Five Days
This will sound overly dramatic, but it’s simply a realistic assessment of where this team stands – the Mariners have exactly five games to save the 2008 season. If they don’t perform well between now and Sunday, the rest of the year will simply be playing for second place, because the hole will be too large to overcome. They have five home games against below average teams, starting with Bedard and Felix to close out this series.
If the Mariners don’t win three of the next five games, they can cash in their chips and go home. The deficit would just be too large to believe that anything short of a miracle could cause them to win the division.
With the Angels and A’s both off to strong starts, you have to set the bar for minimum wins required at 93. It would take some historic collapses by both clubs for the division winner to finish with less than 92 wins, especially considering how well they’ve both played through the first five weeks of the season. So, the M’s target has to be 93+ wins.
If they go 2-3 to finish out this homestand, that would put them at 16-23 with 123 games to go. In order to finish the year with 93 wins, they’d have to play .626 baseball, a 77-46 mark. This team is just not capable of playing that well. Very few teams are, and this is certainly not one of them. If they can go 4-1, they “only” have to play .608 baseball the rest of the year – very freaking hard, but at least within the ream of possibility.
Two wins or less and it’s pull the plug time. The season will officially be over, and they can start shopping the veterans around and setting up the interview list for the new front office and management staff. Three wins keeps the team on life support, with a faint chance of contending still hanging by a thread. Four or five wins gives them a little bit of life as they head out on the road next week.
But they’re at a crossroads. Finish this week strong or fold your tents, because the rest of the year is a waste of time if they can’t beat the Rangers and White Sox at home with the season on the line.
I don’t see how anyone can be anything but pessimistic about this season … and I believe right now that pessimism like many people on this blog have so eloquently put it is actually: realism
pointing to 95 is what I’ve been doing SINCE 95 … and here I am more than 10 years later never getting a run of that type – and do not ever expect to see something that surreal again.
Also … lets be real – that lineup BLEW our current one out of the water… we have no (zero, zilch) comeback ability this year.
yeah. sorta like those losers Bob Melvin, Felipe Alou, Joe Torre … boy, when that Torre sits on the dugout bench during a game, hands in pockets, stone-faced during a game, you can really see why it keeps his team from winning.
If fire in a manager means one who goes out on the field and then gets tossed– McLaren was ejected 3 times last year. In a half-season of managing. The same number of times Scioscia did, in a full year.
Oh my god, I hope Tex doesn’t own a racing horse.
One of the reasons I read this blog is because of the understanding of statistics such as those referenced in the mentions of the gambler’s fallacy and monty hall problem. Excellent.
Sexson can approach .250ba (which puts his OBP around .400)
Even in his prime, Sexson has never OBP’d .400. He’s not going to do it in the twilight of his career.
I root like crazy for Richie and still think he has the chance to put up a non-awful season, but an OBP of .400 is out of the question. If Richie hits .250, his OBP is more likely to be .350, probably even lower considering the difference between his BA and OBP over the past few years.
My point is that I never really thought Joh was a .250. I thought he would be a .230 hitter some of the time and a .270 hitter some of the time (and everything in between and beyond). In coming to the original conclusion that Joh would hit .250 you have to assume that the “colds” and “hots” AVERAGE out to that. If you didn’t think “Joh” could hit .270 or better for a good chunk of the time then predicting that he would hit .250 never made sense.
Similarly, if you didn’t think the Mariners could play like a 100 win team for extended stretches it never make sense to pick them to win 93 game in the first place.
There is a difference between expecting a player/team with a certain history to “turn it around” and planning on your team playing at a 100 game level for an extended period just because you expected them to win 93 games and they are 6 games under .500.
Example: we can say that Ichiro will probably still end up with a very good average because we have regularly seen him start slow and have seen him maintain very high batting averages for extended periods. If I said that Ichiro! could still hit .330 this year some people might disagree, but most probably wouldn’t dismiss it as irrational based on past information. Unfortunately, these Mariner’s don’t have Ichiro’s history to fall back on. I don’t believe that falls under the Gambler’s Fallacy.
The slippery slope, as always, is when does a slump (or hot streak) stop being an anomaly and a new point of reference going forward.
Also, thanks for the discussion. I’m hope you can see what I’m driving at, or at least why I think it differs from the gamblers fallacy.
@ Jeff Nye – No problems. I didn’t think you were picking on me – you were perfectly clear about that :D. I just wanted to clarify.
Fascinating thread. And a huge example of why we cannot trust thinking that merely feels or sounds logical. We see so much of it in politics – messages designed not to be true, but to SOUND true, that it is interesting to see it at play in what is really math logic.
The idea that it will all even out starting… now… sounds right. It sounds so right that this happened:
1. Someone makes the claim.
2. Dave refutes the claim – very logically, I thought.
3. The power of “it sounds right” is so strong that even faced with clear refutation, they still stick by their original argument.
4. Colm politely and clearly shows by example why the logic is flawed. The coin flip.
5. The belief is STILL held, claiming that the logic doesn’t apply to their premise.
6. Then DMZ and others weigh in, putting it in other ways, trying to make it clear.
7. Then, a NEW PERSON comes in to defend the original premise, convinced that all the others are the ones that don’t understand the idea.
It’s actually kind of awesome to see this level of discussion on a baseball board.
But really… it is the Gambler’s Fallacy. It really is.
Jeff, I liked your definition of realism. To that end, the most likely outcomes at this point (if you believe model that the outcomes of games can be thought of approximately independent coin tosses of a biased coin) are:
If you believe the M’s are a…
95-win team: 89 wins (78, 99)
92-win team: 87 wins (76, 97)
89-win team: 84 wins (73, 95)
85-win team: 81 wins (70, 92)
82-win team: 79 wins (68, 90)
79-win team: 76 wins (65, 87)
The numbers in parentheses are 95% confidence intervals.
RoninX:
Thank you for trying to help me clear up my point.
Dave:
The gambler’s fallacy is dependent upon each hand (or coin flip being independent). My argument is that each game, each at bat is not independent. We cannot assume that each game is completely independent, or each at bat is independent. Hitters figure out their swings, start seeing the ball better, or start guessing pitches better.
Johjima, as such a perfect example had stretches last year where he .191/.217/.326 (during the month of July). I’m saying that only taking the sample at the beginning of the year and stating that all of the remaining games of the year he will only hit at his career averages is wrong. Johjima also hit .375/.402/.568 during August last year. Those also aren’t his true talent numbers either. USSM discusses reverting to the mean and my point is that Johjima’s (and the rest of the team) will eventually revert to the mean.
This team, if most of the players play to expected, extrapolated abilities (which means that most of the players remaining will have to overplay their expected talent levels*) they should win about 80-85 games. Not the 93 they need to win the division, but also not the 72 to fall behind the Rangers.
I guess that if you can make the argument that each game, each at bat is independent of the previous then yes, I am biting hard into the Gambler’s Fallacy. However, I don’t believe they are independent.
Finally, to get to 80-85 wins, the team has to play better than .530 baseball the rest of the way (81 or 82 wins). Difficult, yes, but far from impossible. If they had played .530 baseball from the beginning of the season, they would have won 85 or 86 games (I’ll let you round how you like).
*So yes, I am saying that they will overplay their true talent level-for a portion of the season-as every player does every year. Are Mark Reynolds or Nate McClouth the best players in the game? No, but they sure played like it for a bit.
RoninX:
YES.
RoninX, your argument for Johjima is the same as the coin flip. When we flip a coin a hundred times, predicting a tails rate of fifty percent, we know in advance that there will be thirty percent and seventy percent stretches. That doesn’t mean that if it starts out thirty percent, it will be balanced by a later streak of seventy percent. Another thirty percent streak is just as likely. Similarly, when you say that Johjima will of course have .230 stretches and .270 stretches, and he starts out with a .230 stretch, the chance of another .230 stretch is just as likely as a .270 stretch. Do you see? So if he starts the season with a .230 stretch, he has made it less likely that he will be a .250 hitter on the year, if .250 is his true level of ability.
Every streak, bad or good, has a narrative ready to be imposed on it. We like pattern. Experimentation has shown time and again that humans believe in explanations for things that are overwhelmingly likely pattern distributions: people don’t understand how statistically common streaks are, just randomly.
Um. These SOUND good, but of these three, only the first really might occur as batters develop bad habits due to, say, injury.
Seeing the ball better? Athletes talk about it, but, rationally and mechanically, does that happen? And guessing pitches betters is just another way of saying random variation—there’s nothing changing here and there’s nothing DEPENDENT going on.
#102
You have to see a MAJOR difference in the teams they are winning with, and the M’s….Melvin won 95 games with the M’s…..but the talent was there.
The DBacks have TALENT…and more importantly, talent that PERFORMS….I’m pretty sure Beltre is VERY talented, but just isn’t performing. Ichiro…no argument on talent….but Pitching talent…we have a severe lack of that. And even the decent talent isn’t performing.
Torre, and Melvin don’t HAVE to have fire….they’re winning.
I didn’t want that entire thing bold. Sorry.
okobojicat,
Independence and the expected number of games that the M’s will win are separate issues. Mathematically, dependence between coin flips (at bats, games) does not affect our “best guess” about how many successes (hits, wins) we expect over the long run. What *is* true is that in a finite sample (say a season), we might expect more uncertainty about that “best guess” due to longer or more frequent streaks than we might have expected.
So even if you don’t believe in the independent coin tosses model, you shouldn’t quibble with the expected win total. And if you’re predicting that the M’s will “pick it up”, then you’re putting your faith in a strange sort of dependence which prolongs winning streaks but doesn’t do the same for losing streaks.
I think this is truer of probability than any other subject. Every argument about probability happens like this thread: a bunch of explanations with examples trying to demonstrate a counterintuitive truth, with a few folks who seem earnestly and sincerely unable to see it.
It is true that this is not as random as a coin flip; it’s possible that Johjima’s chances of hitting his true level have been depressed by some effect. It is still illogical to say “those slumping will get back to career averages”, as a group. As a group, they very likely will not.
Q: What is the difference bewtween Hillary Clinton and Bill Bavasi?
A: Hill throws her own money into a sinking ship while Bill uses other peoples’ money.
okobojicat,
Please read The Book. There is no evidence that hot streaks exists. People have looked in baseball and even in basketball, and there is not much evidence for it. Human’s are good at spotting patterns, even when they aren’t there.
Your basic premise is entirely wrong. Lets go with your argument that Kenji is going to be a .270 hitter for the year. He’s likely to have 25 game stretches hitting anywhere from .150-.390 (i’m not doing the exact math here, but bear with me). All of these streaks are not as equal however. We know that over most 25 game stretches the most likely scenario is for him to hit closest to his true talent level. Sure he could hit .390 for 25 games and erase the hole he has dug himself, but this is no more likely than him hitting .150 for 25 more games.
The only way you can guess at his final numbers is to take his true talent level going forward and figure out what he will do from this point, and then add it to what he’s already done. If you want to break out the math, we could even figure out how likely it would be for Kenji at this point to end up at particular averages at the end of the year.
Ronin/Objicat – probability doesn’t work the way you think it does. I’m sorry, I wish it did, but it doesn’t. We know what you’re trying to say, and it’s just empirically wrong. It would be fantastic if regression to the mean worked to cancel out improbable stretches, but they don’t – the consequences of an improbable stretch (in this case, the M’s going 14-20 in 34 games) still exist, and math doesn’t wipe out the deficit with a corrective positive improbability later on.
Unless you accept that paragraph as true, I’m not sure how we can help you.
@ Scraps – Believe me. I do understand what you are saying, and in some ways comparing a teams record the the coin flip is more apt than to a player’s average, but let me ask you this: how do you square your position with what I said about Ichiro?
Ichiro has never hit less than .303 (sorry that .330 in my post was a typo) – he is currently hitting .282. Don’t you expect Ichiro! to hit .303 by the end of the season, even though his start has made it more difficult than if he was hitting .372 right now?
Maybe the .303 example is too low, since if Ichiro hits hits career average from here on out (the coin analogy) he’ll exceed that. However, would you bet against Ichiro! hitting .320 or better this year knowing what we know about previous slow starts?
We actually have more information about players than we do in the coin flip analogy, which is what makes it imperfect. IF we have reason to believe (not just hope) that a team/player can substantially improve upon their performance based on actual information (not grit/character) then suppositions that a team/player can “turn it around” aren’t necessarily a fallacy.
I can’t help but think of the warning slapped to my mutual fund statement:
“Past performance is not a guarantee of future earnings” (or something like that).
Every at-bat is independent, based not a whit on the one before it and in no way influencing the one after it.
That said, I don’t think the gambler’s fallacy applies as easily to team performance. If the pieces change, so should our expectation of what the team will do going forward.
I accept this as true! Its true! It just doesn’t tell the whole story unless additional information/statistical layers are added.
This season is mostly playing out as I expected, though I am surprised by the lack of production of Ichiro, Johjima and Putz (I suspect the latter isn’t entirely over his injury), and the bullpen has been worse than I expected. I expect Ichiro and Putz to improve (unless Putz is badly injured), but I’m less sure about Johjima because of the tendency for catchers to go downhill quickly. Overall, I saw this as a .500 team, but while I think some regression to the mean will occur and they’ll improve somewhat, they probably won’t make .500 at this point, since they’ve dug such a big hole for themselves. Alternatively, they could start mailing it in (I think some players already are doing so), and get even worse (shudder). This team, like last year’s, never was a realistic contender, which is why the Bedard trade was so awful (and yes, I know I’m repeating myself here). And no, this has nothing to do with the team “not clicking” yet, or the manager “lacking fire” or any other intangible explanation; it’s a lack of talent, pure and simple. The 1995 team, by contrast, had a lot of talent (Johnson, Griffey, the Martinezes), including two sure first ballot HOFers (3 if you count the young A-Rod, though he didn’t really contribute much to that team). So there’s no point in making those comparisons. Let’s just hope that ownership finally starts learning from its mistakes.
Ronin,
Hate to be a parrot here, but Ichiro’s seasonal performance to date is “in the bank”. Part of the logical gap you’re making is applying that to what you’re expecting going forward.
Start from zero – pretend the season starts today. Your argument is suggesting Ichiro will hit roughly 10% better than his career average, based on… what?
His .282 is booked, so to speak. We should revise his season-end numbers downward because of it, expecting him to hit his career average from here forward.
#124
We’ve been hoping that ownership learns from it’s mistakes since about 2003.
RoninX,
I expect that Ichiro’s true talent level is around .330. He currently has 40 hits in 142 at bats. I expect him to get 534 more at bats (based on his at bats per game this season). If he hits .330 in those at bats he will have 176 more hits. that would give him a total of 216 hits for the season. This means that I expect Ichiro to hit .320 at the end of the year.
The argument that you and Objicat seem to be trying to make is that Ichiro is most likely to end up hitting .330 at the end of the season. For him to do that, he would have to hit .343 for the rest of the season. This is entirely possible. However, it is much more likely he will hit .330 for the rest of the season. He really is no more likely to hit .343 for the rest of the season than he is to hit .317 (which would make his final average .309).
I will say that in Ichiro’s case, we have another set of data that says he has historically hit less well in April, so one could reasonably argue in his case that when considering what he’s going to hit for the rest of the season it might be defensible to use his career May to September average.
The point is, though, you have to make an argument like that for every players who is at variance with their norms. And most of the arguments people make will be anecdotal, not supported in any way by data, such as the argument that a player — Johjima, say — has streaks because he is taking a good or bad approach, instead of simple random variance.
I just wanted to point out that in 2005 when Ichiro hit .303, his average after his 24th game was .356 and last year after his 24th game he was batting .281 (he ended at .351)
Ichiro
2008 BA: .282
2008 BABIP: .292
Career BA: .332
Career BABIP: .357
If his BABIP goes to career numbers, he’ll hit .344 for the season.
Don’t worry about Ichiro.
And please stop discussing whether or not he’s done or w/e.
Just checked: Ichiro’s career average from May on is .340. For what it’s worth.
But throughout his career Ichiro’s BA after April March (based on Baseball-Reference.com) is .340 (1387/4079). Its not that I hope he’ll hit .340 from here on out, Ichiro! has established a pattern of hitting much better after the calendar turns to May (.292 -> .340).
This is what I’m talking about in terms of having a reason to assume a player/team will turn things around, not just hoping they will. For whatever reason.
As I said, not much reason to think these Mariners will be a 93 win team. But 85ish still seems perfectly reasonable based on players past histories… though the two rooks are a bit of a wildcard.
not really related at all, but pretty ridiculous the last 96 games of Ichiro’s 2004 season he batted .401 he is the man
Agreed. With so many players at various “variances” at the moment (including Jose Lopez on the + side) this is a tough team to figure out right now.
Sorry for all the hubbub. I really just meant to say that the coin flipping analogy was simplistic, but then that is the nature of analogy :shrug:
All of this reminds me of the Chaos Theory….
“Lorenz had discovered that small changes in initial conditions produced large changes in the long-term outcome.”
So one small thing, or many in the M’s case, is causing large changes throughout the season…we’ll see how it all ends up after game 162.
I WANT to have hope, but it’s tough right now…10 walks a game…that’s tough to watch, and still have hope.
But throughout his career Ichiro’s BA after April March (based on Baseball-Reference.com) is .340 (1387/4079). Its not that I hope he’ll hit .340 from here on out, Ichiro! has established a pattern of hitting much better after the calendar turns to May (.292 -> .340).
If you want to state that you believe that Ichiro has a higher true talent level in May-September than he does in April, that’s a totally different conversation than the one we’ve been having. The earlier discussion was based on the presumption that the opinion of the team’s true talent level was an unchanging thing.
That’s why I presented two options to objicat – either he belives the team is better than he thought it was when the year started (true talent level increase) or he was buying into the gamblers fallacy. With Ichiro, you’re arguing the former point, but with the team, it’s still the latter.
Yeah to sum it up we’re at this point because the 2007 M’s overacheived, the FO ‘tweaked’ the team and now can’t figure out what the hell is going on…and we’re stuck with another losing season…am I getting it yet?
Sorry I have a headache…
Griffey anyone? *grin*
#137
I was wondering when the “Griffey” name would get started back up…
He’d fill the seats, but as for producing for the cost? Not so sure.
Yeah I’m past caring about that…
I just want a reason to watch this team again…
Like I saidbefore sign him now before he hits number #600
Dave’s already said that he’s working on a Griffey post. Please save that discussion for that post when it goes up.
Thanks!
and why would the Reds get rid of him before he hit 600 for them?
sorry jeff.
I’ll respond to that question in the upcoming greatly anticipated GRIFFEY POST!!
Sorry Jeff…I need a Red Bull
Players do not have a “true talent levelâ€, at least not in the sense that so many are using the phrase here.
When Raul Ibanez is facing a LHP, and his shoulder is hurting, and the shellfish he ate for lunch isn’t sitting too well, and he’s got a speck of dust in his eye, he’s not going to perform as the “true talent level†Raul we expect from his career or seasonal averages. Raul is not a .450 slugging “coin†– he has skill sets that are affected by context. If you’re playing poker, and you know that historically your opponent bluffs an average of 70% of the time, do you call his bluff without a second thought of the cards in play? The point is, you can’t just assign someone a “true talent level†derived from an non-context specific average and claim that your “true talent level†is a good representation of probabilities involved in any one event.
Yes, you can. It’s this belief that you can’t, and that everyone is special, and every situation is an exception, that leads to terrible, horrible roster decisions, and ridiculous notions like Miguel Batista reinventing himself into a frontline starting pitcher at age 38.
When dealing with probability, it is implied that you are speaking in generalizations. Trying to make adjustments for specific contextual situations, and throwing out the rules of probability in the process, is how you end up with Jose Vidro as your DH.
Trying to make adjustments for specific contextual situations, and throwing out the rules of probability in the process, is how you end up with Jose Vidro as your DH.
What about playing platoon splits? I can see how micro-managing based on small sample size contexts is folly, but so is making decisions without a player’s skill set and health in mind.
It makes me feel a little better to look around the league and see who else is pretty much done already. Pittsburgh, Cincinnati, and Texas are no surprise, but Colorado and San Diego? Last year’s World Series team and the team they beat in a one game, 13 inning playoff to get into the postseason? Those fans must REALLY be feeling the sting.
A player’s true talent level doesn’t change with varying conditions; it might change the way that talent gets expressed on a per-situation basis, but the underlying talent level doesn’t change. Raul Ibanez doesn’t become a better or worse player for even one at-bat based on the dinner he just got done eating, or any other random variable.
That’s why we call it true talent level; it is “filtered” from outside characteristics. The only way to do that, of course, is to look at a large enough sample size; random variation is exactly why you can’t look at one play on Sportscenter and decide, based on that highlight reel, that Derek Jeter’s a great defender.
Well, not if you expect anyone to care about your defensive evaluations, anyway.