THE BOOK--Playing The Percentages In Baseball

I’ve used the tool extensively for my CSFBL teams.

I think the tool’s guideline would be, 1-2-4 as in the book, then the next two in 5-9, with the better OBA hitter at #9, and the other hitters 6-7-3-8, descending in overall hitting.

1-2-4-5-9-6-7-3-8?

How’d you come up with that, and in what context?

If putting your 8th hitter means slotting him in the 3rd slot, you are doing something wrong.  Is that how I’m reading that chart?

Yeah. It’s just a gut feeling from using it 50+ times with realistic lineups.

Oh my.  I’m really disappointed to learn that Pinto has kept this tool active.  It’s construction is simply wrong, for reasons not addressed in the Fangraphs post.  My comment there:

“...the entire construction of the tool is completely misguided.  It applies a fixed multiplier to the offensive production of each position.  But those multipliers (even accepting the questionable method used to generate them) are based on traditional lineup constructions.  So, for example, SLG from the cleanup slot is very valuable because the three prior hitters get on base a lot (typically).  The tool, however, allows you to construct untraditional lineups, but continues to assign the exact same value to each “slot” even while the lineup being tested places much better or worse hitters in preceding and subsequent slots.  The value of a cleanup hitter’s SLG is obviously not the same if you put your best three OBP hitters in the 5, 6 and 7 slots. 

The lineup tool as constructed cannot work.  It does not work.  And it never has worked.  I’ve never understood why David keeps it active.”

re: Bukanier. When you say ‘better OBA hitter’, do you mean the guy with better attributes (ex. high contact skills) or one who actually posts a better OBA? When i played CSFBL, i noticed that some of the results were unintuitive, based on what i thought the attributes meant - guys with good contact skills (but without good power) would end up posting absurdly low BABIPs and grounding into a lot of double-plays.

guys with good contact skills (but without good power) would end up posting absurdly low BABIPs

I don’t get how that’s counter-intuitive. Guys without a lot of power aren’t going to hit the ball as hard as a powerful hitter, which should naturally result in fewer balls in play falling for hits. The harder you hit a ball in play the more frequently it’s going to be a hit, with lots of random fluctuation of course.

It would be a fun little project to create a new version of this tool.  What would be a better approach? Using H/BB/2B/HR/etc instead of OBP/SLG?

I might have a go at this this week.

I meant just of those two hitters, the one with the better OBA relative to SLG.

I’ve strictly avoided CN players in CSFBL, but as you said, only with outstanding hitters it doesn’t hurt.

#5 wrote “The lineup tool as constructed cannot work.  It does not work.  And it never has worked.  I’ve never understood why David keeps it active.”

IIRCC Dave has tested it on teams actual runs produced with fairly decent results for an estimator.

Given the fact that managers will use traditional lineups with high OBP guys up top, high SLG in the middle, and wort hitters at the bottom, what exactly does not work?.  Adrian Gonzalez is never going to leadoff for the Red Sox.

If you have a problem with this, then how do you accept many of the offensive saber stats for players which treat the value of a walk and XBH the same regardless of where the player hits in the order.  Unlike at the team level, there is no way to validate these estimates, just an unproven assumption that it works at the team level so must work for individuals.

Most people understand the lineup tool is an estimate, and some are more interested in relative accuracy than absolute accuracy.

I guess I would understand the objection if there were other public options “with the same ease of use and free access”.  I do not know of any.  If there are, it would be interesting to see a comparison of the lineup tool vs the other options for 2010.

IIRCC Dave has tested it on teams actual runs produced with fairly decent results for an estimator.

So? Team-seasons are incredibly poor tools for judging run estimator accuracy, much less for going from evaluating the tool’s merits in extrapolating down to the individual lineup spots.

Lemme go ahead and expand upon that point. A best-fit regression model is designed to correctly predict what occurs given inputs within the range of the observed values used to construct the regression.

Okay, so you feed in a series of OBP and SLG values based upon observed lineups. But the tool won’t only put values in places where they match the observed range - you can test lineups with OBP and SLG values that don’t occur in real life, such as batting the pitcher fourth. Those values - are they meaningful? Not at all.

pft/10:  The tool “works” compared to what?  Look, there are two reasons you would value SLG or OBP more or less than usual in a slot.  One—the most important—is how many PA the slot gets.  That would be a simple adjustment to make.  The second is the context for that slot—how good the hitters before and after are.  This is where the lineup tool is fundamentally illogical:  it uses fixed multipliers for each position, while trying to estimate the impact of placing offensive assets in different slots. If you want to know the impact of having your pitcher hit 8th, does it make any sense to use the same value for OBP production in the 8th slot (despite it now being followed by a much better hitter)?  Or use the same value for the 9th slot SLG, despite it now following the pitcher?  No.

Anyway, the test is whether the tool can outperform simply weighting each slot by its PA, adding the total SLG and OBP, and estimating RS.  I doubt very much that it can.

I think maximizing the run potential of your lineup is a great idea.  What I am very uncertain about is that there are any public tools available that give you one of the most efficient lineups.  Seems like everything is theory at this point.
vr, Xei

Hey guys,

So, playing around this week I’ve put together a little tool that simulates seasons for a specified lineup.  Would like to get your take on the approach.  The tool is actually a simulator, to take into account the things Guy mentioned in this thread.  Because of this, I obviously can’t calculate every permutation of a given lineup as it would take too long, but you can tweak the lineup and run the simulations again to see the difference. 

This is very much in the development stage but I thought I’d post it here initially to get some feedback on my approach. 

Basically, all players are loaded with their 2010 stats and teams.  This is just a default but all stats can be adjusted up or down to get to what you want.  So, if you want to simulate Pujols as hitting 60 homeruns or 20 homeruns, you can just type it into his homerun column.  Once you’ve entered the stats as you want them, drag the players name into the box on the right. 

Once you have nine players, you can re-arrange at will or you can run the simulation.  The slider at the bottom controls how many seasons you want to simulate.  Obviously, the more seasons you do the longer it will take. 

If you’re interested, please have a look and tell me what you think about the approach.  You can comment here or email me by clicking on my name.

Here is the link

Some quick thoughts:
1) While it works in IE, it functions a bit better in Firefox right now (the dragging is a bit screwed in ie)
2) Works best with minimum resolution of 1280x1024

Thanks in advance. 

Ryan