Week 9: What Are Analytics Good For?

Brian stumbled upon Max and Arthur eating at the Brass Ring Pub last week.

Year to Date Power Ranks through this week
TOTAL			HITTING			PITCHING				Manager	Team
1	Paul	3.50	1	Dean	2.00	1	Brian/Josh	2.17		Cory	Hebrew Nationals
2	Michael	3.57	2	Paul	3.50	2	Cory	3.33		Arthur	PURE DOMINATION
3	Dean	4.71	3	Michael	3.75	3	Michael	3.33		Dave	I Hate Fantasy
4	Cory	4.79	4	Arthur	5.25	4	Dave	3.50		Brian/Josh	Smoak That Ish
5	Dave	5.29	5	Matt	5.25	5	Paul	3.50		Paul	CWS champion Gators
6	Arthur	5.43	6	Cory	5.88	6	Arthur	5.67		Dean	First Future father
7	Brian/Josh	5.43	7	Keith	6.25	7	Keith	7.17		Matt	615 for the win
8	Keith	6.64	8	Dave	6.63	8	Max	7.67		Max	[some inside joke acronym]
9	Matt	6.79	9	Brian/Josh	7.88	9	Dean	8.33		Keith	Bourbon Street Blues
10	Max	7.93	10	Max	8.13	10	Matt	8.83		Michael	Future Father Too

Michael and Paul maintained their separation from the field but Dean and Cory took a step in the right direction. Max is bottoming out in a week where he was skunked by Cory. We’re at a bit of a crossroads right now in the league. Right now the first place team has a winning percentage of 67.9% and the last place team is a 32.6%. While I don’t have records of what the most extremes of these numbers have been through week 8 in league history, the highest end of season winning percentage is 58% when Paul set the league on fire in 2011 and the lowest ever league winning percentage is 33% by my old roommate Chris Michaelis in 2009 who made 5 moves all year and was not asked back to the league the next year. I bring up these numbers to illustrate how big the gap is this year between the haves and the have nots. Time tells us these numbers are likely to level out over the next 14 weeks, but it is something to watch. Also, I’m not saying we’re going to be relegating Max to the B league if his trend continues, but maybe he should read up on this blog in particular.

The blog is going to be a little bit different this week. Removed from the time constraints of a work lunch or pre-work activities, on this Memorial Day I’m going to wax freeform about Analytics. If this isn’t your bag, feel free to change the channel. But if that is how you feel,  this is probably directed at you, so I encourage you to read on.

I text most of you in the league on a regular basis about something cool that a player of yours is doing, the fantasy matchup, or just to shoot the breeze. Occasionally I hear back from you guys with cracks at a player I may have called out as lucky in the past who is performing well, insulting my team’s performance, or maybe just mad about fantasy.

Let’s flash back about 12 months ago to the blog about a few pitchers who had significant gaps between their Earned Run Average (ERA) and Fielding Indendent Pitching (FIP) and see how they fared the rest of the year. These were the pitchers specifically called out

Pitcher	ERA on 5/20/2017	FIP on 5/20/2017	ERA between 5/20/2017 and end of 2017
Mike Leake	2.02	4.80	4.68
Ervin Santana	1.62	4.20	3.77
Jason Vargas	2.03	4.70	4.95

As you can see, the FIP was a far great predictor of ERA moving forward than a player’s ERA up to that point. Remember FIP measures what a player’s ERA would look like over a given period of time if the pitcher were to have experienced league average results on balls in play and league average timing. So, what can we do with this information this year? Let’s look at the FIP minus ERA leaderboard to find out who is due for some regression.

The Fortunate (Notable Starting Pitchers, minimum 40 IP)
Pitcher	ERA 5/29/18	FIP 5/29/18
Carlos Martinez	1.62	3.35
Charlie Morton	2.04	3.33
Gio Gonzalez	2.10	3.07
Corey Kluber	2.17	3.37
John Lester	2.37	4.11 !!!!
Jake Arrieta	2.45	3.25
Patrck Corbin	2.47	3.03
Clayton Kershaw	2.86	3.68
Kyle Hendricks	3.16	4.28
Sean Manaea	3.34	4.04

I kept the guys off the list like Verlander and DeGrom who have ERAs below 2.00 who we know will come back to Earth, but their FIPs are only in the low 2s. They’re studs. Most notable here is of course Jon Lester. Arthur has been reaping the rewards of an overperforming Jon Lester for two months now, but the analytics say he is due for regression based on his balls in play, strand rate, and strikeout rate. Also notable here is Sir Clayton Kershaw, and this brings in an element to analytics that isn’t measureable, injuries. Kershaw pitched hurt for much of the first month of the season and his performance suffered. The question now is, which Kershaw will come back on Thursday? One other item to keep in mind is pitcher style. Guys like Gio Gonzalez, Kyle Hedricks, and Dallas Keuchel often out-pitch their FIP, this is due to their ability to command their pitches and generate soft contact. FIP minus ERA is not a be all, end all of pitcher regression identification, but it’s a pretty good place to start

The Unfortunate (Notable Starting Pitchers, minimum 40 IP)
Pitcher	ERA 5/29/18	FIP 5/29/18
Noah Syndergaard	3.06	2.56
Kenta Maeda	3.38	2.60
Nick Pivetta	3.74	2.91
Jon Gray	5.40	3.16
Alex Wood	3.75	3.20
Lance McCullers Jr.	3.98	3.53

There are fewer names on this list because it’s harder to underperform your FIP. This normally happens with very high K rate guys (see Syndergaard, Gray, and McCullers) that have suffered bad luck with their strand rate. Often times giving up untimely Home Runs with people on base or having your bullpen be unable to strand guys on based after exiting with men on. These pitchers would be the buy lows, because they’ve gotten unlucky and are due to for turnaround.

The hitting side of things is a bit more complicated, because there is no single stat available for a similar comparison to FIP. Luckily, we live in the Statcast Era and there are measurable numbers for just about everything. Statcast (with information avaialable at BaseballSavant.com) is able to measure how hard each player hits the ball, runs around the bases, runs to catch the ball in the outfield, and how hard each throw is. It’s awesome. Statcast has created ‘expected’ stats, bases on the exit velocity and launch angle of batted balls, to determine what the reasonable outcome of each Batted Ball Event is. Then, when the defense makes a great or terrible play, this can be accounted for.

For example, Jose Ramirez hits a ball at 15 degrees off the horizontal at 90 miles per hour with a certain spin rate, this would be a line drive that would travel approximately 350 feet with a certain amount of hang time. Statcast has the data to determine that this would be a double 90% of the time, but Mookie Betts is in there. He is super fast, made a great read off the bat, and took an excellent route to catch the ball and makes the out. Statcast gives Jose Ramirez a ‘barrel’ because it was a batted-ball event whose comparable hit types (in terms of exit velocity and launch angle) have led to a minimum .500 batting average and 1.500 slugging percentage since Statcast was implemented Major League wide in 2015. Barrels translate to positive outcomes for expected stats.

Lets look at expected weighted On Base Average (xwOBA), note that weighted On Base Average combines all the different aspects of hitting into one metric, weighting each of them in proportion to their actual run value. xwOBA attempts to predict what a players wOBA should be based on their Statcast data. BaseballSavant has leaderboards that track expected versus actual stats.

Here is the current leaderboard for wOBA minus xwOBA (minimum 150 Plate Appearances this year)

The Fortunate
Player	wOBA	xwOBA	Difference
Albert Almora Jr	.355	.288	.067
Mallex Smith	.336	.272	.064
Scooter Gennett	.397	.349	.048
Starling Marte	.362	.320	.042
Odubel Herrera	.390	.348	.042

The Unfortunate
Player	wOBA	xwOBA	Difference
Kole Calhoun	.178	.279	-.101
Jason Kipnis	.250	.346	-.096
Teoscar Hernandez	.334	.417	-.083
Matt Carpenter	.324	.404	-.080
Stephen Piscotty	.272	.349	-.077

If you know anything about these players, you can note quickly one asterisk to xWOBA, it punishes (meaning makes them look fortunate or lucky) fast players that are able to get infield hits or extra bases. The unfortunate list is full of players that hit the ball hard but don’t run very fast. Therefore these too must be taken with a grain of salt. The full leaderboard can be seen here: https://baseballsavant.mlb.com/expected_statistics?type=batter&year=2018&position=&team=&min=150. I encourage you to go and see where some of your players are: have they been lucky? Are they due to regress? Note that these searches/clicks to get to the leaderboards take less than 10 seconds, no excuses for this being too much work.

Last year I used this data to identify Hanley Ramirez as someone who was underperforming his wOBA, I acquired him in a trade from Max and he took off like a rocket…before he went all Hanley the last two months of the season. At the same time I could have used this information to note that Eric Thames was far overperforming his xwOBA. With my biased view (remember he was lighting the world on fire on my roster) I ignored this data and some good trade offers that came my way only to watch him fall off a clff the second half of the year. Although this data did convince me that Manny Machado was a buy low, and that one has worked out like gold.

Given that I called my shots above, I’ll make a few calls here. I’d buy low on Teoscar Hernandez (I literally did this a week ago before my team got all injured), Carlos Santana, Bryce Harper, Mike Moustakas, Jose Martinez. I’d be selling on Scooter Gennet, Odubel Herrera, Andrew Benintendi, Yoenis Cespedes.

Analytics are everywhere y’all. You can either embrace them or get beat by them.