This article is part of our The Saber's Edge series.
Last week, I looked at which pitching stats a fantasy owner should put more value into for the season's second half (2H), the previous season stats or first half (1H) numbers. Well, this week it is the hitters turn. For hitters, I am going to look at a few more stats than I did with pitchers. The hitting stats are not as predictive as the pitching ones, but some conclusions can be drawn. More importantly, the conclusions which should not to be drawn can be known. Like with the pitchers, walk and strikeout numbers are more predictive over time with power and stolen base numbers also showing some consistency. Batted ball values on the other are widely inconsistent and few conclusions should be drawn from them.
For this work, I wanted to focus primarily on stats used in fantasy and few other related ones. I didn't look at Runs or RBI. I feel they don't show the hitter's personal talent level and are determined by his teammate's production. In the end, I looked at walk (BB%) and strikeout (K%) rates, home runs (HR), stolen bases (SB), batting average (AVG), on-base percentage (OBP), slugging percentage (SLG), isolated power (ISO) and batting average for balls in play (BABIP).
One issue I ran into is to how to turn counting stats (home runs and stolen bases) into a rate stat. I just could not use the counting values because the first half values would be closer to the second half numbers because
Last week, I looked at which pitching stats a fantasy owner should put more value into for the season's second half (2H), the previous season stats or first half (1H) numbers. Well, this week it is the hitters turn. For hitters, I am going to look at a few more stats than I did with pitchers. The hitting stats are not as predictive as the pitching ones, but some conclusions can be drawn. More importantly, the conclusions which should not to be drawn can be known. Like with the pitchers, walk and strikeout numbers are more predictive over time with power and stolen base numbers also showing some consistency. Batted ball values on the other are widely inconsistent and few conclusions should be drawn from them.
For this work, I wanted to focus primarily on stats used in fantasy and few other related ones. I didn't look at Runs or RBI. I feel they don't show the hitter's personal talent level and are determined by his teammate's production. In the end, I looked at walk (BB%) and strikeout (K%) rates, home runs (HR), stolen bases (SB), batting average (AVG), on-base percentage (OBP), slugging percentage (SLG), isolated power (ISO) and batting average for balls in play (BABIP).
One issue I ran into is to how to turn counting stats (home runs and stolen bases) into a rate stat. I just could not use the counting values because the first half values would be closer to the second half numbers because the half totals are lower than full season totals. For home runs, I used home runs per plate appearance (HR/PA) and for stolen bases, it was stolen bases per time on first base (SB/(1B+HBP+BB)).
For each of the values, I looked at the R-squared between the values for the previous season and first half from 2002 to 2014. Additionally, I got an R-squared using both values to predict the 1H values. Finally, when using both values, I included how much each value should be weighted in determining the second half value.
Here is some thoughts on each value.
- Plate discipline (K% and BB%): These two stats stay consistent over time. A 50/50 weight could nearly be used, but the first half numbers do get a little more weight.
- Total on-base skills (OBP): The numbers for on-base percentage make perfect sense. They are halfway between the walk and batted ball values. The previous season values do carry more weight than the first half numbers. For any change in a player's OBP value, I would concentrate on the BB% component for a verification of a change instead of AVG.
- Batting Average (AVG, BABIP): The batting average values correlates a little better than the BABIP values (.21 vs .13) because of the strikeout component in AVG. The r-squared values when both values are used are by far the lowest of any of the stats I examined. The next highest r-squared value from the combined time frames is SLG at .33 and this low value is because AVG is a major component of SLG. Truthfully, I would stay away from using either value as a predictor for future values. I would instead get as many years' worth of data as possible and regress the value heavily to the league average. If a person really feels they need to quote either one of the two datasets, they should go back and use the previous season's data.
- Hitting the ball with authority (SLG): Slugging percentage is AVG+ISO. Like OBP which is a combination of BB% and AVG, SLG takes the middle ground between ISO and AVG. It weights the previous season more, but not as much as AVG. For any changes in its value, I would concentrate on the ISO component for a verification of a change instead of AVG.
- Power (ISO, HR/PA): The power number are next in line after the plate discipline values for being the most predictive. If you look at the weighting when both are used, it is a 50/50 split for the two previous time frames.
- Stolen bases (SB/Times on First): Stolen bases don't fit into the flow of the other stats, so I stuck them here at the end. They do have some of the best correlation, with the current season getting a decent amount of weight.
Conclusion: Plate discipline, power and stolen base stats have a predictive power with the recent half season possibly getting more weight. Once a stat begins to include batted balls/singles, the predictive nature drops significantly with the previous season getting more weight.
We can take the preceding information and look at a few players who are seeing the largest changes in their AVG from 2014 to the first half of 2015. For the players, I used a minimum of 400 PA for 2014 and 200 PA for the 1H of 2015.
Mike Moustakas (2014: .212 AVG, 1H 2015: .297 AVG): Moustakas may end up being the poster child for a player changing his approach to beat the shift. In his first two seasons, he pulled the ball ~40% of the time and posted a near .250 AVG. Then, Moustakas began to pull the ball at a 50% clip which happened at the same time when almost every team began to implement infield shifts. His average dropped to .233 in 2013 and to .212 in 2014. He made some offseason adjustments after the 2014 season and is now only pulling the ball 36% of the time.
Besides the change in the way he sprays the ball around the field, Moustakas has cut his walk rate to a career low of 11%. While Moustakas may not be a true .300 hitter, his change in AVG does have some credence behind it.
Danny Santana (2014: .318 AVG, 1H 2015: .225 AVG): Santana's 2014 AVG should be a warning for owners to not believe in players with a .400+ BABIP. Going into 2014, Santana's best minor BABIP value was .353 and a career value near .320. A .300 to .340 BABIP value should have been expect from him. Another issue with Santana is a near 25% K% in the minors. Most hitters with AVG over .300 don't even have anything close to 25% K%. His AVG screamed regression. In 2015, Santana is on the extreme sides of his projections with a .302 BABIP and 26% K% which has led to a .225 AVG. The .225 1H AVG is closer to his true talent level than the .318 AVG from 2015 and could see a move up as the season goes on.
Jason Kipnis (2014: .240, 1H 2015: .323 AVG): Kipnis's season did not completely come out of nowhere, but compared to his 2014 season, it seems extreme. Before 2014, Kipnis had a career .270 AVG with a ~20% K%. His 2014 AVG of .240 was quite a disappointment, especially with only 6 HRs. Part of the decline was because he dealt with injuries throughout the 2014 season.
He has really turn it around 2015, with a .323 1H AVG which he has pushed even higher in the second half. A career high .369 BABIP fueled most the increase which should regress some in the 2H. Besides the high BABIP, his K% has dropped from 18% to 15%. Kipnis could see his AVG drop quite a bit over the rest-of-the-season.
