Welcome back, Jon Lieber. I remember you well. A lot of innings, and even a couple Cy Young votes. Getting you for Brant Brown - that was awesome.
Now he's back, one year, $3.5 million plus $4 million in incentives. This says a few things, some or all of which may be true. Or none.
- Marquis is on his way out - with the Cubs picking up most of his contract
- Dempster is back to the bullpen, but not as closer
- Marshall to Baltimore
- Gallagher to Baltimore
- No one seriously considered Kevin Hart for a spot
Who knows, the next move will be the most telling.
For now, I'm left wondering what the guy throws. A lot of ground ball outs, if I remember correctly....
I ran a couple passes thru the one and only one game of PITCHf/x data we have for Lieber. y=55, n=104
First, eyeballing it....
Ummm, OK. Not sure what I see here - perhaps 3 pitches. Perhaps 10.
Tried another old friend, k-means cluster analysis, nothing satisfying there.
Using an even older friend, I went to my book shelf and pulled out The Neyer/James Guide to Pitchers and found this, on page 280, right between Ted Lewis and Kerry Ligtenberg.
1. Sinking Fastball 2. Slider 3. Change
They cite The Scouting Notebook 2000.
Checked one new friend, the PITCHf/x tool by Josh Kalk - his technique finds two pitches. Scroll down at the link to see the table with the pitches. For your immediate viewing pleasure, here's the pitch location chart, also available at the link.
Once the season gets going, we'll revisit Lieber and see what his 2008 stuff looks like to a computer and a bunch of cameras.
2 comments:
Harry,
You don't show speed on those graphs, and I don't have my own data handy at the moment, but just eyeballing it, I see fastball and slider. What I see as fastball could well include changeup that can be seen as a separate group when speed is considered.
Josh Kalk's tool appears to be missing the slider entirely, but it's pretty clearly there to me. It's the group with pfx_x > -0.5 and pfx_z < 9, or alternatively, the group with spin rate < 1000 and spin direction < 190 or so. The other group is fastball/changeup, which should be easily differentiated by speed.
As a side note, have you looked at Alan Nathan's new equations for pfx_x and pfx_z? He's got an adjustment for drag that makes a ton of sense and affects the values by roughly an inch in each direction. It's not a huge deal, but it does improve accuracy somewhat. I assume Sportvision/MLBAM will be incorporating Alan's change in next year's data.
Mike - Thanks for the comment. I agree with your read of the data, but the noise makes me think the axes are not accurate - perhaps the pfx adjustments will show some tightening of those clusters/correction.
Regarding speed, I do have that, but I'll hold off until I check out Alan's work.
Post a Comment