10 Games In and Mariners Have Narrowed Gap
by Matthew on Apr 17, 2009 2:00 PM PDT
in Seattle Mariners News, Good
Email
PrintDave beat me to a point I was going to make so instead I will do what I do best, add more numbers!
Projection systems were pretty even in their voices on the AL West coming into Opening Day. CHONE, Pecota (now that they seemed to have patched defense) and ZiPS are three systems I feel worthy of paying attention to. Here was how they pegged each team's win total in that order.
Anaheim: 85, 82, 85 (avg: 84)
Oakland: 81, 84, 81 (avg: 82)
SeattleM: 78, 77, 79 (avg: 78)
Rangers: 72, 70, 69 (avg: 70)
It is still too early in the season to suggest that we should be altering any of these true talent predictions*. At best, you might bump a couple more innings to Bedard since he looks healthy. That's not going to change much. Assuming those predicted winnings percentags hold for the remaining games, here are the revised standings with the first nine (ten for Seattle) games taken into account**
ANA: 83-79
OAK: 81-81
SEA: 80-82
TEX: 70-92
We've gone from six games back to three. It would have been all the way down to one if they had won yesterday, but c'est la vie.
* I know it's tempting to think the fast start has shown something, but I am skeptical. Coming into the year, we had concerns over the lineup showing patience or power at the plate. We thought the defense would be good, possibly great. We thought the rotation would be average to slightly above with upside depending on Bedard's health. We thought the bullpen would be below average, especially with control, though not a disaster.
We quantified all of those and came up with a runs scored and allowed projection that fit in quite nicely with those of the listed systems. Has anything in the first ten games proved to you that any of those were off base?
** MATHS!
The past is over and done with and since we are assuming that our sample is too small to affect our overall prediction of team strength, we treat the past results as independent of the future results. That is, it's just like coin flips. Predict how many times a fair coin will land on heads out of 162 flips. That prediction should be 81 times. Now imagine that the coin has been flipped ten times and landed heads seven times so far. The correct prediction is still that the coin will land heads 50% of the flips going forward, which would yield an additional 76 heads ((162-10) * 50%). Those 76 heads plus the seven heads from before equal a revised predicted for the 162 flips of 83 heads.
Questions?
51 comments
|
0 recs |
Do you like this story?
Comments
Are you being funny Graham?
Cuz unexpected winning streaks are no more a predictor of sudden losing streaks than they are of more unexpected wins.
If we thought this was about a .500 team last week, we should expect them to go 76 and 76 over the rest of the season and finish with a record of 83 and 79. It’s a fallacy to predict that the M’s must now reel off a four game losing streak to ‘correct’ their hot start.
As someone posted yesterday in response to Dave Cameron on USSM, that’s the differnce between regression “towards” the mean, and regression “to” the mean.
Danny Blanchflower Lives!
Sweet.
The number of things that have to go right for the team to make up three wins is pretty small and well within possibilities. For instance, if Yuni plays average defense, Griffey is a slightly (5 runs) better than league average hitter, and Bedard pitches, say, 180 innings we’re pretty much there.
If on top of that Ichiro bounces back from his ‘poor’ performance last year, Lopez and Beltre maintain their 08 level of play and Silva and Washburn get results that align with their DIPS predicted talent then there is every reason to think this team can be playing for something meaningful in September for the first time in what seems like ages.
by Bearskin Rugburn on Apr 17, 2009 2:59 PM PDT reply actions
If nothing goes wrong we're a 90 win team
by Graham MacAree on Apr 17, 2009 3:17 PM PDT up reply actions
Only in baseball
can you lose 70 games in a season and say that nothing went wrong.
I'm more like I am now than I've ever been.
I'm not saying that if those three things happen the team WILL win 83-84 games.
I’m saying if those three things happen it’s a true talent 83-84 win team, maybe more as I haven’t crunched the numbers.
I know that things going wrong are as likely as things going right, but I also think that the 80 win prediction is based on certain things going wrong, ie Bedard gets injured, Griffey is a poor DH, Yuni is an abysmal defender, Ichiro’s eight seasons in the majors are unsustainable and so on.
by Bearskin Rugburn on Apr 17, 2009 7:14 PM PDT up reply actions
A good point was raised on USSM about strength of schedule
and while it’s tempting to say that Minn/Oakland/LAA is a pretty fair sample of the talent we’ll be playing this season (in otherwords, that they’re collective approximately all .500 ballclubs), it’s worth noting that none of them were at their best. Mauer was out, we never saw any of Escobar/Lackey/that other guy and faced Loux instead, and the A’s are out Duchscherer as well (I think his fill-in was Outman?, not sure about their pitching deptch chart right now really). In all reality, the true talent level of the teams we actually faced were probably less than .500, so perhaps it would be better to only add 2 wins onto the M’s projections rather than three. But then, we were without Ichiro and the A’s got to go against Jak, though turns out he might be our 5th best SP afterall. I saw his curveball for the first time last night in the 5th inning or so, I did a double take. Very nice.
Of course, then you probably have to ding Oakland and LAA a little from their original projections as right now they’re NOT true talent ~84 and ~82 win teams given all the pitching they have out. So we probably are only about “3 games back” all things considered.
There was a thread on Tango's website about this very topic.
Here’s a link. The main point is that Dave & Matthew are right.
Going just a bit deeper, the 9 or 10 games played (depending on the team) is not enough of a sample to change your true talent projections for individual players. Also, the schedule already played should not change projections more than one win. Even that one win change would only be in the most extreme case and the Mariners schedule to this point has not been against extremely good or bad teams.
Analysts
It seems really strange to me that so many analysts predicted in March for us to win 55-65 games this year, we have an at least average team talent wise, so even before the season I would have predicted us to be around .500, barring any major injuries. Guess I just see it differently than some of these experts.
It's a pretty easy way of telling whose opinions you should totally ignore
by Graham MacAree on Apr 17, 2009 3:46 PM PDT up reply actions
Like anyone who said this was a 55 win team??
by seattlesundevil on Apr 17, 2009 3:51 PM PDT up reply actions
Yeah, link please.
Considering that a sub-60 win team is pretty rare- you have to be pretty “special” to be the 2002-2003 Tigers (who are the last team to really be that wretched).
by eponymous_coward on Apr 17, 2009 4:00 PM PDT up reply actions
BUT THEY LOST JJ AND RAUUUUUL!
I will smash your face into a jelly.
by Phil Hatzenbuehler on Apr 17, 2009 4:07 PM PDT up reply actions
Steve Phillips said we'd gotten worse since last year.
It was on tv, I don’t know about a link.
by Rollo Tomasi on Apr 17, 2009 4:10 PM PDT up reply actions
I think he's using the term 'analyst' in the talking head, Steve Phillips sense of the word
because I really can’t think of ANY analyst (in the fangraphs, looking-at-data sense of the word) who’d have been close to 65.
If we're talking ESPN 'analysts'
These are the same guys that thought for sure that the Mariners were the favorites to win the AL West last year.
I wasn't really talking about any specific people
I’ve just been reading and watching a lot of 2009 predictions and the vast majority that I’ve seen have put us at under 70 wins.
Most people around here had 75-80
By the way, using ‘reply’ makes comment threads make a lot more sense. Please do so in future.
by Graham MacAree on Apr 17, 2009 4:23 PM PDT up reply actions
Huh. It may be that the much-discussed insular/cult-like nature of LL/USSM readers is true,
but I just didn’t see any 60-65-69 win projections (again, aside from ESPN or something that’s generated more for entertainment). Confirmation bias and all that.
I don't understand the coin flips...
I understand your point about it always being a 50-50 opportunity, but it doesn’t seem like that is totally accurate. If I flip a coin 100 times and the first 10 come out heads I’m still banking on the total number coming out near 50 for all of them.
Sure, the Mariners picked up 2-3 more wins then I thought they would through this part of the schedule, but predicting the season results includes my belief that they’re going to win games here and there that I didn’t foresee (and by the same token lose some I thought they would win).
I’m a golfer. If I normally shoot an 80 and I start out even par on the first 3 holes that doesn’t indicate that I’m going to play in the mid 70’s – it just means that the bogies are coming later in the round. It’s possible that I’ll play better then normal, but wins/losses and strokes on the golf course are not a 50/50 probability.
Teams have good days and bad days just like I have on the golf course. All the wins racked up now don’t increase the M’s predicted finish to the season (these streaks were already included with the original projections). What has changed is the teams wins potential. These 2-3 wins increase the likelihood that they finish better then expected because they’ve ‘stolen’ some games that we didn’t think they would win. However, it’s just as likely that they will go for a similar cold streak and lose some they expect to win – evening out before the season is over.
Even if my coin starts by landing on heads 10 times, it’s just as likely to land on tails 10 times before the season is over.
Your counting on the total number coming out near 50/50 fails to adjust to the new situation.
At the beginning, yes, your odds predict the most likely outcome at 50/50.
But now you’re in a new situation. The coin has already fallen 7-3, and that’s in the past. If you still think the odds of each individual flip are 50/50, the remaining 90 flips should fall 45-45, ignorant of what’s happened previously. Therefore, you get 52-48 as the mean at that point.
It would only normalize if you think we’re a worse than 50-50 team in any given game.
I can elaborate but I gotta run to the game. Hope that helps make more sense out of it for you.
RIP Nick Adenhart. You will be missed by baseball fans across all organizations.
by seattlecougar on Apr 17, 2009 4:49 PM PDT up reply actions
This is wrong.
If I flip a coin 100 times and the first 10 come out heads I’m still banking on the total number coming out near 50 for all of them.
You’re making the Gambler’s Fallacy mistake of assuming independent past events have a bearing on future probability. They do not. Under your line of thinking quoted, you are saying that the probability for a fair coin to land heads after the first ten flips drops to 44%. It’s clearly incorrect.
Karma isn't real.
Early results don’t affect later results. To use the coin flip analogy, if you got heads the first 10 flips, you should expect the total number to be 45/55 at that point. The remaining 90 tosses still have a 50/50 chance, and you can’t expect that there will be a second occurrence of a random lucky event to counter the first.
okay, but isn't baseball different?
don’t you already factor streaks like this in just like you would losing streaks?
Unless we’re going through the schedule and projecting wins and losses that way – then if I think we’ve picked up 2-3 based on the schedule so far I could adjust my projection up by 2-3 games. I’m I just misunderstanding is that what we’re doing right now?
I get the whole we should win 50% of the games remaining, but while that works great for coin flips it seems like baseball might not be quite that mathematical…
This early in the season it is that mathematical for the reasons I stated in the first post.
We have too small of a sample to justify changing our model. We basically treat each individual baseball game as a coin flip, but instead of 50/50, it has some sort of weight to it. For the Ms, based on the three projection systems, we give it a weight of 48% heads (win) and 52% tails (lose). If we flipped that coin 162 times, we’d expect 78 heads.
But we only have 152 games left to predict, so we flip the coin 152 times instead. We get 152 * 48% = 73 heads. That’s the important part. The next 152 games have NOTHING to do with the previous ten.
Then you add the 73 to the already achieved 7 heads and viola, 80 heads (wins).
This only works like this now because the season is so new. We would not, for example, model it this way after 81 games because at that point, 81 games is telling us a lot more about our initial projections. If the Ms are 45-36 after 81 games, well then, it’s likely that the individual player forecasts that went into that original 78-84 projection ended up off. If so, we’d adjust the Ms predicted win totals for the second half of the season upward to reflect their improved level of play.
But even then, sometimes they aren’t. If the Ms went 45-36 through their first 81 games, but were playing at a 78-84 level, we wouldn’t touch our model at all and it would project the Ms to win 39 more games.
No
It’s 50/50 every time you flip the coin.You can a million heads in row and the odds for the next flip won’t change.
The golf analogy is completely wrong. The two things are not at all related.
Your score at the end of 18 holes is the result of your cumulative score on 18 different holes. What you did on the 3rd shouldn’t have any impact on what you do on the 15th, but both will impact the final score. Each hole is its own event. And really each swing of the club is its own independent event that will eventually add up to a total number of independent events from which your final score is derived.
Let's not suck!
Right...
I’m just saying with the golf analogy that: while each of those holes are completely independent, I consistently come out right around 80 regardless of how well the first few holes go…it seems that the mariners record would be more similar to that (total record reflecting true talent) then a 50/50 coin flip.
And that's the fallacy of looking at the completed picture vs the ongoing picture.
See my response above.
Strokeplay vs Matchplay
mwalter
The golf thing is a bit misleading because stroke play is not a good analogue for a baseball season, which is like a 162 hole matchplay tournament.
Imagine you’re playing matchplay with another golfer sporting the same handicap. The form book says you should tie. It doesn’t say you’ll halve every hole, but the bookies would predict a that you’ll end up even after 18.
Now imagine you’ve chipped in a couple from 40 feet (this is very hard for me to imagine) while your oppenent has fluffed a sitter. You’re two up after four holes. Do you still think the bookies are going to predict a tied game? No. The form book now predicts that you’ll win by two holes with one to play. You haven’t suddenly become a better golfer, and your opponent hasn’t become a worse one, but I few chance events have swung things in your favor. No force exists in nature to swing them back.
Danny Blanchflower Lives!
Are you serious?
If so, ’that’s life’. You can use google for things like that nowadays.
by Bearskin Rugburn on Apr 17, 2009 7:17 PM PDT up reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
In Association With
Big League Stew - MLB - Yahoo! Sports
- Photo: Padres fan regrets letting mother capture moment with Trevor Hoffman
- Marlins agree to unretire No. 5 for Logan Morrison, who will honor his father’s favorite player
- Weatherman Derek Holland does Dallas TV
- Ask Alex: Will Julio Teheran or Matt Moore win rookie of the year?
- Ku Klux Klan, umpires and bowling: Why the minor leagues tossed out Jim Evans academy
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
