Baseball is known for its plethora of advanced statistics -- part of its appeal for many -- but go too far down the rabbit hole and things can get overwhelming pretty quickly. Whether it's poring over the in-depth player pages of FanGraphs and Baseball Savant or digging into the latest and greatest stat with the funny-sounding acronym, there's a whole lot to take in.
While that's all well and good, when it comes to daily fantasy baseball, whittling down the myriad of options at the statistics buffet is a necessity. There are a lot of delicious-looking choices, but we have limited time to sink our teeth into a given slate every day.
With that in mind, last year, I took a look at which statistics matter for both hitters and pitchers on FanDuel. Those studies both reaffirmed the importance of certain go-to metrics in DFS research while also proving that not every new piece of Statcast data is helpful when evaluating players when fantasy points are what matter most.
Of course, that tells only one part of the story when it comes to building lineups. As important as individual hitting talent is, matchups are one of our primary considerations when it comes to rostering hitters, and that's especially the case when it comes to stacking.
This includes external factors such as ballparks and weather, yet it often begins with deciding who the least talented starting pitchers are on a given slate. We want to attack the "worst" pitchers, but what are the best indicators of true pitching talent?
This isn't an attempt to reinvent the wheel but rather a chance to take a step back to try and simplify our daily process as best as possible.
Taking all of that into consideration, which metrics should we prioritize when it comes to evaluating opposing pitchers?
Click here if you wish to skip to the results.
How Do Stats Actually Correlate With FanDuel Points Allowed?
At the end of the day, FanDuel scoring for hitters boils down to getting on base (singles, doubles, triples, walks, and hit-by-pitches) and scoring (runs and RBIs) with the almighty home run combining those two aspects for our most sought-after outcome.
Stolen bases are the one other avenue for tacking on additional points, and while that's a reliable source of scoring for a shrinking subset of players, our focus here will be strictly on batters versus pitchers.
For the purposes of this exercise, we'll start by looking at how various metrics correlate with FanDuel points allowed by starting pitchers on a per-batter basis from 2015 to 2021. We'll be looking only at pitchers who accrued at least 100 innings in a given season, which effectively eliminates the COVID-shortened 2020 campaign entirely. But that still leaves us with 751 entries across six full seasons.
Let's begin by first looking at how key stats correlate with FanDuel points allowed. The closer the correlation coefficient is to 1, the stronger the relationship. Conversely, if the number is closer to -1, it shows an inverse relationship.
First, here are the relevant stats I tested that provided a positive correlation of at least 0.20.
|Stat||Correlation with FDP/BF|
|Earned Run Average (ERA)||0.95|
|Fielding Independent Pitching (FIP)||0.84|
|Walks and Hits per Inning Pitched (WHIP)||0.80|
|Batting Average Against (AVG)||0.77|
|Home Runs per 9 Innings (HR/9)||0.73|
|Skill Interactive Earned Run Average (SIERA)||0.67|
|Expected Fielding Independent Pitching (xFIP)||0.67|
|Home-Run-to-Fly-Ball Rate (HR/FB)||0.54|
|Barrel Rate (Barrel%)||0.43|
|Statcast's Hard-Hit Rate (95+ MPH%)||0.42|
|Batting Average on Balls in Play (BABIP)||0.40|
|Average Exit Velocity (EV)||0.40|
|FanGraphs' Hard-Hit Rate (Hard%)||0.37|
|Walks per 9 Innings (BB/9)||0.34|
|Walk Rate (BB%)||0.25|
Okay, there's a lot to unpack here, but good ol' ERA leads the way and comes incredibly close to replicating FanDuel scoring.
Needless to say, a high ERA means said pitcher has been getting rocked in both real life and fantasy life. Other results-based stats such as WHIP, AVG, and HR/9 also fare well.
While we know better than to use ERA as a predictor of future performance, it's of little surprise to see it at the top of this list.
All of these stats literally contain the different components that go into FanDuel's scoring system, so it's only natural that they are closely aligned to fantasy points. But given that these metrics are subjected to factors that are often outside a pitcher's hands (hand?), they aren't the best ones to rely on when it comes to predicting future performance.
That's because each is affected by at least one of the following "luck" stats: BABIP, HR/FB, and left-on-base rate (LOB%). Generally speaking, these three metrics tend to regress to league average over time for pitchers -- hence going well above or below can be considered unlucky or lucky.
Both BABIP and HR/FB show up with moderately strong correlations to FanDuel points allowed, and as we'll soon see, LOB% actually leads all metrics I tested in inverse correlation. Alas, as we know, luck and variance play an important role in daily fantasy baseball.
Although it's true that certain pitchers will consistently fall above or below average in these categories -- so there's some skill involved here -- they're exceptions to the rule and difficult to single out.
ERA is influenced by all three, demonstrating why it can often swing wildly for pitchers from year to year.
But that's what we have ERA estimators for (FIP, xFIP, and SIERA), right? They remove some of that pesky outside noise and give us a clearer picture of a pitcher's performance.
You're probably already familiar with these metrics, but it can't hurt for a quick refresher on what they actually try to do.
As its name suggests, FIP strips fielding/defense out of the equation and assumes league-average BABIP, evaluating a pitcher solely on strikeouts, walks, hit by pitches, and home runs. xFIP takes that one step further by allotting a league-average HR/FB rate. Finally, SIERA applies batted-ball data (ground balls, fly balls, and pop-ups) in an attempt to dial in pitching talent further.
Ultimately, FIP is the most descriptive and least predictive of this trio, which is partially why it correlates so closely to fantasy points allowed. But both xFIP and SIERA certainly aren't slouches here, reaffirming their utility as helpful tools in our DFS arsenal.
The bottom half of the table is where a smorgasbord of batted-ball metrics enter the fray. The Statcast trio of barrel rate, 95+ MPH%, and average exit velocity all further the notion that giving up hard contact is bad for pitchers -- makes sense -- and all three get a slight bump above FanGraphs' original hard-hit rate.
When it comes to hitters, we know that hard contact correlates strongly with FanDuel points, and a barreled ball is easily the most desired outcome.
However, past studies have suggested that batters are in the driver's seat when it comes to power, making it difficult to know how much a given pitcher is at fault when he allows a barreled or hard-hit ball. This is something we'll have to circle back to.
Walks round out the list, and they don't move the needle much on their own. But as we'll find in a moment, they're still important nonetheless.
With that in mind, let's flip over to the other side. Here are the significant metrics that posted a negative correlation of at least -0.20.
|Left-on-Base Rate (LOB%)||-0.66|
|Strikeout-Minus-Walk Rate (K-BB%)||-0.61|
|Strikeout Rate (K%)||-0.57|
|Strikeout to Walk Ratio (K/BB)||-0.51|
|Strikeouts per 9 Innings (K/9)||-0.48|
|Soft-Hit Rate (Soft%)||-0.35|
Pick any flavor strikeouts, and they pop as something we want to avoid in opposing pitchers. A strikeout gives a hitter zero FanDuel points, whereas putting the ball in play always gives us a chance at points. That tracks.
Meanwhile, walks demonstrate their utility when used in conjunction with punchouts. This makes sense because a 10% walk rate is far more concerning for a pitcher sporting a 20% strikeout rate than one carrying a 30% strikeout rate, which is why both K-BB% and K/BB stand out.
Both strikeouts and ERA estimators are strong indicators of DFS success when it comes to rostering pitchers, so it isn't shocking to see that they're vital to opposing hitter selection, as well.
As expected, we want to attack low-strikeout pitchers with poor xFIP/SIERA numbers, and it doesn't hurt if they have a high walk rate. But are we only scratching the surface here?
Which pitchers are more likely to give up those lucrative home run balls?
How Do Stats Actually Correlate With Home Runs Allowed?
It's dingers that ultimately win tournaments, and while we know the pitcher gets the sole blame for coughing up home runs, the amount they allow is impacted by circumstances out of their control like park factors, weather, and just plain old luck.
But we also know that pitchers have a heavy influence on batted-ball type, hence why we readily use terms like "ground-ball pitcher" and "fly-ball pitcher." This is one area where a pitcher can "control" how likely they are to allow round-trippers. A ground ball can never be a home run, so a high ground-ball rate should be one of the most effective ways a pitcher can prevent them.
Additionally, it stands to reason that some are better at suppressing hard contact than others and vice versa. But as I alluded to earlier, this is a far trickier trait to pin down. After all, even if a pitcher throws a meatball down the heart of the plate, you know Aaron Judge is more likely to smash it into the outfield seats than Elvis Andrus is.
While there are obvious scoring differences between hitters and pitchers in DFS, perhaps it's telling that quality of contact metrics have a weak-to-negligible correlation with FanDuel points for pitchers.
Despite this, when we see a struggling pitcher coughing up boatloads of hard contact, it's tempting to assume that he's at fault. Is there evidence to back this up, or is this a fool's errand?
For starters, let's see which metrics correlate with HR/9. I'll include all meaningful metrics I tested that have a positive correlation of 0.2 or higher or a negative correlation of -0.2 or lower.
|Stat||Correlation with HR/9|
|Fly-Ball Rate (FB%)||0.46|
|Launch Angle (LA)||0.42|
|Ground-Ball-to-Fly-Ball Ratio (GB/FB)||-0.44|
|Ground-Ball Rate (GB%)||-0.46|
Both FIP and HR/FB rate have home runs built into their respective equations, so it isn't shocking to see them atop the list. But the former is more of a descriptive stat and the latter is a luck metric, so they're included here more for context.
Here's where barrels really shine, though. We still need to clarify whether barrel rate is a reliable measure of pitching talent, but this is promising at least.
It's also encouraging to see a solid correlation between xFIP and home runs, demonstrating the predictive value of using league average HR/FB rate in its equation as a way of projecting home runs.
You'll notice that ground-ball rate, fly-ball rate, ground-ball-to-fly-ball ratio, and launch angle make appearances here even though they didn't make the cut earlier when we looked at stats correlated with FanDuel points allowed. Independently, they have solid correlations with home runs allowed, but they have little bearing on FanDuel points without added context.
While a bit counterintuitive at first, this stems from the fact that having a high ground-ball rate doesn't necessarily mean you're a good pitcher, and having a high fly-ball rate doesn't mean you're a bad pitcher. In 2021, a good example would be comparing Dallas Keuchel (54.9% ground-ball rate; 5.01 SIERA) to Max Scherzer (48.3% fly-ball rate; 2.90 SIERA).
Still, whether we're stacking or rostering one-offs in DFS, we're always looking for home runs. We probably don't want to throw out these stats entirely.
In a bit of a surprise, strikeout rate has a modest influence on home runs. Perhaps this is like the reverse of how grounders and fly balls don't correlate strongly to FanDuel points allowed; some elite hurlers with high strikeout rates still allow lots of fly balls, so it's possible for them to give up more dingers than low-strikeout pitchers with high ground-ball rates.
Strikeouts have already proven to be important in the grand scheme of things, so that shouldn't deter us from weighing them heavily in general. But it does suggest that when we're home-run hunting with one-offs in tournaments, perhaps a high strikeout rate shouldn't scare us away when the circumstances are right.
How Do Stats Actually Correlate Year Over Year?
Okay, we have a pretty good idea of which stats have the strongest relationships with FanDuel points allowed by pitchers, as well as which ones correlate with home runs.
But at the end of the day, we really only care about the stats that are most predictive. ERA may correlate the most closely with FanDuel points allowed, but it's about as dull and useless a tool as we'll find in our box of DFS tricks.
I've already mentioned the perceived predictiveness -- or lack thereof -- for some metrics. We should back that all up with some real data.
Let's see which key statistics have the most "stickiness" year to year. In essence, how strong is stat X's ability to predict stat X the following season?
Due to the shortened 2020 campaign, we'll be narrowing our sample to just 2015-19, but that should be enough to get us meaningful results.
|Stat||Year over YearCorrelation (2015-2019)|
Remember how I said that pitchers have a heavy influence over whether they allow grounders or fly balls? That notion is hammered home by the year-over-year correlation of launch angle, fly-ball rate, and ground-ball rate.
On their own, these stats are a smaller piece of the puzzle because of a lack of correlation to FanDuel points. We have to keep that in mind. But home runs are the big-ticket items in MLB DFS. Producing an elite ground-ball rate is very much a pitching skill, and we want to avoid that more often than not.
Ground-ball rate and fly-ball rate stabilize quite quickly, too, requiring just 70 balls in play to do so, per FanGraphs. And according to FreezeStats, launch angle stabilizes at only 45 balls in play. In other words, these are among the first stats we can trust early in a new season
We may need to include a slew of other factors when it comes to picking which pitchers to attack, but at the end of the day, grounders are a pitcher's best friend in the battle to suppress home runs.
Strikeouts just barely follow those three metrics, and avoiding punchouts continues to look like one of the biggest boxes to check off in pitching matchups. The year-over-year correlation of peripheral strikeout stats like swinging-strike rate (SwStr%; 0.74) and called-strike-plus-whiff rate (CSW%; 0.70) check out, as well, further emphasizing the importance of all things strikeouts.
Strikeout rate is also the gold standard when it comes to stabilization (70 batters faced). Its counterpart, walk rate, which also fares well above, is another stat that stabilizes quickly (170 batters faced). As we saw earlier, using the two together makes a lot of sense.
xFIP and SIERA also pop once again. While other studies have shown SIERA to be the slightly more predictive stat, they appear mostly interchangeable for our purposes, which is particularly helpful when it comes to looking at pitcher splits on FanGraphs, which lists only xFIP between the two.
Following the seven metrics I've touched on, we start to see things drop off, and it's notable that both versions of hard-hit rate fall in the same range as FIP, WHIP, and AVG.
But arguably the most discouraging result is barrel rate. Not only is it one of the least sticky on the list, but it's below good ol' ERA and barely above HR/9.
To make matters worse, while barrel rate is quick to stabilize for hitters, that isn't the case for pitchers.
According to a 2016 study by Russell A. Carleton, barrel rate doesn't stabilize until roughly 400 balls in play, and even then, he concludes that its reliability remains questionable. For some context, 115 starting pitchers logged at least 100 innings last season, and 50 of them didn't reach 400 batted-ball events. Not ideal!
As much as we want a pitcher's barrel rate to be a useful tool, it appears subject to too much variance to trust as part of our DFS process.
Finally, for context, those luck stats I mentioned earlier round out the bottom three, demonstrating exactly why they are regarded as such.
So, What Really Matters?
Alright, congratulations if you made it this far (or skipped ahead with the handy link)! We've run a plethora of stats through the wringer, and it's time to lay out the clear winners.
Strikeout Rate: Nearly the whole package. High inverse correlation with FanDuel points allowed, sticky year-over-year, and stabilizes quickly. Less influence over home runs, though. Strong relationship with SwStr% and CWS%.
xFIP and SIERA: Best "catch-all" metrics. Strips luck out of ERA and more predictive than FIP.
Walk Rate: Should be used in tandem with strikeout rate (K-BB%). Clear pitcher skill. Free base-runners are good!
Ground-Ball Rate: Pitcher's best weapon against home runs. Stabilizes early. Oversimplification but -- pitchers control batted-ball type; hitters control power. Secondary to other metrics and factors due to low correlation with FanDuel points allowed. Can also use fly-ball rate or launch angle.
Honestly, that's about it!
No, this isn't a groundbreaking list by any means. But it's a good reminder that despite the abundance of advanced stats out there, most of them aren't actually important for DFS.
Strikeout rate, walk rate, and ground-ball rate are pitching skills. These are among the handful of metrics that are largely within a pitcher's control; hence why they tend to stay consistent from season to season and stabilize early within a given campaign.
Meanwhile, ERA is perhaps the closest stat that mirrors FanDuel points allowed but is the product of a myriad of factors a pitcher can't control. ERA estimators aren't perfect, but xFIP and SIERA eliminate some of that variance.
On the other hand, a hurler's hard-hit rate and -- especially -- barrel rate have questionable utility at best. The results back the notion that while pitchers can manipulate the type of batted ball, hitters have far greater impact on power. As Russell A. Carleton states, “the pitcher gives up the fly ball, but the batter hits the home run.”
As tempting as it is to lean on these quality-of-contact statistics, when it comes to home run power, we should instead be focusing on the hitting talent of our potential stacks.
In the end, the opposing starting pitcher is just one -- albeit important -- factor when it comes to choosing which stacks to prioritize. And this doesn't diminish the fact that evaluating arms in daily fantasy goes beyond looking at a handful of stats.
Whether it's adding a new pitch, a change in velocity, or altering their pitch mix, pitchers are constantly changing and adapting, so we can't always take their season marks at face value. They can also have wildly different splits depending on the handedness of the batter.
But as far as outlining a basic foundation, these are the metrics we should actually care about. Sometimes, simplicity is key.