Advanced MLB DFS Strategy: Using Recent Stats vs. Season-Long Stats in DFS

Thunder Dan Palyo identifies some of the most important stats to consider when building daily fantasy lineups in MLB.

The first half of the MLB season is in the books and the second half is set to kick off tonight with the Yankees and Red Sox being featured on national TV. Then, tomorrow we get the rest of the league back in action with a monster Friday slate.

I hope you had a chance to enjoy some of the All-Star Break festivities this week while regrouping and recharging your batteries for this fantasy baseball season. I enjoyed watching Pete "the Polar Bear" Alonso crush homers in the derby to defend his title and seeing Ohtani being featured as the starter and leadoff hitter for the AL in Tuesday's game was pretty cool, too.

If you're reading this article, it's because you are interested in diving into some stats a little deeper than most folks, and what I want to do here is explain part of my process when researching both hitters and pitchers now that we have fairly large sample sizes of data. The problem at this point in the season for us DFS players is this - maybe the sample size is TOO big? At some point do we start focusing on some of the micro-trends instead of macro trends when targeting our hitters and pitchers? I say yes, we should - although it can't be the only data we consider, it should be part of the process.

This article is a follow-up to a series of strategy articles that I wrote here earlier this spring for our MLB DFS strategy guide. You can find ALL of our DFS strategy articles here. If you ever want to chat with me about MLB, DFS, or fantasy sports in general, hit me up on Twitter @ThunderDanDFS.

Last 14 Days vs. Season Long For Hitters

So one trap that I think all of us can fall into very easily is continuing to pick on the same teams with pitchers (or fade these offenses in our lineups) all year long. Once the first month of the season is in the books, we decide that Detroit, Pittsburgh, Arizona, and Miami suck on offense and we should use just any decent pitcher against them. The reality is that the baseball season is incredibly long compared to every other major sport and that any given team's performance is going to vary greatly when we zoom into some two-week windows.

One of my favorite things about DFS is being able to take calculated risks on a day-to-day basis. On any given day, a crappy pitcher like Merrill Kelly can strike out 12 Dodgers (he did this earlier this year) and a relatively unknown slugger like Ryan Mountcastle can hit three homers in a game (this actually happened this year, too). When playing season-long fantasy baseball, you learn to live with the ups and downs and variance of players' performances on a daily basis. We DFS wackos are trying to guess outcomes correctly every day - which at times can be maddening but you have to lean into it and embrace the unexpected. Otherwise, you'd just stack the 2-3 best offenses every day and whichever pitcher has the best matchup on paper - and what fun is that?

(click to enlarge)

What I like to do is take a look at each team's collective performance against the handedness of the opposing starter over the last 14 days. I know some guys like to use 10-day samples, and that's fine, but with a travel day or two in there now you are looking at maybe only 9-10 days worth of data, which could be just three series. What I've done in these charts (vs. RHP pictured above) is taken each team's K rate, wOBA, ISO, hard contact rate, and ground ball/fly ball ratio and compared their last 14-day sample to their season-long numbers.

What I am looking for here is some big disparities, either positive or negative, in each team's performance against RHP over the past two weeks. These numbers are from the final two weeks of games leading up to the All-Star Break. A few teams with noticeable positive trends across the board against RHP would be the White Sox and Phillies, while the Cubs and Cardinals both had negative trends in nearly every category.

I wouldn't put too much stock into a team striking out 1-2% more/less often, but 5% less often (like the Red Sox) is significant! One of the sneakiest stats here is the final column - the GB/FB ratio. It's one of those stats where an increase is a BAD thing, as it means that the team is hitting more groundballs and fewer fly balls. Fly balls are GOOD because fly balls = home runs. It's not surprising that the White Sox saw the biggest increase in ISO (nearly a full point!) and also one of the biggest increases in fly balls. As a team, they swatted 20 homers over the final 14 days before the break.

(click to enlarge)

Here are the number against LHP. One of the teams that I targeted with lefties for the first month of the season was Detroit as they were striking out over 30% of the time and hitting for no power against LHP. But the recent trends here suggest that they have improved considerably against lefties over the last two weeks, and that has been a pattern that has extended beyond that window, too. Their overall numbers against lefties are still some of the worst in MLB, but I personally want to factor in these recent samples before making my decision to target them, stack them, or fade them.

Remember, just because a team HAS been hitting better doesn't mean they will continue that trend. And if anything, you could even use these numbers in reverse and look for teams that might be due for regression - with the logic that they may be under/overperforming in the recent sample compared to the larger data set. Those are decisions that you have to make - do you roll out of a hot offense even if their overall numbers are bad or fade a good offense that is slumping?

Last 30 Day Samples for Pitchers

If we are looking at the last 10-14 days for hitters, then why should we extend it to a 30-day sample for pitchers? Well, it's simple. You are going to only get two starts worth of data from a starting pitcher in a 14-day window. By extending it to 30 days, we should get at least five and possibly even six of a pitcher's most recent starts. One really bad start can still skew the data here, but not nearly as much as it would using only a two-week rolling window.

(click to enlarge)

With all this talk about hitters, what do we do with pitchers, right? Well, I just threw together a much smaller chart here with a few pitchers who are scheduled to start on Friday for their respective teams. What I did here took their last-30 days' stats and reconcile them with their overall season data to look for increases or decreases in a few key stat categories - XFIP, K rate, walk rate, WHIP, ground ball rate, and hard contact rate.

You might have other stats that you look at on a daily basis, and those aren't the only ones that I consider, to be honest. But I just wanted to illustrate the principle here of looking at a pitcher's recent performance when compared with their overall numbers.

Let's talk about Martin Perez. While he enjoyed a really good stretch of starts in May and early June, his numbers have all got worse with the exception of walks (neutral) and a few more ground balls. A big drop in strikeouts and a big increase in hard contact would make me think twice about using him, even if his overall numbers are decent.

On the opposite end of the spectrum, you have Kyle Hendricks. He was a guy we were all stacking against early in the season as his velocity and movement on his pitches were way down and he was giving up a bunch of homers. But he seems to have righted the ship and is giving up less hard contact recently while most of his other stats have stabilized. Without looking I am going to guess his HR/FB rate has normalized, too, making him a guy we are less likely to stack against than earlier in the year.

Julio Urias is walking a lot more hitters and Andrew Heaney has seen the biggest increase in WHIP and xFIP of the guys on this list. Both lefties have shown some poor recent trends and are guys I would avoid in the interim even if they both have solid strikeout potential.

CONCLUSION

I'll keep it short and sweet here. You can't rely solely on recent trends in DFS, the larger sample sizes are important, too. How much you weigh each sample against the other is up to you and honestly might be different from player A to player B. But if you're only looking at overall numbers or surface numbers and not willing to dig a little deeper to see which players are performing better or worse in recent samples, I think you are missing a chance to get a leg up on the competition. There are always going to be opportunities to get ahead of the field in MLB DFS as long as you are willing to put in the time and crunch some numbers! Until next time, good luck in the second half of the season!

More Fantasy Baseball Analysis