Some use a spreadsheet, some are more abstract. But every fantasy baseball enthusiast forms expectations (some take issue with calling them projections). It really doesn’t matter what they’re called or how they’re derived, one of the primary elements of this great hobby is player evaluation translated to an expectation.
Truth be told, just like professional scouts need to combine what the numbers say and what their eyes see, fantasy players do, too. Relying solely on a spreadsheet with all sorts of statistics and algorithms doesn’t tell the whole story. But neither does watching players in small doses, be it live or via television or video.
However, numbers are the heartbeat of player evaluation. Observing the player, especially in-season, is a great means to help explain a quirk or inconsistency with the numbers. But conjuring an initial expectation is driven by past performance, which is best captured by numbers. What follows are some tenets for player evaluation for draft preparation and in-season roster management.
GENERAL
1. Temper expectations – Playing baseball is hard. Repeating the previous season’s success is even harder. While it depends on exactly how the data is parsed, between 55 and 70 percent of players produce worse numbers than the previous season. We all have our guys—players we feel will have a great year. Just keep in mind you’ve got gravity (and probability) working against you. This is especially true for players with a limited track record, as expectations can get carried away when owning the shiny new toy.
2. Skills are only part of the story – Rolled into your expectation should be injury risk (fewer at bats or innings pitched) and role. For batters, the closer to the top of the order, the more chances received and the greater the number of opportunities to produce runs. Speedsters hitting eighth in the National League are likely to lose stolen base opportunities, since managers don’t want the pitcher leading off if the runner is caught. With pitchers, it’s crucial to consider the number of expected innings and not focus on ERA and WHIP. More innings mean more potential wins and strikeouts and a greater impact on team ratios. Home venue and quality of the lineup are factors as well.
3. Beware of small sample anomalies – While some players may perform better early or late in the season, first and second half splits are not predictive. The same goes for contract years and batting order protection. It’s best to look at the whole body of work, unless there is a tangible reason (like an injury) to serve as a before and after time-stamp.
DRAFT PREPARATION
1. Establishing a baseline – Most objective projection systems set a foundation using a weighted average, incorporating at least three and sometimes five years of data with the most recent factoring in the heaviest. Tweaks are made according to age and possibly injuries. Players with fewer than the requisite number of Major League seasons have minor league numbers included, adjusted for age and level. As a general rule of thumb, discount Triple-A numbers by 10 percent and Double-A by 20 percent. Whether or not you believe in projections, there’s a reason why this is used to establish a baseline—it works. It serves as an excellent jumping-off point.
2. Predicting expected skills growth – Pop quiz: A hitter stroked 12, 15 and 18 homers the past three seasons, 18 the most recent. Assuming equal playing time throughout, how many will he swat in the upcoming campaign? Using projection theory, 16 or 17 would be the answer, depending on weighting, age, rounding, etc. But what if he is still climbing the learning curve? A minority of players do improve one season to the next. Even if you believe the player is getting better, remember the golden rule: temper expectations. Algebraically, 21 is next. Maybe you think he’ll do better. Your team, your call, but keep your hopes under control. If performance spikes were predictable, they wouldn’t be surprises.
Skills growth usually come in bunches. The more skills improve as a group, the better the chance gains are maintained.
For hitters, strikeout (or contact) and walk rate are the two principle metrics to judge skill. Additional hitting skills include line drive rate and hard hit percentage. As data collection improves, more second level stats are accessible. Using our home run quiz, if this player is displaying some combination of better contact, more plate patience and hitting the ball harder, there’s justification for more than 18 homers.
Whiffs and walks are also basal pitching skills. Inducing a ground ball is a skill that minimizes homers, but realize, more grounders end up as hits than fly balls. For the purpose of this discussion, grounders are good. It’s still unclear the extent pitchers limit hard contact, so we’ll consider strikeout, walk and ground ball rates the leading indicators. It follows if a pitcher’s ERA is trending downward, it’s likely to continue if skills are improving (and vice versa).
3. Discerning luck from skill – Players with superior skills ultimately realize better outcomes and teams with more skilled players enjoy greater success. But on an individual basis, production isn’t always reflective of skills. Since a certain skill level should result in a certain outcome, these unexpected results need to be fleshed out or at least normalized.
HITTERS 
Batting average on balls in play (BABIP) – Each hitter establishes his own baseline. Contributing factors include hit distribution (line drives result in hits about 70 percent of the time, ground balls about 25 and fly balls about 15), speed, power and amount of hard contact (regardless of the trajectory, hard contact is most likely to result in a hit, soft contact is next, with medium contact most apt to become an out). If a player’s batting average history is inconsistent, check the BABIP components. A high line drive rate or hard hit ball percentage can support an elevated BABIP. There’s no assurance the elevated level will repeat, but at least the positive results weren’t all luck. If there’s nothing to support elevated BABIP, there’s an outstanding chance it regresses to career norms not captured in the player’s baseline. Lowering your base hit expectations also diminishes runs, RBI and steals. A BABIP lower than normal can be a result of dumb luck or a decrease in line drives and hard hit balls; thus, more hits and associated production is in the offing.
Home runs per fly ball (HR/F) – The number of homers is a function of the number of fly balls multiplied by the HR/F rate. Much of a player’s HR/F is intrinsic skill, but there’s also an element of luck. The same swing can clear the fence in one venue while nestling into leather elsewhere. Perhaps the first swing resulted in a warning track catch, but next time, wind was blowing and the ball left the yard. Since home runs are so critical to a fantasy ranking, contributing to four categories in standard 5x5 roto-scoring, it’s imperative that your home run expectation considers if any of the previous seasons homers were lucky and not likely to recur, or if he was unlucky and a couple more should be anticipated. Like BABIP, a batter develops his own baseline, so comparing year to year isn’t that difficult.
Run production – Another element that is often left to chance is timing of base hits. Clutch hitting is a myth. As such, there are seasons a player’s runs or RBI are out of whack. The easiest way to check RBI is batting average with runners in scoring position (BA w/RISP), which should be reasonably close to the overall average. If it’s significantly higher, expect a drop in RBI (and vice versa). To investigate runs, use batting average w/RISP of those likely to drive in the player.
PITCHERS
BABIP – Pitchers tend to cluster around the league mean, as opposed to hitters controlling their baseline. Some interesting current research is investigating if a pitcher can consistently induce weaker contact, but even if possible, the happenstance when ash strikes horsehide will mask the skill. Fly ball pitchers do tend to sport a BABIP a little lower than average, while ground ball specialists carry one a bit higher. Overall, the analysis is like hitting; if BABIP is out of whack, make the expectation adjustments.
HR/F – Like BABIP, pitchers don’t control HR/F, though their home park can influence. However, a fly ball pitcher will sport a higher HR/9 than a ground ball pitcher, assuming similar HR/F rates. When analyzing a pitcher’s homer total, look at the components. After adjusting for venue, a high HR/F will likely regress, leading to fewer homers and an improved ERA. The opposite is also true.
Left on base percent (LOB%) – LOB percentage measures the allowed base runners that score. League average is about 72 percent, though elite starters can sit around 78 percent. If a pitcher’s LOB percentage is lower than average, he’s been victimized by an unusual cluster of hits, thus expect a better ERA going forward. If his LOB percentage is unusually high, this isn’t likely to continue, meaning more runs scored.
Expected ERA – A specific skill level should portend to a specific ERA. The previously listed luck factors can result in an actual ERA different than what it theoretically should be. The best predictor of future ERA is not actual ERA but expected. There are several different versions of expected ERA, most of which are readily accessible on Fangraphs.com. Examples are FIP, xFIP and SIERA. Space limitations prohibit detailed explanations, but suffice it to say: if as a group a pitcher’s expected ERA is higher than his actual, anticipate a better mark going forward. Similarly, there are instances where a hurler outpoints his peripherals and is blessed with an ERA lower than what it should be. Unfortunately, regression is likely to correct the mark to a more reasonable level.
Swing Strike Rate (SwStr%) – Strikeouts are not only a skill leading to better ratios but a scoring category. Anticipating an impending rate change affords a sizable edge over your competition. In brief, the number of times a pitcher makes a batter swing and miss correlates quite well with his strikeout rate (K/9). The league average SwStr% is 9.4, which should result in 7.7 K/9. A pitcher sporting a K/9 of 6.0 corresponds to a SwStr% of 7.9 while a hefty 9.0 K/9 should emanate from a 10.6 SwStr%. Using these benchmarks, you can determine if a pitcher’s K/9 is in sync with his SwStr%.
NOTE: Be sure to visit FantasyAlarm as these metrics and more will be discussed in greater detail all spring as we prepare you for your drafts and auctions.
IN-SEASON ROSTER MANAGEMENT
1. Skills-based analysis – The statistics we look at are the same. Initial player expectation is based on a three- to five-year foundation, while in-season, deciding if weeks or months of unexpected performance is a change of skills or luck-based naturally derives from a smaller sample size. There’s some elegant research estimating when various skills stabilize, providing benchmarks when new skill levels can be considered real. Before, regression to historical levels should be expected. After, the skill remains better or worse (don’t pinpoint an exact amount—expect better or worse than originally projected). The benchmarks are a matter of debate, but for the purposes of managing a fantasy baseball team, the following is a reasonable guide:
Strikeout rate – Two months for hitters and pitchers
Walk rate – Three months for hitters and pitchers
Home runs – All-Star Break for hitters, always regress to career level for pitchers
Hit rate – Always regress to career levels for hitters and pitchers
Ground ball rate – Two months for hitters and pitchers
2. Luck-based analysis – The same factors discussed above should be considered, but realize regression doesn’t follow a schedule. Sometimes it occurs rapidly, others taking multiple seasons. Think “should” and “likely” as opposed to “must” and “will”. Manage like the regression will occur, but don’t freak out if it doesn’t. Over the long haul, it’s better to be wrong for the right reasons than right for the wrong reasons.