Well, we all know that, don’t we? The Oakland Athletics gave us all one of the most memorable, thrilling and unexpected seasons ever. I loved this team, perhaps more than any I have followed because they made everybody look stupid, including you and me, for predicting them to fail miserably, while they refused to be shut down. The thrill was second to none, the Bernies leaned, the pies were tossed and the masses rejoiced.
Mighty Angels fell. Mighty Rangers fell.
Now, everybody feels different about the A’s. Now, the predominant feeling is that A’s should be poised to repeat, as both of their main rivals got weaker and Oakland is, after all, the defending AL West champion.
But, any prediction on how strong the teams will be in 2013 with all the personnel changes has to start with how strong the teams were in 2012. And there are better ways to do that than just looking at the W/L record. Unfortunately.
Okay, let’s start with the real 2012 standings, just because they look so pretty:
First quick plausibility check is to look at first order Pythagorean wins. If you don’t know what that is, it is simply an algorithm that takes in account the number of runs a team scored and the number of runs a team allowed, and basically says: “With so many runs scored and allowed, the most likely scenario would be a team finishing with a so-and-so record. Or, in other words, it tries to take out the randomness of converting runs to wins. Unless we have a very good reason to believe it is a skill, and we normally do not, we should not think that teams who won, say, more than their fair share of one-run games will repeat that tendency in a year to come.
The A’s were 25-18 in such games, for a .580 winning percentage, which is good but also in line with their winning percentage in all other games last year – .581. And Old Greek concurs:
The A’s “lose” a win, but so do the Rangers and the Angels. What this says is that the A’s converted their runs to wins at an expected rate. A team that scores 713 runs in a season and allows 614 should expect to win 93 games, give or take a few. This process is also called first order wins.
As you could guess, there is also a process called second order wins.
Just as we do not associate turning runs into wins at a higher than average level with a repeatable skill, we don’t associate turning hits into runs with a repeatable skill, either.
Remember when Adam Dunn opened the 2006 season with hitting his first six home runs with nobody on base? When he finally took Mark Mulder deep with a man on second for his seventh HR of the year, he joked how he wanted to be the first guy to go 40 HR / 40 RBI in a season. Sometimes home runs happen with men on base. Sometimes without. The sequence of hits is something that is mostly out of control. The players will not hit better because there is a scoring opportunity nor will teams in most cases significantly optimize their hit flow to score more runs. A team can hit a double five innings in a row and not score. However, if those five doubles happen in a same inning, they will lead to multiple runs. And if they do happen to occur in the same inning it is for the biggest part not because the players tried harder, concentrated better or because the manager was a genius. It is chance. Luck. Distribution. Whatever.
So, to eliminate that factor, we look not into the runs the teams actually did score and allow, but into the hitting components of wOBA. The A’s had a below league average wOBA, yet they scored more than league average number of runs. They allowed their opponents a bellow average wOBA and kept them well underneath the league average in scoring. Overall, they both converted their own “hits” into runs at a pace much better than expected and suppressed the opponents conversion. The Angels and the Rangers did the opposite.
You may argue two things when you see this. One, that wOBA doesn’t cover enough to be used as a predictor and two, that beating your wOBA predictor is a repeatable skill, some combination of Melvin genius, A’s being better human beings than anyone else and unicorn power.
First, wOBA really does cover a lot of ground and was proven in test after test to be the best available descriptor/predictor of actual performance out there. As for run prevention, it is a perfect combination of pitching and defense. Who cares if a pitcher induces weak contact or the fielders have great range? The combination of that will be the number of singles, doubles and triples given up and that is all we want to know. We don’t need to pin specific contributions or divide them precisely among players involved, we are interested at the team as a whole.
Second, you might say that wOBA does not cover specific managerial strategies. For example, out is an out for wOBA, strikeout is just the same as a sacrifice bunt, and they should not be, you argue. If teams consistently use sacrifice bunt, they should convert more of their hits into runs then the ones who don’t, because they have productive outs.
Mariners sacrificed 32 times in 2012. They scored 34 more runs than what their hit distribution would indicate.
Rockies sacrificed 74 times in 2012. They scored 17 fewer runs than what their hit distribution would indicate.
Overall, the correlation (r2) between teams* outscoring their hit distribution and frequency of bunting was 0.00. Yes, that is zero point zero zero. Correlation with sacrifice flies (which, one can argue, are equal part decision and chance) was 0.02. With strikeout frequency of batters, it was 0.09. The only somewhat notable correlation was with grounding into double plays (0.31), and one can argue that there is not much skill or managerial control in hitting into those or staying out of them.
* I only looked at 2012 data, because it was a lot of work
Made short, wOBA covers pretty much all that happened on the field that is associated with skill, with two notable and measurable exceptions. One is the base running element (since 2012, wOBA on FG does not include base running) and the other is my own pitch blocking. If we add those, it looks like this:
Having looked at all the elements of run creation and prevention that we have, the most likely distribution would land the Athletics at 89 wins. Still awesome. But also 5 wins behind both the Rangers and the Angels. Were the A’s lucky? Were they clutch? Were they smart? Probably a bit of all, but there is nothing to indicate that they can repeatedly outperform their components based on any combination of intangibles.
So, if 2013 featured exactly the same players on every AL West roster and they all made same individual contributions as they did in 2012, the most likely scenario is the A’s missing the Division by 5-6 games. Just in case your Giants friends read this and start molesting you about how your team got lucky, here is the complete MLB overview:
Delta wins is number of additional wins a team had as compared to the expected number of wins due to their run creation and run prevention components. One standard deviation is 4.7 wins.
What does this all mean?
It means that while the A’s are deserving AL West champions, we should not treat them as such when evaluating their chances for 2013 and justifying roster moves. It means that while the A’s won 94 games in 2012 they were just as likely to win 85 and have both Angels and Rangers win 100. Before we start to calculate how much stronger or weaker any team in AL West has become, due to personnel changes, aging curves and regressions, we have to recognize that the base the A’s are building on is about five wins weaker than the base the Rangers and Angels have (The sum of individual fWAR values is even more decisive, with the Rangers 9 wins ahead of the A’s and the Angels 6 wins ahead. It is also more imprecise as it is muddied with UZR, because defensive contributions had to be split up to individual players).
Or, in other words, imagine that the Mariners, the Angels, the Rangers and the Athletics replay 2012 season. Not only are all the players same, but they all have exactly the same contributions as they really did. The only difference is that the timing is random – the hits or the strikeouts will not necessarily occur at the same situations as they did in the real 2012. Now let’s say they do that 100 times. That’s hundred seasons, each one with Josh Reddick hitting 32 home runs, each one with Cespedes getting 142 hits in 540 plate appearances, each one with Michael Young having .312 OBP and Erwin Santana giving up 39 home runs. Do you know in how many of those 100 repetitions the A’s end up in the sole possession of first place? Eight times out of a hundred.
This is not to take anything away from those 2012 A’s. Hell, to me they were not of a eight-in-a-hundred variety, they were of a one-in-a-hundred variety. I loved them. But they were still probably only the third best team in AL West.
The thing to me, which is also usually how I respond to those dumbass luck-invoking Giants fans, is that the A’s team at the end of the year was nothing like the A’s team at the beginning of the year. To me, saying that the true talent level of the 2012 A’s was 89 wins is quite impressive, given the fact that between the first and the second halves of the season, Brandon Moss, Chris Carter, Josh Donaldson all went through an apotheosis, Tyson Ross turned into Brett Anderson, Sean “The Kraken” Doolittle was awoken, etc. All of these are cases where I’d be very surprised if they reverted to anything resembling their previous forms. My point is simply that I think the A’s team going into 2013 is already much better (not including the small changes so far, though Young may turn out to be more than small) than they were as an aggregate 2012 team.
Also I truly think that Josh Donaldson is going to be the best player on the A’s for the next 4 years.
Sure, that’s a valid point of view. If you think the 2013 A’s are better than 2012 A’s you might have a case, or might not. But if you, say, think that having Doolittle and Carter for a whole year makes the A’s 1 win better, then my argument is – you should expect them to win 90 games, not 95.
That’s a reasonable argument.
They would be the experts on luck, I suppose.
The Orioles might have a few thoughts on the subject too though.
That’s quite a statement. Do you just mean he is the best player likely to be in Oakland for the entirety of the next 4 years (eliminating Cespedes), or that he will be the best on average over the next 4?
Josh Donaldson, by my calculations, had a 132wRC+ after coming back from the minor leagues, along with his +14.5 UZR/150 at 3B for the year. Last year, Adrian Beltre had a 140 wRC+ to go along with a +15.1 UZR/150 (and was worth 6.5 WAR). I think that the mechanical epiphanies are real, that Josh Donaldson is every bit the hitter he showed, as well as the defender. Yes, I’m crazy, but I think Josh Donaldson is better than DFA thinks he is. I essentially think that he’s Adrian Beltre.
With double the balls
Does Beltre only have 1, or does Donaldson have 4?
I think that argument is borderline crazy. You are talking about what, 35 games?
Wasn’t it more like half a season?
Nope, 47.
And really, Donaldson had an amazing run in August. That’s basically it. He accumulated almost all of his value in those 17 August games. Long term he strikes out too much and walks to little for what he provides offensively.
His K/BB percentages after coming back from AAA were at about 17/7, which is right around league average.
I’m also including his performance in AAA this year, which demonstrated the exact same player and skills as he showed after returning.
Is what I’m saying the objectively correct assumption? No. It is my prediction though, based on some statistics and some intuition. I don’t expect to be believed.
Ok, I can respect that. If you are going to stick your neck out on your claim about Donaldson, good on you.
That said, I disagree and if you care to make it a wager, I would be happy to take the counterpoint to your statement.
hee, I would make a retzips joke, but it wouldn’t register with aardvark
no, no it wouldn’t.
I’ll happily take the over on 4 WAR from Donaldson in 2013, unless he misses significant time to injury. Care to take the under?
This reminds me of the 4 years in a row when I put down $20 at the beginning of the season that Rich Harden would win the Cy Young.
I’ll take the under and hope I lose.
Any DL stint longer than 20 days voids the bet, wager is $20?
Excellent, we’re on.
Well that bet is already taken. Care to make another one along the lines or your original premise? Josh Donaldson vs the rest of the team over 3 years?
Sorry for betblocking you. You can take mine if you wish, no problem
oh, I didn’t mean to suggest that you did anything of the sort. I could have offered to make the same bet, but I’d prefer to come up with a different one.
ha, that one’s tougher. how about, Josh Donaldson will be in the top 3 in cumulative fWAR over the next three years on the A’s? voided if he gets traded or misses over 60 games total due to injury?
60 games total?
He came back up in mid-August.
Very nice – thanks.
Obviously the 2nd half of the A’s 2012 was much better than the 1st, but did the amount of luck change between the two halves too?
Good question.
First and second half pythagoreans?
Though one of EC’s points is that the A’s met their 1st order (Pythagorean) wins, and the discrepancy only shows up in their 2nd order wins.
A well written piece but…
I disagree.
Projection systems have play in their “final” results. (Granted, there weren’t many that forecast 94 wins for Oakland in 2012 but that’s why you play the games!) The A’s were able to max-out their projected performance last year and it was a joy to watch, even if the odds of it happening again are low. Because the real question isn’t what would happen if we replayed 2012 100 times, it’s what will happen when everyone starts out 0-0 in 2013. That 5 Win base advantage you say the evil-doers had over the 2012 A’s does not carry over into next year.
I don’t think the 2013 A’s will be the dominant team in the Al West… not even if Santa gives them a legit SS for Christmas! But I think we saw enough real improvement last year to predict they’ll be in the mix for the division crown next year. How much they’re in the mix, the odds of their repeating as the division champs, rests entirely on the roster changes, aging curves and regression/projection for the 2013 rosters. Weigh those numbers and place your bets and watch them play the games to see what happens.
I am not really sure what you are arguing, but I am, of course certain that you are wrong for disagreeing with me.
Seriously, though, this is wrong:
To best estimate how good the A’s (or any team for that matter) will be in 2013 you need to know two things. How good they were in 2012 (that’s the part I did here) and how much they changed in 2013 (this is what we have to estimate next, with the roster changes, playing time changes, experience, age curves and so on). You need both parts of the equation. And if you are using 94 wins as the first part, you are doing it wrong.
I respect your certainty in my wrongness.
But excuse me while I go nuclear with an example that, in my mind, goes neener-neener pthhhhbbt to your certainty: If the Angels completely changed over their 25 man roster from last year’s band of LOSERS to an entirely new 25 man squad of soon-to-be LOSERS you would not have any realistic means of, nor basis to, evaluating how good the team was in 2012 while predicting their 2013 performance.
1. You’ve already noted the inherent random nature of timing and how it affects outcome. Random cannot be predicted but your argument seems to suppose a change in the random to the detriment of the 2013 A’s by 5 Wins before accounting for the change in variables. You’ve made a very clear case that based on 2012 talent the odds are against Oakland winning the division in a replay scenario. I’m not disagreeing with that.
2. We can project player performance heading into 2013. We don’t know when guys are going to get base hits or strike out or any of that but folks try and predict how many times a given player will do those things during the course of a season. Now, as I understand it, those individual data sets go into creating an estimate of how a team will do in 2013. If after doing all the math said team’s projected 2013 wOBA is .330 then you base the expected number of runs that team will score on .330. It doesn’t matter if in 2012 our hypothetical team scored fewer runs than they should have based on their actual wOBA. They don’t get penalized or rewarded for that discrepancy in 2013.
Simply put, the projection systems will account for whatever wOBA discrepancies existed in last year’s numbers when they project next year’s numbers. They don’t care about luck or random… or unicorns. When I say the Angels look weaker than they did a year ago it’s because Torii Hunter took his 5 Win performance to Detroit and I’m betting against Mike Trout posting another 10 Win campaign. And that is what the projection systems are going to account for when they spin out their numbers.
Here are just three quick thoughts.
1. In your nuclear example – yes I would. I would have the same chance as everybody else. I would say that X (2012 true talent level) of them suckers is 94 wins. Then I would add Y (2013 changes to the team) which consists of three components: Y1 (players lost), Y2 (expected performance changes of remaining players) and Y3 (players gained).
X = 94 wins
Y1 = -94 wins
Y2 = 0
Y3 = ?
So, I’d be saying the same as everyone else.
In real life, though, teams don’t change 25 roster, so you are better estimating taking a known component which accounts for the majority of performance and then adding and subtracting.
2. While calculating team wOBA might help you estimate team run creation, you will not have any such luck estimating run prevention. Another reason to go with X + Y approach
3. When you say that the only thing that matters are roster changes, it’s akin to saying that the only thing you need to know how tall your son will be on X-mas 2013 is how much he will grow from now till then. And I say that it’s not correct. You need to estimate both how tall he is now and how much that will change. Otherwise, you can only estimate that he will grow two inches, but have no clue if he will end up 4’10” or 5’10”.
And finally, me showing the low odds of the A’s winning 2012 was for illustration purposes. I don’t care about 2012. It’s in the books. But 2013 team is the same as the 2012 team plus the changes that happened on the roster and expected regressions or improvements. So, I’m taking that as a starting point, or X if you will. Next, I (we) have to make educated guesses at Y. I am confident that this approach gives you better idea of how good the A’s are in 2013 then adding individual wOBAs or WARs.
I’m guessing it’s a typo in the above equation when you have Y = 94. I think you meant X = 94 because you wrote out that Y = Y1 + Y2 + Y3. So following that assumption, let’s plug in the numbers.
94 – 94 = 0
0 +/- 0 = 0
Meaning your answer depends entirely on Y3. As it would for everyone else. But since there is no way the new players’ performances affected the run scoring or run prevention output of the 2012 team you would have no choice but to base the 2013 team performance projection on individual projections.
You seem to be arguing that there’s carryover from the previous season. There isn’t. But the most accurate method for making a projection is to base it on historical data, i.e. a player’s previous performance. What you’re calling the true talent level of the team is already being factored into the individual performance projections.
You’re right, of course, when you say that teams don’t do complete roster make-overs. (Although it seems like the A’s came awfully close at one point last year, save for Pennington and Crisp!) The point to my example is to show how it is the individual pieces that are key to making up the whole. And if a player’s run production was below his actual wOBA in 2012 then he does not get any kind of bonus or penalty when projecting his 2013 wOBA.
Your third point isn’t comparing like things. If Pujols hits a HR in his first at bat in a game there is no tangible carryover to his second at bat. He doesn’t get a 4th strike or walk after Ball 3. It doesn’t matter if it’s his first HR of the year or his 30th. Nor will the ball carry an extra 50′ because of the previous blast. If my son stands 4’10” after one measurement then there is a an actual, physical carryover to the next measurement.
It was a typo. Thanks and fixed.
Yes, we both agree that past performance of the team is irrelevant if all 25 players are exchanged. But back to the real world.
If I seem to be arguing that there is a carryover from previous season, it’s because there is. The carryover is the amount of talent the team carries over. The most accurate method for making a projection is, indeed and as you say, to base it on historical data. That is exactly what I am doing. I am basing it on historical data, just not the one you seem to agree with.
See, if Pujols hits a HR in his first at bat, it has a carryover to his second at bat. Not because the result was a home run, but because it told us something about his true level of talent. I don’t care that he hit a home run. I care that he was capable of hitting a home run, and will adjust my projections for him accordingly.
I will, however, value that information only infinitesimally, because it is based on only one plate appearance. The information about the A’s true level of talent is based on more than 12,000 PA and is therefor much more significant. If Pujols hit a HR in his first 12,000 at bats, would that be significant to what we expect of him in his 12,001st? You bet it would.
The 12,000 plate appearances I analyzed tell us that our best bet is that the A’s were a 89 win true talent team in 2012. Now imagine there were no roster changes at all, that nobody is really on the steep part of the aging curve or due to a significant regression. We would have every reason to believe that 2013 A’s are also a true talent level 89 wins team.
There were some roster changes, and some aging curves and probably some regressions are due. So what I say is, take the 89 wins as a starting point and then add and subtract. I agree that this is not always the best approach (it would be a bad one for 2012 A’s and what seemed to be 40 rookies on the team, although there is hardly a good approach there at all). But in most cases it is the best approach, much better that looking at individual projections.
This is like when I go mountain climbing. I have a backpack I use all the time and a certain set of gear, that I manage to squeeze in the best I can. Say I have 800 cubic inches free when I pack everything in. Now, I get new crampons and their pack mass is 200 cubic inches more than the old crampons. My approach is to say that my best estimate is that I will have 600 cubic inches free when I pack my gear this time. Yours is to measure every single piece of equipment and add those values up. I am convinced my method is the better one.
This is a key point. Historical data is about what actually happened. By using 2nd order runs you are basing your 2013 projections on what could (or perhaps more accurately, should) have happened for the A’s in 2012. For all teams, obviously, but we’re using the A’s as the example. But team numbers are created by the individual performances and those performances are already being screened by an analytical process that includes all relevant factors to produce a real talent projection for 2013. Since we have no way of knowing which players will over/under perform next season we have to base our team projection on the numbers produced from the individual projections.
OK. I won’t argue otherwise.
Fair enough. And admittedly, while rare a more likely scenario than my Total Roster Replacement example. Using your example the Y1, Y2 and Y3 would equal 0. I acknowledge that if you can keep all those variables constant then it makes sense to use the overall data from 2012.
However, real world example, the 2012 Angels included Torii Hunter, Ervin Santana, Dan Haren, Zack Greinke and Jordan Walden on the 25 man roster. Barring something strange, those players will not contribute to the 2013 team. I’d argue that a 20% roster turnover from Game 162, 2012 to Game 1, 2013 puts a serious hurt to the relevancy of the Angels’ true team talent level from last year.
I cut you off there because my approach would be to put in the new crampons and then prioritize the remaining gear and see how much of it I could cram into the bag. Also, I don’t climb so the metaphor is largely lost on me. The one part I question is, the 800 cubic inches in your backpack is a hard cap and I don’t see win projections as anywhere near as solid. You’ve already noted the fluctuations between expected and actual run creation. I’d add to that the wide range of win totals possible during a season. The 2001 Mariners won 116 games. The 2012 Astros won 55. You can fill your backpack to the very last inch but I imagine it borders on the impossible for any single team to win 162 games in a season.
My stake in this discussion is simple. I don’t know how strong the A’s, Angels and Rangers are projected to be in 2013. I leave that to Bill James and PECOTA those of that ilk. I do know that Torii Hunter and Josh Hamilton were significant contributors to the success of their respective teams in 2012 and their omission from the 2013 rosters of those teams (1 down, 1 to go) will have a negative impact on the projections of the true talent levels for the Angels and Rangers next season.
Fair point. Again, I am not sure where the cut-over is to say that a certain percentage of roster retention is a minimum for basing the projection off of a last year and adding changes is the best way to go. Perhaps 20% is already too much. I think it’s not, though, especially because of my point of run prevention.
The data that I use is historical data that actually happened. I am counting singles, doubles and home runs, instead of counting runs and wins. This is actually a fairly common approach when projecting future performances. You don’t best predict future ERA based on past ERA, you best predict it by accounting for components that lead to it – strikeouts, walks, home runs, ground balls and similar.
I should clarify, because I didn’t mean to imply you weren’t using some historical data. The hits, walks, HBP and SB/CS you used is historical data.
But when you redistribute those hard numbers to normalize scoring and then convert those normalized runs into an expected win total, that is when I feel you have left the bounds of historical data. I understand what 2nd order wins are trying to do. But since we’re already using arbitrary age curves and playing time estimates and weighed seasonal performance data to produce future player projections… why adda team based wOBA bonus/penalty to the mix.
You’re right, there is a talent carryover from one year to the next. But the performance that talent will create can and will flucuate. Runs scored (both actual and normalized) is a product of performance, not talent.
You also make “projecting 2013 team wins from what their projected wOBA is” sound really simple, while it is not, at least on the run prevention side.
I challenge you – Bill James projections are out. Tell me what you think the run differential of 2013 A’s will be and therefor their likeliest win-loss differential. So, basically let’s say all my gear is measured by Bill James, you only need to add it up and see how much place is left in my backpack.
I will stick with my approach. I will use the 2012 information I collected and apply only changes based on the same Bill James projections. Let’s see how thorny each way is by keeping everything transparent and publishing our ways in separate posts. I think it would be a great way to evaluate the two approaches (not the results we come up with, but the way itself) and I think we both and everybody following could learn something.
Deal?
Also, is there interest from those reading this in such an attempt? If so, say “Ay”.
Ay
This. I mean, “Ay.”
I would absolutely read and be fascinated by both accounts. Then again, I’d read almost all baseball analysis by you or grover.
Ay Ay Cap’n.
Holy Zeus, no! I think that’s what has to be done… I never said it was simple to do. I don’t know the math well enough to accept your challenge. I’m not even sure where to find the wOBA formula for 2012; I plugged the Angels’ 2012 numbers into the 2011 formula listed on FG and came up with an wOBA +.004 higher than FG’s listed 2012 wOBA.
Here.
My point was not that much that is hard to do from the math standpoint, but that the data that bests describes run prevention (wOBA against) is per definition only measurable on a team level. When you split it into individual statistics, you have to go with one of the FIP/xFIP/SIERA and some sort of UZR-like statistics, which are in combination not that precise.
Thank you for the link.
Would you mind telling me where I’d find wOBAA and how to average it? Doesn’t seem to be a Fangraphs staple.
I had to string it together from a combination of FanGraphs and ESPN outputs. Perhaps there is a simpler way, but I was first looking for a RoE and didn’t find it, so I made a shortcut and just counted all Es as RoE. I don’t think there is a systematic error with that, though.
I’ve not a clue what you just said.
I’m not trying to be difficult, it’s just that I stopped keeping up with the advanced metrics right about the time people began to argue about FIP vs. xFIP vs. tERA.
No problem.
RoE is reached on error. This is probably why your home made wOBA will not be the same as FanGraphs one, as you don’t have the information on how many times a batter reached on error. My team wOBA in run scoring doesn’t have it either, but I don’t consider it to be important. Reaching on error is only marginally a skill a batter has and when you look on offensive side of things, the plate appearances are spread against multiple defenses and their qualities.
However, when you look on wOBA allowed, RoE is much more important, because it is the part of what you want to qualify, namely how often your defense allows for base runners. As I couldn’t find a site that keeps track of RoE (there is something out there, because I used it in the past, but I just don’t remember where), I just used the total errors instead (meaning both “E” and “e”, if you are a scorer). I have no reason to believe that there is significantly different part of all errors that are reaching base on errors between teams, so I just stuck with that.
So, if you want to calculate wOBA against you need to combine teams fielding stats to get errors and team pitching stats to get TBF (Total Batters Faced, same as PA), HBP, niBB, 1B, 2B, 3B and HR.
OK, forget everything. I’m an idiot. I just realized that it’s BR that has RoE, so you can use it from a single source, like here.
Time to go to bed.
Ay ay cap’n
If you’re using fangraphs’s wOBA, I don’t think it’s park adjusted. Given our home turf, I think park/league adjusted wRC+ would be a better measure of the offense’s abilities and would probably make ’em look a little nicer. Not sure if it makes that much of a difference in the final conclusion but thought I’d point it out.
This is from memory, but I think fangraphs wOBA is NOT park adjusted but wRAA is.
It doesn’t matter. The plays exactly the same number of their games in every ballpark as their opponents did. These are relative numbers – we look at the relation between wOBA they created and wOBA they allowed. If you don’t think that a ballpark has huge influence only on players of A’s skill set and none on their opponents, they could have been playing on the Moon for all I care.
This sort of confuses me, but I think it is just me. You mean they play the same # of games in other parks, right? Obviously they dont play the same # in all parks, TX gets an Offense boost from 81 and we get a pitching boost from the OACC, right?
Even if we are just talking about other opponents I dont think we play exactly the same #. Heck, one argument for it being closer than 5 would be that we had to play SF while the Rangers got to play the Astros, which will not happen again. But at that point I would assume it is really picking nits and would not cause major variation in your numbers typically.
When I say opponents, I don’t mean division opponents, I mean opponents in each game. In the end, I am looking at wOBA differential between what A’s did offensively in their Game 1 of the season as opposed to whet they allowed Seattle to hit in the first game of the season. They both played in Tokyo Dome, so whatever park effects there are, affected both teams. And so on. I go from Game 1 to Game 162 and add those. In every one of those games there were same park factors affecting the A’s run creation as there were creating their run prevention. The wOBA differential is then converted to run differential, and run differential to wins.
Clearer like that?
Strength of schedule (we got to play the Giants) is the third order wins. I am not going to get into that, because to do it properly, I would have to analyze every pitching match-up and I don’t have the time. I will account for AL West realignment when I come to the part where I try to measure differences from 2012 to 2013.
I knew it was just me, but yes now it makes sense. Thank you.
I just wanna find a way to get Trevor Bauer from Cleveland please. I wish Billy would have gotten in on that.
Thanks, and go As.
Chloroform, and a hidden compartment in a car trunk?
They gave him up for an SS that can’t hit. We could have given them Weeks!
Thanks, and go As.
Or, Pennington for Bauer, instead of Young?
yesplz?
Thanks, and go As.
That I could have been excited about…
Why not both? Just get the Royals involved.
It does seem like the dbacks gave him up pretty easily.
Very interesting. Now can you make it so that I feel optimistic?
Are you planning on doing a prediction for the division in 2013 accounting for age curves etc? If so, I look forward to it.
I think the point about our evolving team is important. Hell, not giving innings to Ross has to be worth ~15 wins!
What’s the effect if we just use the A’s second-half wOBA and project from there?
(and no Godrey!!!)
Speaking of that, has it been confirmed that Parker accrued a full year of service time last year due to Billy’s inability to tolerate Godfrey for even one more game? I remember the math being very close.
What would have been different if he had one start fewer?
One week less service time.
(I may be misunderstanding the rules here)
But one week less would not have made any difference
That’s what I was wondering. At the time it seemed like it might have.
Unless I am horribly mistaken, it doesn’t whatsoever.
You get three years before the player goes to arbitration, but there is still Super-two, meaning that some of the players who have more than two, but less than three years of service go to the arbitration. The cutoff this year was 139 days, meaning that everybody with more than 2 years and 139 days of service went to the arbitration.
Even with one fewer start, Parker would have been way over that, anyway.
isn’t it calculated differently depending on what year you it is in year 2-3?
You it is in?
I SAY WHAT I SAY!!!!
Put a Milo on yourself.
It wasn’t that he accrued a full year last year (which I think happens if you’re down for < 10 days), it's that last year's service time plus what he had in 2011 adds up to a full year, making him eligible for free agency going into 2018 instead of 2019. b-ref now says Service Time (01/2013): 1.000 which is what we thought.
But he would have been a super two anyway, wouldn’t he?
For arbitration yes.
For free agency it’s a hard cut off at exactly 6 years service.
OK, forgot about that.
OK, so how much longer would he need to have stayed down last year in order for it not to add up to a full year? Billy obviously timed the move appropriately considering how the season played out, but I’m still curious.
If Godfrey started just one more game that Parker started, the A’s would probably not have won the AL West…
Maybe in that alternate universe Parker is pitching in game 163.
It was obviously the correct decision in hindsight, but not obviously so at the time (although I endorsed it because the thought of seeing Godfrey pitch again — ick). Interesting parallels with the Giants’ decision to wait even longer on Posey in 2010 and then also winning the division on the last day.
I think this was the one time that everyone agreed no matter if you prefer stats, “grit” contracts, whatever. No one wanted to see GG pitch again.
Not even his family…
No one wanted to see Godfrey again, but several people didn’t think it was the right time to bring Parker up because of that service time issue.
I was ok with it but only because I thought (wrongly) that they would do something like send him down for 10 days around the all-star break to get him back below the limit.
Sure, but that is the part you cover in what you expect to be different in 2013, not in what happened in 2012.
So you can say, the 2012 team was a 89 win team, but I think the 2013 will be 5 wins better because players evolved and will be half run better each. I am not going to split it in halves because it’s a hell lot of work and possibly impossible without writing a program, and frankly, I think people who try to do this are missing a point of two component projection. What the A’s were in 2012 and how much different do you expect them to be in 2013. First part is rather clear. Have it at the second part, we all should.
The A’s weren’t the best team in the AL West in 2012?! This makes me Pythed off!
No way! A whole hour??!!! That’s like 60 minutes!!
Fear not, I am certain the bnk will start locking the doors at like 10 til and not let you in.
Phew!!
a harsh reality in these economic times.
The economic times have hit the banks hard. Bank CEOs need food tuu!
They can eat tuubers and rotting tuumatoes.
You forgot tuuna fish.
And they can drink Tuu Buck Chuck.
Are you sure this isn’t how the A’s will stock their post-game spreads next year?
Bartolo will NEVER go for that!
No can do.
I don’t get the impression that he’s all that picky about what he eats.
True. He didn’t even wash the baseball before he ate it during the game in the winter league last month…
He definitely needs to have a full-time ball washer on his payroll.
“ball” washer would refer to Beltre, right?
Even those who arrange and design shrubberies are under considerable economic stress in this period in history.
This is great work, Elcroata. I had a feeling, even looking at the 1st order Pythag (I didn’t know about 2nd order), that this team got lucky last year. But I didn’t have to knowledge to find the evidence. Plus, I was too busy enjoying the ride to stop and think about it too much.
Another thing why I think this approach is helpful is because lot of people think in terms “this team is just as good as the last year’s” or “exchanging Hunter for Hamilton is a wash, so I expect them to win the same number of games next year”.
And this is where I think people are mistaken. If you really think that, for example, Angels are as good in 2013 as they were in 2012 (definitely point worth discussing and not one I necessarily agree with) then you should be expecting them to win 94 games, not 89.
I always expect the A’s to win 162 games…. it just never quite works out that way. :(