As we head into the second week of the season, 2018 sample sizes will continue to grow and fantasy owners will continue to analyze player performance to determine whose performance–good or bad–has staying power. Below I’ve shared some skills I look for and tools I use when analyzing players during the season to help identify which players to target and which ones to avoid.
Contact & Plate Skills
Any player can do nearly anything in a small enough sample size, so it’s important to look at underlying skills in addition to outcomes. Some key plate skills I look for in hitters are improvements in contact rate and o-swing % (the percentage of pitches outside the zone a player swings at) and velocity, swinging strike rate, contact rate and o-swing % for pitchers. These skills–along with others–help show whether a player’s performance is supported by changes in their underlying skills or approach.
Early in the season, I also like to look back at the end of last year to see if the improvement (or erosion) in skills and performance aligns with late season trends (i.e. the adjustment already happened and the early season performance is an extension of it).
Justin Smoak made tremendous strides with his contact rate in 2017, but the improvement actually started in late 2016. His results late that year were horrendous, but he was able to continue improving his contact skills (and plate discipline) and turn then into a breakout year in 2017. There is no perfect way to do early season analysis (or there probably is and I just don’t know it), but I find it helpful to look for trends in player skills and not just performance. For instance, if you had looked at Smoak’s 40-game rolling average for contact in late April last year (right before he broke out), you would’ve seen it at 85.3%, well above any previous point in the past three years or his entire career.
Rolling average graphs on Fangraphs player pages are a great tool for analyzing skills and they’re easy to create. Simply go to the player page, click on “graphs” in the top left, select “by game,” set the rolling average and select from various options for hitters and pitchers at the top. I prefer 80-game rolling averages for hitters, 10-game rolling averages for starters and 20-game rolling averages for relievers, but you’ll want to look at smaller intervals, too. Larger rolling averages can mask more recent changes, though the smaller samples also make the data less reliable.
Batted Ball Profile
Early season signs for batted ball profile can also be a bit messy, since one good game with four hard hit balls (or just facing the Marlins rotation) can really skew actual performance. Still, it’s a valuable exercise.
In particular, I look for changes in fly ball rate, which may signify an increase in power potential (or a drop in BABIP and batting average), or ground ball rate, which may signify the opposite. If you prefer more precision, you can also look at launch angle. Drastic changes in batted ball profile can also indicate an unreported injury (for instance, a big jump in ground ball rate).
I also look at hard hit rate and hard hit fly ball rate (you can also look at exit velocity, though try not to just look at overall average exit velocity but instead look at average exit velocity at key launch angles (i.e. 19-39° for fly balls) using MLB Savant’s Statcast advanced search).
Logan Morrison is a great example of someone who went undrafted in most leagues last year who benefited from a change in his batted ball profile–an increased fly ball rate. His hard hit fly ball rate also improved a little, but wasn’t special at 43.2%. However, the substantial increase in fly ball volume (11.5% more than 2016) resulted in a huge power spike (he also got a bit lucky, but let’s pretend he didn’t). Morrison began hitting more fly balls in the middle of 2016, and the trend accelerate early in 2017. If you had seen the spike early last season on top of late in 2016, it would’ve been a good indicator of a change in approach and potential breakout.
The same applies to pitchers. Look for changes to their batted ball profile or underlying skills. Has their velocity gone up or down a tick? Are they throwing a new sinker and their ground ball rate has gone up? (If so, goodbye strikeout rate! Just kidding…sort of.) Or better yet, are they throwing their curveball more, which has both increased their ground ball rate and their swinging strike rate? The key is connecting changes in batted ball profile and underlying skills to a different approach, like a new pitch mix or velocity increase.
Jimmy Nelson is a good example of a pitcher whose performance increased dramatically as last season progressed. One potential reason is that he began throwing his curveball more beginning in May and saw an increase in both his swinging strike and ground ball rates over a similar period of time.
Looking at his pitch splits on Fangraphs, you can see Nelson’s curveball has both a high ground ball rate (53.8% in 2017) and an above average swinging strike rate (14.3% in 2017). This may help explain some of the improvement.
Beginning with his May 11 start (the start when his curveball usage started increasing on the graph), Nelson had a 3.19 ERA (2.78 FIP), 1.22 WHIP and 174 strikeouts in 143.2 innings. Before that he had a 4.83 ERA (4.26 FIP), 1.39 WHIP and 25 strikeouts in a small sample of 31.2 innings in 2017 and had struggled throughout 2016.
Whether or not the curveball in isolation was the key is difficult to know, since so many variables go into how a pitcher performs. Over the same period of time, Nelson also decreased his fourseam usage, which had a high fly ball rate. That probably made a difference, too. The key is identifying what has changed and trying to piece together a rational narrative on what might have been the catalyst for improvement and whether it looks sustainable.
The Luck Factor
One of the greatest inventions of modern times is xStats.org, the brainchild of Andrew Perpetua who writes for Rotographs.com. xStats uses exit velocity and vertical and horizontal launch angle to determine the likely outcome of each batted ball (hit? home run? double? out?) and what each players expected stats should be. In essence, xStats allows you to see how lucky or unlucky a player has been and to use that data to factor into your assessment of whether their performance is likely to improve, stay the same or get worse.
Justin Smoak is another good example for using xStats to see if actual performance aligns with expected performance (to rule out as much luck as possible). Smoak started off OK in April, but turned it on in May hitting 8 home runs. If you look at Smoak’s xStats for May, it shows him hitting 7.6 xHR. Half a home run shy of his actual total, but it supports an increase in home runs. When he hit 10 home runs in June, his 9.9 xHR would’ve given fantasy owners more confidence the newfound power was real. Some good news: xStats.org will update daily this season, so you can track how a player’s performance lines up so far with their expected stats.
Corey Dickerson is an example of how to apply xStats for overperformers. Last April he hit .330 on a .359 BABIP and 6 home runs. According to xStats, his expected stats were a .292 xAVG, .301 xBABIP and 6.1 xHR. The expected stats supported the power but did not support the very high batting average and BABIP. Still, the performance was strong.
In May, Dickerson hit .349 (.289 xAVG) with a .418 BABIP (.326 xBABIP) and another 6 home runs (6.9 xHR). The underlying skills pointed to strong performance yet again, but also a lot of luck on balls in play. Midway through the season, he was an All-Star. But Judgment Day was on the horizon:
Corey Dickerson is a top 2H drop off candidate: .312 AVG (.275 xAVG) w/17HR (17.8 xHR) on .361 BABIP (.310 xBABIP) #fantasybaseball
— BatFlip Crazy (@batflipcrazy) July 10, 2017
The expected regression took place in the second half when he hit .241. He would’ve been a great sell-high candidate for fantasy owners in July.
Dickerson’s drop off also illustrates how you can use multiple tools to identify a player who may be overperforming. At the same time that Dickerson was outperforming his xStats, he also had one of the highest swing rates on pitches outside the zone. Pitchers had noticed, too–they stopped throwing in the zone to Dickerson (yellow line in the graph below). Combining these two data points made me suspicious of Dickerson’s ability to continue outperforming his expected batting average stats while also having very poor plate discipline.
Dickerson: O-swing% 2nd worst in #MLB (46.8%), contact OK, but pitchers adjusting (.265 AVG since zone% started falling) #fantasybaseball pic.twitter.com/h0DnF9Xeac
— BatFlip Crazy (@batflipcrazy) July 10, 2017
The prediction ended up being right this time (You didn’t think I was going to include an example where I got it wrong, did you?), but it’s no guarantee. The best you can do is make an informed decision based on available data.
All the Signs
Each skill and tool for analyzing players early in the season is imperfect. You won’t have the opportunity to track a player’s improved contact rate for months to make sure it isn’t just small sample variance. By that time, another owner in your league is enjoying the player’s breakout season.
The best way to get the clearest picture of whether a player has really improved is not to look at one skill in isolation. In the Justin Smoak example, a dramatic increase in contact rate could also mean he’s making more contact but it’s weak contact. Verifying that he hasn’t sacrificed hard contact for contact is one way to feel better about the improvement. Logan Morrison could hit more fly balls, but if only one out of 10 is hit hard, he’ll hit a lot of routine fly balls and his batting average will plummet. Jimmy Nelson could have a sweet ERA after throwing his curveball more, but he could also have a .250 BABIP when xStats tells you it should be .330.
Look at the skills, use the tools at your disposal and make an educated decision about which player performances you trust. You’ll definitely get some wrong, but hopefully you’ll get more right.