Odds of Making the Playoffs

In this space, I pretend to use and understand sabermetrics.

In actuality, while I would love to know what BABIP means in essence and how it effects what we can further expect from a batter or a pitcher, I truly have no idea.

I often want to apply mathematics to my analysis, but I am never quite sure how to begin. I sometimes have ideas about how to proceed, but they are usually half baked. For example, when I wanted to figure out if Joe Torre has too quick of a hook with his starters, I had an idea of how to start, but once I had my pretty little table, I didn't quite know how to analyze it.

I try and look up the stats (such VORP), and understand the stats, and pass on the information in a meaningful way, but ultimately, it never happens to the extent that it could have been.

So when I want to figure out the odds of the Yankees and their 22-29 record of making the playoffs, I can go to Coolstandings.com and see they have a 10% chance, or I can go to Baseball Prospectus and see they have a 19.5% chance.

The question is, is looking at how the Yankees have done to this point in 2007, the best way of looking at how they will continue to do in the remaining 111 games this season?

To better answer that question, I turn to the email I received from sometime commenter and Knower-Of-All-Things, Anthony. He, unlike me, understands stats, and math, and can come up with some sort of formula to discover all things.

His interest regarding the Yankees' playoff hopes piqued from this conversation on Tango Tiger's website, and specifically a few comments made by one, MGL:

No matter how much we (analysts) show how current performance affects performance going forward (not much as compared to past performance, unless we have little past performance of course), EVERYONE, including people like Neyer and Law, and of course, every mainstream fan and commentator on the planet, attaches WAY too much importance to a team’s performance thus far in the season. People cannot imagine that the Yankees may be a true .575 team...That just boggles their mind.

It is MGL's belief that what seems like a large enough sample size (51 games) to determine some sort of projection moving forward, it is still too small to completely ignore preseason projections for individual players. Ultimately, MGL thinks that actual 2007 numbers (at this point in the season) should only account for 10-15% for a player's expected value going forward.

With that thought process, according to Anthony:

The correct way of predicting how the Yankees will do in the final 111 games is to look at the players, not the team. That is, figure out each player's true talent level, figure out how much playing time each will get, then add them up for the team totals.

At this point, I have to completely turn to the Knower-Of-All-Things, as the math leaves me slightly confused. That, and there is no real way to summarize this information any better than Anthony's original statement.

(Slightly-involved math - Should we use MGL's suggested 85/15 split? Tango uses a 5/4/3/2 split for the Marcels. That means he weights 2006 by 5, 2005 by 4, 2004 by 3, and league average by 2. Someone in the above discussion suggested a 6.5 weight for 2007. Let's say we bump it up to 7. A 7/5/4/3/2 split weights this year as 33% of the projection. If we assume 600 PA in all previous years, and 200 PA this year, that makes this year's weight just over 14%. Looks like MGL's guess was pretty darn close.

Considering most players don't yet have 200 PA, and we'll be using a "smarter" projection system than Marcel, that 14% should really be the absolute max. Instead, what I'm doing is adding 1000 PA of the preseason PECOTA projections to every batter's actual 2007 stats. For pitchers, I'm making it 1500 BFP worth of PECOTA. Derek Jeter leads the Yankees' in playing time thus far; he gets an 81/19 regression. Andy Pettitte tops out the pitchers: he gets an 84/16 regression. If anything, I'm actually overweighting 2007.)

Some interesting things come out when we do this. Check out what happens to Bobby Abreu:

2007: .228/.313/.289
PECOTA: .277/.389/.447
Updated: .267/.374/.417

Now here's Posada:

2007: .357/.414/.560
PECOTA: .259/.365/.443
Updated: .275/.372/.462

Repeat for every player, add together, wipe hands on pants. For offense, we'll use a simple Runs Created formula. So far the Yankees have scored 268 runs versus 268 runs created. I'd say we can live with that accuracy. For pitchers we'll use Component ERA. Regular ERA won't take into account luck and FIP won't take into account defense, so Component ERA is the best available option.

So that's the method. What's the result, then? What is the Yankees' true talent level?

.614.

Seriously. .614. Even with the awful start, this is a true 99-win team.

So how do the Yankees go from being a .431 team to a 99 win team? Well, Anthony answered that as well looking at three specific areas:

Luck - 13 wins. Believe it or not, the team has scored more runs than they've allowed, 268-244. They should be 28-23. A lot of this is due to a dreadful 2-10 record in one-run games. Even if you hate "luck" as an explanation, merely positing that Mariano Rivera will even vaguely resemble his old self from now on will dramatically improve the Yankees' fortunes in close game. Banning Luis Vizcaino from the park will help, too. That their performance thus far more closely reflects a .545 team than a .431 team is worth 13 wins the rest of the way ([.545-.431]*111).

Pitching - 3 wins. Two words: Roger Clemens. Replacing DeSalvo/Wright/Igawa with arguably the best pitcher in baseball is easily worth three wins by itself. Two more words: Phil Hughes. Even if likely improvements by Rivera and Mussina are canceled out by Pettitte and Bruney coming down to earth, the additions of Hughes and Clemens easily make the pitching far, far better. I was surprised this analysis only saw the team's runs allowed per game falling from 4.78 to 4.49, to be honest.

Hitting - 5 wins. The offense has scored 5.25 runs per game this year. This analysis says they'll score 5.76 the rest of the way. Essentially, while Rodriguez, Jeter and Posada are playing over their heads, everyone else is likely to improve. We know Cano and Abreu and Mientkiewicz and Damon are better than they've shown, to various extents. Giambi is old and chubby and may be on his way out; but there's still a very good shot his power returns. And Wil Nieves couldn't hit any worse if he used a whiffle ball bat.

Essentially, Anthony took reasonable offense and pitching improvements and the all important luck improvement ("the luck improvement, while huge, makes sense when you consider all the in-game pitching injuries and Rivera's late-game failures") and came to the conclusion that this Yankee team is not only capable of winning 90 games, but quite likely to do so.

How likely? Well, using a binomial distribution, and the aforementioned true talent level evaluator, here are the odds of various win totals:

• 81 games: 96.9%
• 90 games: 55.4%
• 95 games: 19.9%
• 100 games: 3.3%

So while it sucks watching the Yankees fumble games like this, it is still far too early to give up hope on playoff chances.

The question is now, does Giambi going on the DL for at least 3 weeks in favor of Kevin Thompson have a significant effect on anything from above?