It won't be long now. The return of ESPN's fantasy baseball Daily Notes is right around the corner. A major feature is a ranking of the slate's starting pitchers using projected Game Score. While Bill James' Game Score does a solid job capturing the elements relevant to fantasy baseball, it's impossible for everything to be distilled down to a single number.
As such, each day a handful of pitchers likely to be available in standard ESPN leagues will be highlighted, with supporting analysis. While the rotating Daily Notes crew each have their own favorite evaluation methods, there's significant crossover in terms of terms and processes. Today, many of the common threads used to identify pitchers in a favorable spot will be discussed.
Glossary of terms
The following terms are often incorporated into the pitcher analysis in the Daily Notes.
Weighted on base average (wOBA): Features the components of on-base percent but with coefficients for each term. The result is a gauge for team run production. Since 2011, the league-average wOBA is .320, matching last season's mark. Note, wOBA is not park-corrected. The lower the team wOBA a pitcher faces, the better chance he has at limiting runs.
Weighted runs created plus (wRC+): Expressed as an index with 100 being neutral, wRC+ is a normalized version of wOBA, accounting for park factors and era. The lower the team wRC+ a pitcher faces, the better chance he has at limiting runs.
Park index: Also called park factor, it measures the extent the venue benefits or hinders specific events. The formula fleshes out team bias, comparing a team's performance at home and on the road. Since they're variable from year to year, a three-year average is conventionally used. The most useful indices for evaluating pitchers are hits, homers, strikeouts, walks and runs. A factor of 100 is neutral. A venue aiding the metric by 10 percent is 110, while 90 suppresses by 10 percent. Rough guidelines:
Over 115: Extremely detrimental
106-115: Moderately detrimental
96-105: Neutral
86-95: Moderately favorable
Below 86: Extremely favorable
A negative index for hits, homers, walks and runs favors pitchers. A positive factor for strikeouts benefits pitchers.
Strikeout percent (K%): Strikeouts/Plate Appearances. 2018 league average: 22.3 percent. Excellent is above 27 percent. Troublesome is below 17 percent.
Walk percent (BB%): Walks/Plate Appearances. 2018 league average: 8.5 percent.
Batting average on balls in play (BABIP): (Hits - HR) / (At bats - HR - strikeouts + sacrifice flies). 2018 league average: .293.
Left on base percent (LOB%): LOB% = (Hits + Walks + Hit by Pitch - Runs) / (Hits + Walks + Hit by Pitch - (1.4 x Home Runs)). 2018 league average: 72.8 percent. Elite pitchers approach 78 percent.
Home run per fly ball (HR/FB): Influenced by park factors, the 2018 league average was 12.7 percent.
Expected ERA: A pitcher's ERA based solely on skills. Examples are FIP, xFIP and SIERA.
Swinging strike rate (SwStr%): Pitches resulting in a swing and miss / pitches thrown. SwStr% correlated to K%. 2018 league average: 10.7 percent Dominating starters are above 12 percent while risky starters are below 9 percent.
O-Swing%: Commonly called chase rate, percentage of swings on pitches outside the zone. 2018 league average: 30.9 percent.
Factors for evaluating pitchers
Quality of opposing offense
Much like a batter facing a weak pitcher is the best measure of likely success, the run-scoring potential of the opposing offense is the single greatest tool when identifying pitchers to spot-start into your lineup.
The two primary measures are wOBA and K%, with wOBA paramount in rotisserie leagues and K% the chief concern in points leagues. That said, K% is also relevant in rotisserie whereas wOBA matters in points leagues.
Teams exhibit different levels of both stats facing righty and lefty pitching. The problem is the smaller the sample, the higher the variance -- so numbers versus southpaws, especially early in the season -- are quite suspect. Still, looking at the platoon splits, especially once lineups are announced, can provide an edge.
An area requiring more study is how a team's recent wOBA and K% influences that day's expectation. Individual player streaks are largely thought to be nonpredictive, so by extension, a team streak should also be nonpredictive. That is, if a team averages 3.8 runs a game but has averaged 5.6 tallies the past five games, the baseline expectation should still be 3.8 runs. This is more apropos when picking on a perceived struggling offense than avoiding what appears to be a hot one. If a team usually scores 4.6 runs a game, it's a risk deploying a pitcher against it if the team has averaged just 3.2 for a few games.
Another area in need of research is how pitcher's strikeout rate is influenced by the opposing team's strikeout rate. That is, how beneficial is it for a hurler with a low strikeout rate to face a team that fans at an accelerated clip and how detrimental is it for a dominant arm to square off with a club stingy with the strikeouts? Hmm, this sounds like the topic for an upcoming column.
While on the topic of strikeout rate, there's a movement to eschew the traditional K/9 and use K%. While it's true K% is more indicative of a pitcher's skill, K/9 remains more relatable. When someone says a pitcher spots an 8.6 K/9, most have an immediate feel for what to expect. On the other hand, a 22.4 K% doesn't paint the same picture, even though the two stats can be from the same guy.
A practical example is two pitchers with a 9.0 K/9. The first retired the side in order with a whiff while the other allowed a hit and a walk. The K% for the former is 33% while it's 20% for the latter. Of course, this is an oversimplification, but it shows the difference.
Home-field advantage
There's an intrinsic skills advantage for pitchers working at home. This is data from last season, but it's representative of what occurs.
It may not appear significant, but the better skills make a difference. Keep in mind this is independent of park factors. Be it comfort with the mound, favorable umpiring or just the comfort of sleeping at home and not in a hotel, most pitchers enjoy home-field advantage so it's best to stream them for home starts.
Park factors and game conditions
Park factors are helpful, but the opposing team's production bakes them in. That is, be careful not to double-count a park factor. For an away game, the opposing team's home wOBA already bakes in the park factor. When the pitcher is at home, the opposing team's road wOBA should be tweaked based on the park.
Similarly, keep in mind park factors consider typical weather conditions. That is, one reason Globe Life Park is a hitters' park is the hot weather in Arlington favors offense. So, don't further penalize a pitcher on a hot Texas night since the factor already takes that into consideration.
On the other hand, extreme weather conditions can be used to gain an edge. When the temperature is lower than average, offense suffers, giving the edge to the pitcher. Wind blowing in or out at a higher-than-normal rate can also be leveraged into an advantage.
Regression candidates
It's unfortunate that regression has become synonymous with "play worse" instead of connoting the more statistical use of movement toward the mean. The following refers to the type of regression out of the pitcher's control. A pitcher encountering bad luck should revert to neutral. Similarly, a fortunate pitcher should also revert to neutral. This is an example of gambler's fallacy. Luck doesn't even out. If a flipped coin lands heads, there's a 50 percent chance the ensuing flip is also heads. Some may expect tails so the result of the two flips is one heads, and one tails. That's not how regression works.
When a pitcher is lucky, his outcomes are better than predicted by his skills and vice versa. Part of identifying streaming candidates is looking beyond recent surface stats, focusing on the skills within the player's control.
Expected ERA
The simplest down-and-dirty test is comparing a pitcher's ERA to his FIP, xFIP and SIERA. The differences between the three are beyond the scope of this discussion. The key is pinpointing lucky pitchers you may want to avoid while finding unlucky ones you want to trust. This helps in unearthing options others are avoiding based on a poor ERA. Assuming the pitcher maintains current skill levels, his ERA should regress toward his expected ERA. While others are skipping over the guy with a 4.23 ERA, you note his xFIP is 3.56, suggesting he's been unlucky. In fact, not only is he a solid streaming option, but you might want to hold on to him since he's likely to be a 3.56 pitcher the rest of the season.
While it goes a little deeper, the following regression metrics essentially explain the discrepancy between actual and expected ERA.
BABIP
Given that ground ball pitchers usually carry a higher batting average on balls in play than fly ball pitchers, BABIP is largely out of a pitcher's control. There's growing evidence an individual can influence the quality of contact, but in general, a pitcher's BABIP should range from .275 to .310. Anything lower can expect to regress upward, while a higher mark is likely to fall.
LOB%
Left-on-base percent is mostly a reflection of timing with some bullpen efficiency factored in. Two pitchers allow five hits (including a homer) with two walks in five frames. One gave up a three-run dinger, while the other allowed a solo shot. A cluster of hits is more likely to result in a run than the same number of scattered hits. How many are on base greatly influences run scoring. The same number of hits and walks can lead to different run totals. LOB% takes all these events and estimates the number of runs that are scored on average with those inputs. A low LOB% suggests more runs scored than usual in those conditions. If the pitcher repeats that performance, chances are he'll allow fewer runs. A high LOB% portends more runs the next time, assuming the same inputs.
HR/FB
There are three components contributing to home run rate: contact rate, fly ball rate and HR/FB. Of the three, HR/FB is the most out of a pitcher's control. If a pitcher has surrendered an inordinate number of homers in recent outings, look at his HR/FB. If it's above his usual level, expect it to regress.
SwStr%
Since strikeouts are a scoring category in almost all formats, looking at SwStr% is more than a means to explain expected ERA. If a pitcher is generating more swinging strikes than usual, but isn't fanning as many as normal, it's fair to expect more whiffs in future outings. On the other hand, if he's garnering more whiffs despite a drop in SwStr%, the strikeout spike isn't likely to continue.
Recent performance
Research demonstrates hitting streaks are nonpredictive. However, studies have revealed a pitcher with a string of recent solid performances has a better-than-50 percent chance of staying on a roll. Obviously, the streak will end, but this is about looking for an edge. If the odds are over 50 percent a hot pitcher stays hot, you can gain an advantage over the course of the long season. You won't be right every time, but in the aggregate, you're right more than wrong. This is helpful when scraping the bottom of the barrel, seeing a below-average pitcher has put together a few better outings. Others continue to avoid the lesser arm, but so long as expectations are reasonable, there's a good chance you can squeeze another useful effort from the guy.
Sustained changes from projected performance
Perhaps the biggest strength of the projected Game Score is avoiding recency bias. Formulaic treatments are good like that, but they're slow to recognize actual improvement. One of the keys to streaming pitching is being ahead of the algorithms, correctly elucidating options whose skills have taken a sustainable leap forward. The opposite is also true. Sometimes, water doesn't find its level. If a pitcher's skills have slipped, his usual number may not be there in the end, as we may presuppose.
There's no foolproof recipe. Sometimes, we're tricked into thinking a change is permanent, only to see a reversion to previous levels. As such, it's obligatory to have something tangible to use as evidence. Here are some examples.
Velocity
An increase of decrease in velocity is a great indicator. Higher velocity usually results in improved skills. Sure, the velocity can fall, but if you're looking for something to support improved performance, look at velocity.
Pitch mix
Many times, adding a pitch or eliminating one aids skills. Sometimes, mixing up distribution or sequencing has a positive influence on performance. For the longest time, this sort of thing was unknown, expect to those closely following a team on a day-to-day basis. In today's information era, these changes are no longer a secret as they're blasted all over social media.
General thoughts
There's a plethora of next-level analysis about pitching. However, unless you're gambling or a hard-core DFS player, it's deeper than needed to find viable pitching in standard fantasy leagues. If there's one additional factor to consider, it's umpires. While it's hard to determine who will be behind the plate to begin a series, the umpires rotate the rest of games with the first base ump calling balls and strikes the next game. Umpire tendencies are available and can be deployed to assist in streaming pitchers.
However, similar to hitters, the prevailing characteristic is a pitcher facing a weak lineup. All the rest is icing on the cake. To that end, making last-minute adjustments based on that day's lineup can pay off handsomely. It requires more diligence and the available time, but if a top hitter isn't active, the matchup can easily flip from one to avoid to one to use to your advantage.