How Billy Beane are we really ? - how it ended up Topic

Posted by Trentonjoe on 2/21/2013 9:28:00 PM (view original):
It's been a long time since I had Pre-Calc, but to make a bell curve you need an exponent right?   Log5's are just multiplication?  There's just a ton of multiplication?

Either way, the extremes are  more successful than they mathematically should be in my experience.


That's the log5 calculation.  To calculate estimated run production (using obp and slg) u would use exponents.  I don't think the "extremes" perform better than they should.
2/21/2013 9:45 PM
Also, why do you say that Cy Williams is so good?  I looked at the performance history of the 1923 Cy Williams (a popular choice in OL).  His numbers don't seem out of the ordinary to me.
2/21/2013 9:48 PM
I tend to agree with the OBP guys, but I think a bit of managing is called for here. If the opposing SP allows walks, use the OBP, and vice versa.
2/23/2013 10:34 AM
Thanks pfattkatt, my old friend. Nice to see you active on the forums again. Missed you. Not to mention that boogerlips has been dissing your chess playing on the "what movies do you stay up to watch" thread. 

Good point about managing. I will try to keep up on opposing pitchers. Doing so daily is not always practicable but I will do my best. Thank again everyone. 
2/24/2013 5:02 AM
A quick update - after 7 games, in which I decided to platoon the two OFs to see who performs: 

Jack Graney (Mr. OBP) - hitting .412 with a .412 OBP - meaning zero walks
Chicken Wolf (Mr. AVG) - hitting .467 with an OBP of .500. 

Go figure.
2/26/2013 4:35 AM
Posted by italyprof on 2/26/2013 4:35:00 AM (view original):
A quick update - after 7 games, in which I decided to platoon the two OFs to see who performs: 

Jack Graney (Mr. OBP) - hitting .412 with a .412 OBP - meaning zero walks
Chicken Wolf (Mr. AVG) - hitting .467 with an OBP of .500. 

Go figure.
I'd start both of them with those stats! Keep us informed as to the progress through the season.
2/26/2013 4:42 PM
Frazzman80, I have been platooning them. Oddly enough we have faced a lot of lefties so they have the same number of plate appearances - 27.

After 10 games: 

Chicken Wolf (Mr. AVG)  is batting .400 with an OBP of .444

Jack Graney (Mr. OBP) is batting .318 with an OBP of .318 (no walks !)


2/27/2013 4:14 AM
Hopefully the message board jinx does not apply to forum posts, but so far it has been an embarrassment of riches:

Jack Graney 56 PA .362 OBP: .412
Chicken Wolf 30 PA .357 OBP .400 
 
And since my regular 2B and leadoff hitter Hub Collins has been hitting under .200 and not getting on base, and I noticed that Wolf plays 2B not greatly inferior to Collins, I have shifted Wolf to 2B for the time being so they are both playing every day. 

With two good first basemen, Wally Pipp LH and Dave Foutz (later a good pitcher for whom the SIM does not have stats in the database unfortunately) RH, who I am platooning, I have put Foutz as the back up LF/RF. Defense will suffer somewhat from these moves, so I will have to watch just how badly, but it should generate some offense in an offense-heavy league (at least for these seasons).

3/1/2013 4:32 AM (edited)
Two words:  sampling error.

The sim has no memory, no 'streaks': if someone is 'hot' (overperforming) it's just a statistical anomaly.  Actually, an argument can be made that you should trade/dump a hot player, as he will regress toward his mean.  If you think you were smart/lucky getting an overperforming player, it was the latter.
3/2/2013 4:42 AM
Yeah but nobody said either that these two players were "hot" or on a "streak". As to overperforming, that also can't be determined on such a small sample. In any case, even 162 games could be considered a relatively small sample size, which is why players do not perform exactly the same way each time they play the same season. 

So while I don't expect these particular level stats to last, I don't think taking these two players out of the lineup BECAUSE they are hitting well and getting on base is a strategic move that most people here would agree with. 

As to my being smart, they kind of pay me for that, but it does not apply to baseball on WIS admittedly. So I don't take credit, but don't think I have claimed any particular smarts in deciding to put two players with good OBP in my lineup. 
3/2/2013 7:15 PM
Posted by pinotfan on 3/2/2013 4:42:00 AM (view original):
Two words:  sampling error.

The sim has no memory, no 'streaks': if someone is 'hot' (overperforming) it's just a statistical anomaly.  Actually, an argument can be made that you should trade/dump a hot player, as he will regress toward his mean.  If you think you were smart/lucky getting an overperforming player, it was the latter.
That aguement would be wrong.  The sim doesn't remember the last game anymore than it remember the last season.
3/3/2013 6:10 PM
Posted by ncmusician_7 on 2/21/2013 9:48:00 PM (view original):
Also, why do you say that Cy Williams is so good?  I looked at the performance history of the 1923 Cy Williams (a popular choice in OL).  His numbers don't seem out of the ordinary to me.
Cy Williams averages around 44 HR's per seaon in OL.   If you compare him to other guys with 600 PA and a hr/100AB# of 9 that seems about average for players before the steroid era in baseball.  

Once you get past 1994, those few guys that meet those requirements (Bonds, McGwire, Bagwell) and the hr's per season drops into the mid 30's.
3/4/2013 9:47 AM
Posted by zubinsum on 3/3/2013 6:10:00 PM (view original):
Posted by pinotfan on 3/2/2013 4:42:00 AM (view original):
Two words:  sampling error.

The sim has no memory, no 'streaks': if someone is 'hot' (overperforming) it's just a statistical anomaly.  Actually, an argument can be made that you should trade/dump a hot player, as he will regress toward his mean.  If you think you were smart/lucky getting an overperforming player, it was the latter.
That aguement would be wrong.  The sim doesn't remember the last game anymore than it remember the last season.
The argument wouldn't be wrong; it just wouldn't be particularly useful.

Given an infinite number of opportunities in a computer gaming model - assuming the model is consistent and not flawed - any given outcome will reflect the assumptions when controlling for externalities.  In this case, given an infinite number of plate appearances a batter's AVG will equal the assumptions: a .280 batter will bat .280, all other things being equal.  Therefore, if a player has a streak in which he bats .350 instead of his programmed .280, ultimately his AVG going forward will be less than .280.  Unfortunately, this information does us no good, as it may take thousands of plate appearances for his AVG to 'return' to .280.  As for trading a 'hot' player: all other things being equal if I have a .280 hitter batting .350 and someone's willing to offer me value based on his batting .300 (let alone .350!) I make the trade in a second.
3/6/2013 5:03 AM
Zubinsum is right,  you're wrong.  The Law of Averages is false.  The sim treats every PA in a vacuum.  If you take the same player who "should" hit .300 in the sim on two different teams and they both have 500 PAs, it's possible that one is hitting .250 and the other .350.  The expectation value over the next 10, 100, 1000, 10000, or however many PAs you want is that both will hit .300.  It is absolutely, 100% false to think that the guy who has performed worse to date should get more hits to compensate.  There is no memory of the fact that he's hit worse to date.  The next 1000 PAs for each are just 1000 PAs, they should both hit .300.

No CORRECT argument can be made that you should trade or dump a hot player unless you can trade him for more than he's worth.  You don't trade him for his value based on the assumption that he'll have to play below par.

3/6/2013 6:59 AM
Given an infinite number of opportunities in a computer gaming model - assuming the model is consistent and not flawed - any given outcome will reflect the assumptions when controlling for externalities.  In this case, given an infinite number of plate appearances a batter's AVG will equal the assumptions: a .280 batter will bat .280, all other things being equal.  Therefore, if a player has a streak in which he bats .350 instead of his programmed .280, ultimately his AVG going forward will be less than .280.  Unfortunately, this information does us no good, as it may take thousands of plate appearances for his AVG to 'return' to .280.  As for trading a 'hot' player: all other things being equal if I have a .280 hitter batting .350 and someone's willing to offer me value based on his batting .300 (let alone .350!) I make the trade in a second.

The part in bold is the problem since it straight up isn't true.  That guy's average going forward should be .280, not less than .280.  He's still a .280 hitter and each PA is taken in a vacuum.
3/6/2013 7:03 AM
◂ Prev 12345 Next ▸
How Billy Beane are we really ? - how it ended up Topic

Search Criteria

Terms of Use Customer Support Privacy Statement

© 1999-2024 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.