Which Stats Truly Matter for Pitchers in Daily Fantasy Baseball?

With a plethora of advanced metrics at our disposal, what should we focus on when selecting pitchers in DFS?

There is so much baseball data available at this point that it can be hard to know what really matters.

The implementation of Statcast gives us more to look at -- but how predictive is it? Is it really important for daily fantasy purposes?

When looking into which hitting statistics correlate most closely with scoring FanDuel points, specifically looking at which Statcast metrics we should focus on when evaluating batters, we found that barrels were the big winner.

Both barrels per plate appearance and barrels per batted ball event demonstrated a strong relationship with the power metrics that lead to FanDuel points. It's also noteworthy that barrels showed better than even hard-hit rate, a proven stat in daily fantasy hitter evaluation.

While these results weren't necessarily groundbreaking, whittling down the vast number of advanced stats we have at our fingertips to just the most important ones is vital towards improving our daily research process.

Flipping over to the other side, we have to ask: which stats matter the most for pitchers?

It's no secret that piling up strikeouts is the name of the game on FanDuel, but what else should we be looking at? Are ERA estimators as useful as we think? What value do Statcast metrics bring to the table?

Let's see what stands out on the mound.

Strikeouts or Bust

Compared to how FanDuel allocates points to hitters, you could argue that pitcher scoring is fairly simple and straightforward.

It all comes down to striking out opponents, completing innings, and keeping the opponent off the board. FanDuel deducts points only for earned runs, so while we obviously don't want our hurlers giving up walks and hits, those aren't directly reflected in the scoring system. Throw in the occasional win and quality start, and that's it.

With that in mind, let's check out which metrics demonstrate the strongest relationship with FanDuel points.

I pulled the data of qualified pitchers from 2015 through 2020 from FanGraphs and tested how closely their advanced, batted ball, and plate discipline stats correlate with FanDuel points scored per inning (FDP/IP). Of the metrics I looked at, below are the top 10 with the highest positive correlation.

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 (more on that shortly).

Stat Correlation
with FDP/IP
Strikeout Rate (K%) 0.948
Strikeout-Minus-Walk Rate (K-BB%) 0.917
Strikeouts per Nine Innings (K/9) 0.913
Swinging-Strike Rate (SwStr%) 0.821
Called-Strike-Plus-Whiff Rate (CSW%) 0.810
Strikeouts per Walk (K/BB) 0.657
Swing Rate Outside Zone (O-Swing%) 0.473
Swing Rate (Swing%) 0.318
First-Strike Rate (F-Strike%) 0.259
Soft-Hit Rate (Soft%) 0.216

Well, if it wasn't already abundantly clear how important strikeouts are, this shows it rather emphatically. The top nine stats are all either directly or indirectly related to getting strikeouts, and it's satisfying to see strikeout rate and strikeout-minus-walk rate (K-BB%) pass with flying colors -- they're go-to metrics for a reason.

Strikeout rate is also fantastic because of how quickly it "stabilizes." This refers to how long it takes within a season before a stat has a large enough sample size to use it as new data. A stabilization point doesn't mean we can immediately take it at face value, but the smaller the stabilization point the better, and the larger the sample goes beyond that point, the more we can trust it.

According to FanGraphs, strikeout rate takes only 70 batters faced to stabilize, which is roughly 17.1 innings (per FreezeStats). That means we can start taking strikeout rate seriously after only about three or so starts, which is far faster than just about any relevant pitching stat. It's truly everything we want in a metric for DFS.

Additionally, it's revealing that swinging-strike rate (SwStr%) and the newer called-strikes-plus-whiffs rate (CWS%) also fare well, confirming their utility in conjunction with strikeout rate. SwStr% and CWS% both correlate strongly with strikeout rate at 0.883 and 0.843, respectively, making them both metrics to keep an eye on. They can help alert us when we're trying to sniff out pitchers who could be over- or under-performing or when one exhibits an in-season change to their approach or arsenal.

But What Else Is Important?

Okay, so we want strikeouts first and foremost. Check. But as noted earlier, we also need to prevent runs, too.

What happens when we hop over to what's inversely correlated with FDP/IP?

Stat Correlation
with FDP/IP
Skill-Interactive Earned Run Average (SIERA) -0.832
Contact Rate (Contact%) -0.824
Expected Fielding Independent Pitching (xFIP) -0.797
Fielding Independent Pitching (FIP) -0.794
Earned Run Average (ERA) -0.748
Walks-Plus-Hits per Inning Pitched (WHIP) -0.746
Swing Rate Outside Zone (O-Swing%) -0.724
Swing Rate Inside Zone (Z-Swing%) -0.712
Home Runs per Nine Innings (HR/9) -0.325
Walks per Nine Innings (BB/9) -0.201

Here's where we find our favorite ERA estimators -- because lower is better -- which all show a stronger inverse relationship to FanDuel points than regular old ERA. Sure enough, this reaffirms the status of SIERA as a strong foundational research piece.

At first glance, it may seem odd to see the different iterations of contact rate pop up so strongly here, but contact rate also has a strong inverse correlation with strikeout rate. Intuitively, this makes sense, as contact rate (contact per swing) effectively tells us the same thing as whiff rate (whiffs per swing) but in reverse.

In fact, contact rate has the strongest inverse correlation with strikeout rate (-0.887) of all the stats I checked out. That said, I don't think contact rate is a necessary addition to one's research arsenal, as it doesn't really tell us anything that we can't already find out from SwStr% and CWS%, let alone just using strikeout rate.

The next closest inverse correlation, though? SIERA, at -0.826. Hmm, I'm sensing a theme here.

Lastly, one thing I found interesting is that walks allowed per nine innings (BB/9) is low on this list, and walk rate (BB%) just misses (-0.122). This is presumably because walks aren't incorporated into FanDuel scoring, so both metrics by themselves don't really tell us anything.

This is why we get a much better all-in-one picture through K-BB%. We don't want walks, but pitchers who offset them with buckets of punchouts don't necessarily need an elite walk rate to be successful.

Luis Castillo is one such recent example, as he posted a 10.1% walk rate in 2019, but a 28.9% strikeout rate allowed him to still be a viable option on FanDuel. On the other hand, Kyle Hendricks posted a 4.4% walk rate that season, but a middling 20.5% strikeout rate made him an unexciting daily fantasy option.

How About Statcast and Batted-Ball Metrics?

You've probably noticed that I've left Statcast out thus far. Surely, we can dig up something useful here?

Below is a list of all the typical Statcast metrics you find on Baseball Savant's yearly pitching leaderboards and how they correlate with FDP/IP. Once again, our sample is from 2015 through 2020.

"BBE" is a batted ball event, while "EV" stands for exit velocity. "95+ MPH%" refers to Statcast's version of hard-hit rate, which measures the percentage of times a batter hits the ball at 95 miles per hour or faster.

Statcast Metric Correlation
with FDP/IP
95+ MPH per swing -0.702
Barrel/Plate Appearance -0.324
Average EV -0.199
EV on Ground-balls -0.191
Maximum Distance -0.140
EV on Line Drives/Fly-balls -0.130
95+ MPH% -0.150
Maximum EV -0.102
Sweet Spot% -0.093
Barrel/BBE -0.050
Average Home Run Distance -0.026
Average Distance -0.016
Launch Angle 0.088


Out of all these metrics, only hard-hit balls per swing (95+ MPH per swing) moves the needle, and that's because -- you guessed it -- it has a fairly strong inverse correlation to strikeout rate (-0.698). And that's logical when you consider that it's on a per-swing basis, so it naturally coincides with pitchers who miss bats more often.

However, while that's an interesting find, it doesn't usurp any of the prior key metrics we've discussed.

We also find that both barrel metrics are mostly meaningless, which is in stark contrast to what we found with hitters. In fact, there is evidence that barrels may not stabilize at all for pitchers in a given season.

Exit velocity and distance metrics similarly don't tell us much. This is likely because of the notion that pitchers have more control over the type of batted-ball allowed (ground ball versus fly ball), whereas the batter supplies most of the power. So, much like 95+ MPH% shows little correlation with FanDuel points, the same holds true for FanGraphs' hard-hit rate (-0.047).

Many of us are guilty of looking at these sorts of metrics -- myself included! -- and attributing hard-hit rates or barrels as pitcher-related traits, but that just doesn't seem to be the case. While I don't doubt that certain pitchers are better or worse at allowing hard contact than others, this really looks like something that's hard to quantify.

But circling back for a second, if pitchers have more control over ground balls and fly balls, do those two metrics have value? They both stabilize at about 70 balls in play (around 24.5 innings pitched), which is only a smidge longer than strikeout rate. Great! It also stands to reason that a fly-ball pitcher is more likely to allow home runs and vice versa for ground-ball pitchers.

with FDP/IP
with HR Rate

Honestly, this one surprised me a bit.

Although fly balls do naturally have a decent relationship with home runs -- and the inverse is true for ground balls -- both metrics have a negligible correlation with FanDuel points. The main takeaway is that while rostering a ground-ball pitcher may reduce the chance your hurler gives up a dinger, it has little effect on his overall FanDuel performance.

But if we take a step back, this does make sense when we consider that ground-ball pitchers aren't necessarily elite, and fly-ball pitchers aren't necessarily bad.

In 2019, Justin Verlander led all qualified pitchers with a 45.2% fly-ball rate and posted the second-best strikeout rate (35.4%) and third-best SIERA (2.95). On the other hand, Dakota Hudson produced a league-high 56.9% ground-ball rate, but his 18.0% strikeout rate and 5.08 SIERA were both seventh-worst.

Once again, it all boils down to strikeouts.

Closing Thoughts

Most of these findings probably don't come as a big shock, but what they do demonstrate is how much fluff we can cut out of day-to-day research.

This doesn't touch on other aspects of pitcher evaluation such as park factors and matchups, but when it comes to individual pitching stats, we really only need to concentrate on strikeout rate and the metrics that closely correlate with it. K-BB% and SIERA are great catch-all stats, while SwStr% and CWS% give us additional metrics for predicting strikeout rate.

Meanwhile, traditional Statcast and batted-ball metrics are far more important when it comes to analyzing hitters and can be largely ignored for pitchers. Even ground-ball rate shows a weak correlation when it comes to pitcher scoring and shouldn't be a focal point in pitcher selection.

Ultimately, strikeout rate is easy to understand, correlates most closely with FanDuel points, and stabilizes the quickest. Sometimes the simplest answer really is the best one.