using B-ref’s OPS+ numbers, remember that those are park adjusted, so an Tampa Bay and a Mariners team with the same OPS+ for example would likely have a significant difference between runs. Much better going with straight OPS like Shandler, and ideally sometihng like wOBA–again, I’d like to see the individual #s going into the wOBA calcs.
February 14th, 2008 at 10:09 pm Quote
Don’t know about 97… that would be way, way, way up. 97 wins has happened 4 times in the AL in the last 5 years, none last year, and 3 of those times by the Yankees.
February 15th, 2008 at 12:47 am Quote
.
Fair enough, Fett. That resolves the major question.
……………
A minor quibble or two:
Slow down a second there amigo…
122 - Ichiro OPS+, last year
119 - Ichiro OPS+, lifetime
109 - Vidro last year
110 - Vidro life
104 - Vidro 2004-07
Good fit for the Safeco park
It’s not clear at all to me that Ichiro and Vidro are obvious picks to drop off in any significant way.
……………….
Wilkerson — he had a terrible BABIP last year and if that normalizes, he at least equals Guillen in Safeco.
He could underperform Guillen (30% let’s say), roughly equal him (60%) or overperform him (10%) but let’s not simply peg that as a significant drop.
……………….
Ibanez, sure. He *is* the big candidate for a drop. Not that it’s a given he will…
………………..
It’s that SF water amigo. YuBet is actually our nickname for our SS.
;- )
But: why would Betancourt stay the same? I thought he was an age-26 roto sticky note player?
February 15th, 2008 at 12:49 am Quote
#37 - definitely. Predicting ANYBODY for 94+ wins is tough, much less a team with a 100′ish offense.
But Matt’s saying, if everything clicks. In that sense you could pick five or six AL teams for 97 wins if everything breaks right.
February 15th, 2008 at 4:08 am Quote
To be clear, I’m not predicting the Ms will win 97. I’m on record as expect 91-94 and being willing to bet the high end.
February 15th, 2008 at 4:44 am Quote
Again, good thing I’m from SEA…
February 15th, 2008 at 5:13 am Quote
:- )
Am I mixing you up with somebody else? Thought you were the guy who moved to SF to Admiral the visiting battleship or become a CIA spook or head the IRS compound there or something… You hain’t been the source of our Lincecum and Koufax and Bonds reports and all that?
I’m gettin’ old, cain’t remember anything except the Nimzo-Indian Defense anymore… weird how the chess and sabermetrics and age has you forgetting what kind of car you drive and who writes for your blog…
February 15th, 2008 at 5:34 am Quote
no…that was Fett…(who provided the Lincecum reports)…but I think he’s a Giants fan from afar now…hey Fett..correct me if I’m wrong.
February 15th, 2008 at 6:09 am Quote
Hey Doc (and everyone else)
I posted a discussion of Jose Vidro over at MarinerCentral in my blog section. Consider it my first full blown statistical discussion over there.
I basically argued that his BABIP spike last year was not a fluke and is indeed repeatable…am interested in comments on my thesis.
February 15th, 2008 at 10:48 am Quote
As Matt said, I’m generally a Giants fan from afar, though my father (he grew up worshipping Mays, thus Giants are my 1b team) lives in SF and I get there as often as I can (a few weeks a year). I probably get to 3-5 Giants games per year.
February 15th, 2008 at 5:02 pm Quote
#45 - OK now I got it :- )
Will try to get over there Matty. We’ll see if we can mess up what would otherwise be a clean kill for ya.
February 15th, 2008 at 6:16 pm Quote
Vidro’s numbers are heavily driven by BA. And his BA was heavily driven by an absurdly large number of infield singles given his speed.
February 15th, 2008 at 7:02 pm Quote
I have been ruminating about all these projections and particularly the ferocity with which people argue about the probabilities of different scenarios.
Being a physical scientist, as opposed to a social scientist, I only have a rudimentary understanding of statistical analysis. None the less, us physical scientists try and put error bars on our measurements, but the process of assigning a number with an error always involves judgment. The reason is simple and profound, measurement always results in error, with some errors being normally distributed about the ‘true’ answer and others being systematic. Gaussian statistics addresses the ‘normally’ distributed errors and generates error bars. However, you cannot eliminate the systematic errors, you rarely know their origin, and consequently cannot model them correctly. So a standard deviation assumes a normal distribution of events and says that the chance of 9 heads in ten flips is extremely unlikely and if you increase the number of samples from one set of ten to a thousand this will be born out.
But what about those persistent and troubling systematic errors? In real life 9 heads in ten might be a signal that the coin is rigged. This is where the judgment comes into play and it cannot be avoided. Based on ten flips you have to decide to ante up or figure the game is rigged.
So what does this have to do with baseball? The distribution of performance in baseball cannot be fit to standard probability distribution functions, whether they are Poissonian, Gaussian, or other. At a fundamental level simple probability theory does not model baseball, but this does not impede the confidence of many analysts. As a rule, be leery of model — whether it works or not — if you cannot justify the assumptions that underlie it.
The fact of the matter is that the difference between performances that win and loose in baseball is often smaller than the deviations in performance from the normal distribution. Let me try to clarify. Steve Carlton was a full time starting pitcher from 1967 to 1984. During that time he had 11 seasons with an ERA+ clustered around 110 (109 with a standard deviation of 9) and five season clustered around an ERA+ of 160 (ERA+ of 162 with a standard deviation of 12). It is truly a bimodal distribution. So I ask you, is Carlton’s ‘true ability’ really Gaussian and we just didn’t run enough simulated Steve Carlton seasons, or does this tell us something significant about Carlton himself? The question isn’t academic either, since ERA+ pitcher of 162 or of 109 that is starting 40 games for the home team could certainly be the different between the post season or not.
Let me state my hypothesis clearly; I believe that the difference between winning and loosing is often smaller in magnitude than the systematic errors that result from imperfect information AND overly simplistic statistical models.
In the end I find the sanctimonious disdain for people foolish enougth to state, “And that’s why we play the games…,” sophmoric. After all, Gaussian statistics miss the plateau jumps in skill (see JJ Putz), miss the fact that magic pixie dust can turn Bret Boone into Hank Aaron for a year or three, miss the injuries that result in performance variations that never approach a Gaussian distribution. These facts plus the small number of players on each team, mean that these effects do not necessarily ‘even out’ either. The reason we are wise to temper our confidence in our projections — a disposition embodied by, “And that is why the games are played…” — is that we neither know if Morrow is going to master a second nor how to model the probability of it occurring.
February 15th, 2008 at 8:40 pm Quote
Just to clarify, I see value in projections systems like PECOTA, zips, et al. What I resent is the notion that the distribution produced by PECOTA resembles reality — in other words if we could ground hog season Yuniesky Betancourt the resultant distribution of performance would likely match a Gaussian distribution… And if this isn’t true, what is the justification for the heavy-handed smarter-than-thou pedantry?
February 15th, 2008 at 8:49 pm Quote
Just to make my position clear, I see value in projections systems like PECOTA, zips, et al. What I reject is the notion that the distribution produced by PECOTA or some other projection system resembles reality. That if we could groundhog-season the 26 year old Yuniesky Betancourt a thousand times the resultant distribution of performance would look like a Gaussian distribution… And if this isn’t true, what is the justification for the heavy-handed smarter-than-thou pedantry? Why translate the assurance you have regarding coin flips or water molecules to the baseball where ignorance and systematic deviations rule?
February 16th, 2008 at 12:11 am Quote
USSM’s original position was that YuBet was going to be a .270 hitter…by now it’s obvious that isn’t the case…so they don’t get to claim they had him pegged right anyway.