Guarantee Fairy: Deep League Options

I’ve stolen from the movie before. I’ll do so again…

Guarantee? If you want me to take a dump in a box and mark it guaranteed, I will. I got spare time. But for now, for your fantasy teams’ sake, for your daughter’s sake, ya might wanna think about listening to quality content from me.

If you don’t know where this reference is from, then well…just ring your call button, and Tommy will come back there and hit you over the head with a tack hammer.

I actually will play guarantee fairy here, specifically for deep leagues since there are no uber-exciting names that jump out in my below grid. So here goes…

So long as they pitch to a qualifying level of innings without getting hurt or losing velocity (not ballsy enough to leave out these contingencies), I GUARANTEE these starters won’t be any worse next year (although in the grid below I highlighted in different strengths of green/red both starters and relievers):

Allen Webster, Erasmo Ramirez, Carlos Martinez, Dan Straily, Andre Rienzo, Tim Lincecum albeit as a reliever, Yohan Flande, Jhoulys Chacin, Randall Delgado, Robbie Ross, Trevor Cahill and Samuel Deduno.

I can add Edwin Jackson to the list based on my approach, but history tells me not to.

What an utterly random bunch you might say. I can only agree. I was hoping more attractive names popped out, but they didn’t.  I’ll get to the other side of the spectrum later on. With such a unique bunch, how can I guarantee it? Onto the approach:

First, I wanted to look at a combination of swinging-strike (SwStr)% and general contact (Ct)% to see what pitchers don’t have their strikeout rates matching up. I z-scored SwStr% and Ct% and then took the average z-score between the two. You can find that in the 9^th column below, titled “avgzCt.” The grid below is sorted by this column.

• If you keep the grid sorted by this column, the bold borders will make sense to you. For example, Aroldis Chapman is at the top in a tier all by himself because he leads the majors in SwStr% and Ct% (4+ standard deviations from the mean). Koji Uehara, Joaquin Benoit, Brett Cecil (surprisingly!) and Craig Kimbrell fill the next tier (between 2 and 3.16 standard deviations from the mean). And so on.
• For each of these tiers, I took the average K%. Again, for example, Uehara’s tier’s mean K% is 33.4%.
• Similar to the first few steps for finding RMSE, I then subtracted each player’s K% by the mean K% in their tier. Using a better example lower down the spectrum, Randall Delgado’s actual K% is 24.3%, but the mean K% in his tier (based on their swinging-strike and contact rates) is 26.3%. Therefore he has a -2.0% differential, which I’ve highlighted positively in green i.e. there’s a chance his K% jumps a bit next year if the swinging-strike/contact rates hold true.

Obviously, the assumption with the above approach is that K% and SwStr/Ct-rates tend to match-up, which they do. Feel free to run your own correlations.

Second, I wanted to back up these guys even further, so if you scroll all the way to the right, I used Mike Podhorzer’s expected K% and expected BB% formulas. Note: I only filled in the blanks where guys should jump in value based on their contact rates and expected ERA’s. I used BERA as the expected ERA in this case. You will find their BERA in column 11 and the ERA-BERA differential in column 12 (ERAdiff) and the z-score for the differential (zERAdiff) in column 13.

In summary, the guys I listed above (Carlos Martinez, Randall Delgado, Erasmo Ramirez, etc.) are highlighted in green in all three of these categories:

1. K% not quite up to par with their combined swinging-strike/contact rates
2. Have a positive ERA differential
3. Should have had a better K% (and sometimes BB%) according to Pod’s expected command formulas.

On the other hand, and don’t get upset, here are some starters likely to regress somewhat:

Chris Sale, Corey Kluber, Felix Hernandez, Alex Wood, Jon Lester, Jordan Zimmermann, Johnny Cueto, Mike Fiers, James Paxton, Danny Duffy, Jon Niese and Vance Worley.

It might seem like this approach is picking on studs, and based on their K% vs. SwStr/Ct%, I have to say that it is. Are Sale, Kluber and the King going to remain elite? Obviously. Will Wood, Lester, Zimmermann and Cueto be excellent? Naturally. Do I like Paxton and his glorious curveball? Most def. It’s just that their K% rates don’t match up well enough to their swinging-strike/contact rates; BERA says they’ll regress; and Pod’s formula says they’ll strike guys out a bit less.

In a single player’s terms: James Paxton certainly has the potential to induce more swings and misses which would negate the expected K% drop; his BERA of 3.64 is still solid; and xK% only has him dropping off 1.8%.

Said generally: This post is not associating value. It is pointing out possible jumps/regressions.

Instructions: Use the grid at your own discretion:

1. See whose K% could jump based on their swinging-strike/contact rates. You can see the biggest differences according to their “contact” tiers by sorting (by ascending) column 5, “K%diff.”
2. Again, the grid is currently sorted by column 9, “avgzCt” i.e. the combined swinging-strike%/contact% z-score. Each tier (Aroldis = tier 1; Uehara, Benoit, Cecil & Kimbrel = tier 2, etc.) is broken up by a half standard deviation from the mean.
3. You can see who has the biggest ERA differential (ERA-BERA) by sorting column 12 (actual difference) or 13 (difference via z-score). If you sort by descending, Ernesto Frieri will be atop (no surprise based on his 7.34 ERA/4.32 BERA).
4. Lastly, don’t sort, but check out column 16 (xK% differential) and 18 (xBB% differential) for the starting pitchers with ERA differentials and K%-MeanK% differences larger than .25 standard deviations from the mean on both sides of the spectrum.

Keep in mind that there will be many guys that aren’t highlighted in xK% and xBB% that will still jump/regress. For example:

Last year, G Richards had velo, strong SL & CB, & good MiLB BB9. Jacob Turner at 93mph, 2.5 MilB BB9, strong SL (18% swSTR) & good CB (12%)

— Eno Sarris (@enosarris) September 19, 2014

Jacob Turner as Eno Sarris just pointed out is an obvious candidate based on his pitch effect and 1.88 ERA differential which is 2.34 deviations from the mean ERA differential. I simply only made a guarantee when all three things aligned (contact rates; xK%; and ERA differential jumps or drops).

Finally, the baffling grid: