I got this from this post at Beyond the Boxscore. I ran a binomial distribution for the Reds to see what their probability is of winning a certain number of games given the projected win total of 79. The graph is the result.
Based on this graph, they are guaranteed to win 43 games (woohoo!). They have an 81% chance of winning 71 games. They have a 67% chance of winning more games than last season and a 41% chance of winning 81 games. The last three points are 85 wins (25%), 90 wins (9%), and 100 wins (0.71%). So, at least there's a chance!
10 months ago
Slyde
40 comments
0 recs |
Email
PrintComments
Thank God
you’re back. I thought RR had been taken over by the mafia.
"I'm standing at the plate thinking, Where are my hands? Where's this, where's that? Before I know it, the pitch is there and I'm completely out of rhythm." -
-- Adam Dunn
by Joe Nolan's Glasses on Jan 11, 2009 7:41 PM EST reply actions 0 recs
eh
once you see the stuff I’ve got brewing, you might wish I wasn’t around.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 11, 2009 8:24 PM EST up reply actions 0 recs
unless it's a new non alcoholic beer
i think you’re safe
by SadbutTrue on Jan 12, 2009 8:17 AM EST up reply actions 0 recs
It would be interesting to see
All of the Central on one chart, just to show how much further we have to go.
With that said, I don’t see any way that the Brewers are nearly as good as Marcels projects them to be.
by dougdirt on Jan 11, 2009 8:31 PM EST reply actions 0 recs
that's partly why I used 3 different projection systems for my WAR projections
for fear that one system might exaggerate a team in one direction or another. I was pleased to see the Replacement Level Yankee Weblog come up with a 79 win projection for the Reds as well since it makes the 79 wins I’ve projected for them seem more reasonable.
Then again, I wouldn’t be shocked to see the Brewers win 87 games either.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 11, 2009 8:36 PM EST up reply actions 0 recs
I just don't see the pitching
for them to put together that many wins. They lost some pitching and I see nearly everyone else regressing to the norm with the exception of Gallardo, who just had a small sample. They might score 780-800 runs, but they are going to give up 750+ according to my math. Nearly every single pitcher on their team well outperformed their FIP and xFIP last year. I see some serious steps back being taken across the entire staff. Who knows, maybe their defense is something real special and it continues?
by dougdirt on Jan 11, 2009 9:01 PM EST up reply actions 0 recs
Well
800 RS and 750 RA is 86 wins. I know those were the extremes of your examples, but that’s still within reason for an 87 win projection.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 11, 2009 9:07 PM EST up reply actions 0 recs
it is?
a 50 run differential is five wins above .500, putting them at 83.5 W and 78.5 L. or am i missing something?
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on Jan 11, 2009 9:42 PM EST up reply actions 0 recs
Using pythopat (sp?)
I get a power of 1.85 for a 9 R/G environment like we had in 2008.
using excel, I did power(800,1.85)/(power(800, 1.85)+power(750, 1.85)) = 0.529814, which is 85.829 wins.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 11, 2009 9:52 PM EST up reply actions 0 recs
I just realized the problem with this this morning
a 50 run differential doesn’t mean 5 games over .500, it means 5 wins over .500. Every ten runs gives you an additional win. So that would be 86 wins, not 83.5.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 8:08 AM EST up reply actions 0 recs
You know, you're right. Learn something new every day.
Ten runs = one extra win AND one fewer loss.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on Jan 17, 2009 12:38 PM EST up reply actions 0 recs
i could buy the defense argument
they are strong defensively at every position outside of the infield corners, but even with that i dont think the pitching will hold up. losing CC and Sheets and not adding anyone to make that up is going to be too much to overcome. i can see them winning 83-84, but anymore would be a stretch, IMO.
by Charlie Scrabbles on Jan 11, 2009 9:08 PM EST up reply actions 0 recs
an 83-win team (talent-level wise) has a 67% chance of winning 87 games
so, it’s not out of the realm of possibility.
Damnit, I should go blog at BCB.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 11, 2009 9:11 PM EST up reply actions 0 recs
fair-weather fan, eh?
i cant blame you.
if you leave, can i have your ticket to Fantasy Camp?
by Charlie Scrabbles on Jan 11, 2009 9:26 PM EST up reply actions 0 recs
Where is the Manager's incompetence included in this this complex statistical projection?
Your conjecture that – ’ an 83-win team (talent-level wise) has a 67% chance of winning 87 games
so, it’s not out of the realm of possibility.’ cannot possibly take Dusty Baker’s management ‘style’ into account.
I see it more as an an 83-win team (talent-level wise) has a 67% chance of winning 87 games with a manager that drops the 83-win team talent level wise to a 67% chance of winning 77 games. Dusty has got to be worth a least five unforseen losses.
Mads, he is a crazy old man. Plus we have kinda had the same sabrametric evolution togeather.
Justin
by Madville on Jan 12, 2009 4:50 AM EST up reply actions 0 recs
Hmmm
Think I’m missing something here regarding your methodology. Are you just taking the projected win percentage using 79 wins (48.77%) and running it through a binomial? If so, either I’m way off base or the numbers have gone screwy.
If I understand your methodology correctly, I get a slightly different distribution: 91% for at least 71 wins, 41% for 81 wins, 19% for 85, 5% for 90 and 0.06% for 100.
On your last statement, even if I’m misunderstanding you regarding the distribution or methodology, this would seem to be saying that a team that should win 83 games would be underperforming if it did.
by RedStalk on Jan 12, 2009 10:56 AM EST up reply actions 1 recs
Yep
I screwed that one up. I forgot to subtract it from 1. It’s actually a 33% chance of winning 87 games. Math is hard.
I’m not sure where you got your numbers from though. They seem low. I used the formula in the post at BtB that I referred to above. in Excel, it is
=(1-BINOMDIST(87/2, 81, 83/162, TRUE))
where the 87/2 represents the 87 win prediction. The 81 is because putting in 162 craps out in Excel – it’s also why the 87 is divided by 2. The 83/162 is the winning percentage of our 83-win talent level team. And TRUE makes the function cumulative. I’m not a statistician, so I can’t really guess why your numbers came out different, but that’s the function that I used.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 11:25 AM EST up reply actions 0 recs
Allright. I think I understand...
If I understand correctly then, what you’re doing then is calculating the probability of winning 43.5 games in an 81 game season. While this is proportionally correct it wreaks havoc on the variance and does not produce the correct distribution. You’re looking at the same sample winning percentage, but the increased sample size will decrease the variability within the sample winning percentage.
It would be akin to saying that the probability of winning at least 3 of 5 is the same as winning at least 6 of 10. (Using a win % of 0.5, it’s 50% for the former and 37.7% for the later) Same win percentage of 0.6, but not the same probability of occurring.
Keep up the quality analysis. This is good stuff and consistently saves me from rampant boredom at work. Hope you don’t mind me poking at it.
by RedStalk on Jan 12, 2009 12:14 PM EST up reply actions 0 recs
Yeah. You can't just divide by 2
with the binomial distribution, since it’s a discrete probability distribution.
What values make Excel crap out, I wonder?
We want to build long period of time. I didn’t come here for the shot run.
by Gray on Jan 12, 2009 12:28 PM EST up reply actions 0 recs
probably user incompetence
see below
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 12:29 PM EST up reply actions 0 recs
I still haven't figured that out...
Part of it is probably that I’m using open office, but I did the exact same thing that made it crap out before and it didn’t this time, so whatever, now it works without dividing by 2 for me
---
Juuuust a bit outside!!
http://www.rightfieldbleachers.com
by Jack Moore on Jan 12, 2009 6:01 PM EST up reply actions 0 recs
you and me both
I think the computers were out to get us yesterday. :)
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 6:09 PM EST up reply actions 0 recs
good point
I had an inkling that something screwy might happen with the 81 game season vs. 162 games, but I don’t know enough to be able to tell. I’m the dangerous kind of stathead that knows just enough to do something wrong.
You know what’s funny is that I tried it with 162 games last night and it didn’t work. I just tried again and I got a result (24% for an 83-win team winning 87 games). I’m sure it was my fault, but I’m gonna go ahead and blame Microsoft. Stupid Bill Gates! I’m glad you are no longer the richest man on the planet!
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 12:28 PM EST up reply actions 0 recs
Whenever I'm baffled by math...
Tanzen!
by Verka Serduchka on Jan 12, 2009 1:12 PM EST up reply actions 0 recs
I don't think that was the image you were looking for.
We want to build long period of time. I didn’t come here for the shot run.
by Gray on Jan 12, 2009 1:17 PM EST up reply actions 0 recs
Is this a jedi mind trick?
Tanzen!
by Verka Serduchka on Jan 12, 2009 1:44 PM EST up reply actions 0 recs
pwnd!
Think of me what you will...I gotta little space to fill
by chandrathan on Jan 12, 2009 1:29 PM EST up reply actions 0 recs
God, that's a great "don't hotlink" image
"Aio, quantitas magna frumentorum est."
by jch24 on Jan 13, 2009 10:39 AM EST up reply actions 0 recs
it's great becaue it references lolcats!
Also, I'm given to understand that your mother is overweight.
by chandrathan on Jan 13, 2009 10:45 AM EST up reply actions 0 recs
Cutting things in half changes the numbers.
And 162 doesn’t crap out Excel, actually. It craps out EditGrid.
=1-BINOMDIST(winTotal,162,avgWin%,1)
If you change the 1 to zero, you’ll get a non-cumulative output (will look like a bellcurve if you do it for lots of winTotal numbers.)
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on Jan 17, 2009 12:42 PM EST up reply actions 0 recs
To address the point made in the title of that post:
Flipping 7 out of 10 heads is a lot more common than flipping 70 out of 100 heads. As a .480 win percentage team, it’s a lot easier to win 60% of your games over 81 games than over 162 games.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on Jan 17, 2009 12:43 PM EST up reply actions 0 recs
rec'd for nerdiness
and for smacking down Slyde’s incorrect math.
We want to build long period of time. I didn’t come here for the shot run.
by Gray on Jan 12, 2009 12:29 PM EST up reply actions 0 recs
I like to be smacked down on my math because then I actually learn how to do it right, or at least better
My wife studied Biostats, but she ends up being pretty much worthless to me because I can’t ever explain to her what I want to look at or why. It turns out a philosophy degree really isn’t worth much of anything practical these days.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 12, 2009 12:31 PM EST up reply actions 0 recs
poppycock!
a philosophy degree gets you lots of practical things! like you can…understand and make fun when someone misuses the phrase “per se”. or you can get discounted subscriptions to The Economist. or you can be the only person to get it when someone shows up to your Halloween party dressed as a monad. or you can attempt to logically validate your education decisions on a baseball weblog of questionable repute…and succeed!
by Charlie Scrabbles on Jan 12, 2009 3:54 PM EST up reply actions 0 recs
If you did that, it would be best to use bell curves, not cumulative distributions
That way you could see where the bell curves overlapped. That’s is where the upper tail of one team’s curve actually intersected the lower tail of a better team’s curve, showing the probability of a worse team actually winning more games.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on Jan 11, 2009 9:40 PM EST up reply actions 0 recs
i'd like to see that graph done for the 1990 team
by Daedalus on Jan 13, 2009 12:51 PM EST reply actions 0 recs
I'm not sure what you mean
based on their Run Differential, they were basically a 92-win team and they won 91 games. The graph only makes sense if you are trying to guess the chances of a possible outcome. We know the outcome.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 13, 2009 1:02 PM EST up reply actions 0 recs
i mean, i'd like to see a graph like that done for the 90 team before the 1990 season
I’d be interested to see if they had a chance in hell of winning the world series that year. I don’t think the probability would have been very high, because no one expected much from them that year.
by Daedalus on Jan 13, 2009 10:55 PM EST up reply actions 0 recs
If sabermetrics had been more advanced back then
I think they would commonly had been considered a bounce-back candidate. The drama of the whole Pete Rose scandal could have been considered a serious distraction, and the team clearly had a lot of talent as evidenced by finishing 2nd in 4 of previous 5 years. Their probability of 90 wins may not have been incredibly high, but I’d guess they would have projected in the mid-to-high 80s for wins, which wouldn’t make 90+ unrealistic.
Then again, part of the reason they had such a good season was because Jack Armstrong was surprisingly good and Danny Jackson had a big recovery from an awful 1989. It’d be interesting to see how they would have been projected, sabermetrically.
We’’re in process of trying to a guy with a trade record of working with pitches
by Slyde on Jan 14, 2009 8:27 AM EST up reply actions 0 recs
i would like to be old enough to have seen the 1990 team.
"Oedipus ruined a great sex life by asking too many questions!"- Dr. Stephen T. Colbert DFA.
by justin007000 on Jan 13, 2009 3:43 PM EST up reply actions 0 recs
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recent FanPosts
FanShots
Quick hits of video, photos, quotes, chats, links and lists that you find around the web.
Recent FanShots
Certain photos copyright © 2009 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.
