Feb 28, 2010

Using High Draft Picks on Fantasy Starting Pitchers, A Study (Part 3)

You probably thought our series that studied high investment in starting pitchers was over!  Well, so did I.  But, we're forced to revisit it to a glaring omission.  We examined the myths around highly drafted starters being riskier or more injury prone than highly drafted batters.  Those both proved to be false.  But, those aren't the only reasons that experts avoid starting pitching early on.  One other main reason is that the drop-off at other positions is more severe, forcing you to wisely secure those positions before thinking about taking a starting pitcher.  That's the claim, at least.  So, let's examine it.

Avg. Positional Value by Draft Round (Chone)

Above is a chart that takes a look at the trend in average WERTH values by position each round.  With the starting pitcher trend highlighted in red, you can see that they do maintain value over the other positions throughout most of the draft.  Let's examine this further.  Here is a glance at the data behind that chart:

Much like the chart showed, you can see that the starting pitchers are not only valuable in Round 2 but also the most valuable pick in much later rounds.  So, if you kept drafting for value, you'd end up filling up your entire pitching staff right off the bat then left scrambling.  To show exactly how these positional values stack up each round, here's the rankings from the above data:

For the entire first 14 rounds, either starting or relief pitchers represent the most or second most value of the round.

So, for those who argue that drafting starting pitchers early is a fool's errand, there is certainly strong logic behind that.  Why use a 4th round pick on a starter when they're still extremely valuable compared to the rest of the board in Rounds 9, 10 and 11?  Though, I would argue that even if you start drafting pitchers later than most, you'll want to finish up your pitching staff early because the best values after Round 14 lies in filling in your bench with hitters.

Feb 14, 2010

Using High Draft Picks on Fantasy Starting Pitchers, A Study (Part 2)

We took a look at some data constructed to see how predictable starting pitcher's were last year and we showed how many of those starting pitchers did not pitch over 150 innings despite being drafted highly.  Running a similar analysis on the top 50 drafted hitters from last year, yielded interesting results.  Firstly, here is the odd crop of first 50 drafted hitters from last season on average:

Ramirez, Hanley
Fielder, Prince
Gonzalez, Adrian
Pujols, Albert
Soriano, Alfonso
Kemp, Matt
Reyes, Jose
Morneau, Justin
Rios, Alexis
Wright, David
Pedroia, Dustin
Guerrero, Vladimir
Sizemore, Grady
Ramirez, Manny
McCann, Brian
Cabrera, Miguel
Upton, B.J.
Martin, Russell
Braun, Ryan
Rodriguez, Alex
Ortiz, David
Hamilton, Josh
Lee, Carlos
Granderson, Curtis
Howard, Ryan
Quentin, Carlos
Ramirez, Alexei
Rollins, Jimmy
Suzuki, Ichiro
Mauer, Joe
Teixeira, Mark
Crawford, Carl
Ellsbury, Jacoby
Utley, Chase
Phillips, Brandon
Victorino, Shane
Holliday, Matt
Ramirez, Aramis
McLouth, Nate
Kinsler, Ian
Bay, Jason
Ordonez, Magglio
Longoria, Evan
Markakis, Nick
Soto, Geovany
Berkman, Lance
Roberts, Brian
Hart, Corey
Beltran, Carlos
Youkilis, Kevin

Keeping in the same spirit as we did for pitchers, I wanted to stick to ratios as best as I could to account for variance in at-bats skewing cumulative stats.  The only traditional measures we could use were Batting Average and AB/HR and then we had to get creative.  So, though it's not a terribly practical statistic, we calculated R/AB and RBI/AB for this experiment.  For stolen bases, it didn't seem fair to just did SB/AB so we broke it down by SB per Base Appearance (being a walk, single, double or triple).  So, we took these five metrics we had and calculated them for Marcel and Chone's proejctions for these top 50 drafted hitters from last season and compared that to their actual results.

On average for those 50 hitters, here's the percent difference between the projections and results:

Chone 5.64 -2.13 -7.97 -2.66 3.01
Marcel 3.86 -2.61 -5.53 -1.64 -6.43

If we look back at starting pitchers in the top 35 and the average of their percent difference, we'd see:


For the most part (on average), starting pitching was more predictable.  Of course, I recognize that average is not the most scientific indicator here but it paints a picture for you.  But, perhaps a nicer representation would be to show how many of the top 50 hitters performed over 10% worse than projected, making them unreliable, in these particular metrics.

Chone 20/50 11/50 20/50 19/20 15/20
Marcel 20/50 11/50 16/50 16/50 21/50

And, when we ran that calculation for starters who did 10% worse than projected in their areas, we got these results:

Chone 11/35 6/35 7/35
Marcel 15/35 6/35 7/35

So, really, the starters actually were mostly mostly better off...  Hmm.

The other reason we are told to stay away from starting pitching early on is because of the high risk of injury which caused 11 out of 35 of the top drafted starters to pitch under 150 innings with 3 starters being rendered nearly useless.  With hitters, the story was slightly different but not much.  There were 11 out of 50 hitters who missed time and were forced under 450 AB's.  However, if we tweaked it up a bit and looked at all hitters under 500 AB's, that number would become 16 out of 50 which is almost the exact same percentage as the 11 out of 35 starters. One main difference is that there was only one hitter who was rendered nearly useless on the season (Jose Reyes).

Now, this is only a small study and not using multiple years to get to our conclusion but, regardless, our research seems to point to an idea that starting pitching might not be as volatile as many think.  Perhaps using those early picks on starters isn't as much of a fool's errand as I was lead to believe in the past.  Food for thought.

Feb 13, 2010

Using High Draft Picks on Fantasy Starting Pitchers, A Study (Part 1)

One of fantasy baseball's hardest decisions is determining when to draft a starting pitcher. As a disclaimer, I have usually subscribed to the theory of waiting until the 8th round or later until grabbing your first SP but I visit this idea with an open mind.  If we look at projected roto values of starting pitchers, you could say that it makes sense to grab one of the elite starting pitchers before the projections dip off.  The argument against that is that starting pitching is volatile at best and carries a lot of injury risk and they rarely live up to their projections.  We'll investigate this claim... NOW.

A few notes about what we're doing here: I think it is unfair to evaluate projection systems on whether they correctly guessed an exact number of strikeouts or wins.  Those are both highly dependent on predicting an exact number of innings pitched and I don't think it's a projection system's job to do so by gauging a player's health; that is the drafter's job.  So, I think a better judge of how closely the projections matched the final result is to compare ratios.  So, we'll look at WHIP, ERA and K/9 that were projected by Marcel and Chone last season for the top 35 drafted starting pitchers and we'll compare that to the end-of-year results to determine if they were accurate in their assessment of a pitcher's skill then we'll take a glance at health afterwards.  First, Marcel:

Santana, Johan 3.24 2.86 -9.15
Lincecum, Tim -26.11 -17.92 6.50
Sabathia, C.C. 9.28 -4.18 -9.62
Webb, Brandon

Peavy, Jake 9.34 -9.90 7.84
Hamels, Cole 24.56 8.71 -6.10
Halladay, Roy -17.84 -7.22 17.97
Haren, Danny -9.64 -21.60 7.75
Beckett, Josh -1.90 -6.46 2.94
Lee, Cliff -13.87 -5.57 7.84
Liriano, Francisco 46.30 15.57 -1.08
Oswalt, Roy 17.58 -6.17 -1.54
Shields, James 8.31 5.98 -3.65
Lackey, John 7.11 -3.64 -1.30
Kazmir, Scott 33.22 5.98 -29.58
Billingsley, Chad 17.49 -5.88 -4.89
Santana, Ervin 20.16 11.94 -13.91
Matsuzaka, Daisuke 42.74 29.00 0.90
Lester, Jon -6.29 -10.65 37.86
Burnett, A.J. 1.13 1.85 -2.71
Volquez, Edinson 17.95 -7.15 -1.79
Harden, Rich 34.45 9.06 10.70
Zambrano, Carlos -1.72 0.78 11.38
Chamberlain, Joba 42.95 17.35 -20.18
Gallardo, Yovani 5.46 -1.45 21.41
Wainwright, Adam -29.31 -10.80 21.49
Nolasco, Ricky 26.78 -3.02 23.27
Cain, Matt -24.96 -13.38 -10.56
Verlander, Justin -17.87 -17.29 30.62
Vazquez, Javier -41.10 -25.27 17.14
Greinke, Zack -51.45 -21.29 19.40
Price, David 11.91 -2.53 3.30
Dempster, Ryan 0.36 -4.23 -0.35
Young, Chris 35.05 12.36 -32.59
Myers, Brett 12.15 -2.26 -24.51

5.45 -2.54 2.20

While I generally have no qualms about including an injured player in this comparison, Brandon Webb represents a unique case as he wasn't even able to produce any notable sample size based on his four innings of work in 2009 so for that reason his percentages are omitted. 

And, we'll let Chone step up to the plate before analyzing...

Santana, Johan -6.24 2.69 -9.33
Lincecum, Tim -25.79 -18.67 1.93
Sabathia, C.C. -1.25 -4.96 -2.34
Webb, Brandon

Peavy, Jake 4.29 -8.57 8.98
Hamels, Cole 23.34 10.58 -6.45
Halladay, Roy -24.14 -11.15 15.99
Haren, Danny -17.02 -22.32 3.83
Beckett, Josh 5.29 -4.35 3.00
Lee, Cliff -15.97 -2.88 6.12
Liriano, Francisco 37.50 12.32 1.19
Oswalt, Roy 4.38 -7.77 1.14
Shields, James 11.43 8.25 -2.51
Lackey, John -0.89 -3.50 -0.64
Kazmir, Scott 31.00 8.97 -31.11
Billingsley, Chad 8.82 -3.00 -6.07
Santana, Ervin 28.65 14.83 -13.81
Matsuzaka, Daisuke 38.82 27.76 -2.63
Lester, Jon -24.34 -17.84 39.47
Burnett, A.J. 4.03 1.20 -4.16
Volquez, Edinson 13.85 -5.74 -9.17
Harden, Rich 25.67 9.10 -3.45
Zambrano, Carlos -9.24 -4.48 14.25
Chamberlain, Joba 33.53 17.89 -26.95
Gallardo, Yovani 6.36 1.77 5.76
Wainwright, Adam -35.73 -8.28 28.13
Nolasco, Ricky 26.50 -1.59 25.28
Cain, Matt -24.96 -13.38 -11.16
Verlander, Justin -14.50 -17.44 32.10
Vazquez, Javier -12.73 -14.31 6.16
Greinke, Zack -64.24 -24.23 19.42
Price, David -10.27 -13.43 6.25
Dempster, Ryan -11.10 -9.64 3.01
Young, Chris 27.58 6.43 -23.53
Myers, Brett 19.65 3.86 -19.92
1.54 -2.70 1.43

It's a similar story.  Though they're scattered about on the individual projections, the sum of their parts isn't too far off.  If these results would be considered reliable to come within 10% of the projections, the top of the crop all do fairly well.  Cole Hamels is the one outlier but his ERA and WHIP were both victims of an odd BABIP spike.  Other than that, in the Top 10, the only players that were out of the 10% range were there because they performed over 10% better than projected.  If you're already using a top pick a starting pitcher, there's certainly nothing wrong with them outperforming already high projections.

Using 10% as our ideal range for checking the reliability of these results, let's take a look and see how many of these starting pitchers' ratios were greater than 10% worse than expected to see exactly how many of these were did not live up to their predictions of skill:

# Over 10% Worse Than Projected
# Over 10% Worse Than Projected

Granted, most of these pitchers who underperformed did so because of significant injury and that is a large reason to avoid wasting a high pick on pitchers.  Out of those top 35, there were 11 who pitched under 150 innings last season (Brandon Webb, Jake Peavy, Francisco Liriano, Scott Kazmir, Ervin Santana, Daisuke Matsuzaka, Edinson Volquez, Rich Harden, David Price, Chris Young, Brett Myers).

On their own, these statistics are meaningless.  But, in our next installment in the series, we'll compare these results to the top 50 drafted hitters from last year's drafts and see if they were truly more reliable and less injury-prone.  Stay tuned, loyal readers.