Which Hitters Are Streaky?

I wanted to take a break from my Effective Velocity research, and so, in honor of Anthony Rendon, I decided to take a look at streaky hitters. Before this season, I did not have a perception of Rendon as streaky. But so far this season, he has produced 5 of his 7 total home runs, 7 of his 21 total runs, and 15 of his 28 total RBI in just two games. That’s insane. It may not mean anything in this case; however, it makes intuitive sense that some hitters would be more consistent than others. And beyond even the desire to roster consistent hitters over streaky hitters in weekly formats, I think it is at least worth exploring whether it is possible to identify hot and cold streaks as they are happening and make start-and-sit decisions with them in mind.

The first step toward that goal is defining what hot and cold streaks are. Rendon probably isn’t the best example of that. In general, I think of a hot hitter as one who produces well above his typical level of production for an extended period. As a starting point, I decided to look at players who produced a wOBA over seven consecutive games or more that was either 110 points above their seasonal line (hot) or 110 points below their seasonal line (cold). That 110-point threshold is fairly random, but it is based on the difference in the glossary markers for an excellent and awful player over a full season.

The tricky part is in handling overlapping streaks. When someone has a game like Rendon did on Tuesday, you can create a seven-day hot streak that includes it, an eight-day hot streak that includes it, and so on. My preference is to consider only the longest such streak as a hot streak and throw away all of the other streaks that are contained within or overlap with that streak. That said, I opted to allow hot streaks and cold streaks to overlap with each other. For example, in 2017, Rendon has two streaks based on my specifications. His first is a cold streak that lasted from April 2 through April 29, and his second is a hot streak that lasted (so far) from April 27 through yesterday—not surprisingly, that hot streak is more-or-less bookended by those two monster games. It’s not perfect, but I think it produces a fairly accurate count of streaks for hitters.

So far in 2017, there are 11 players with at least 4 total streaks: Ender Inciarte, Yangervis Solarte, Carlos Gonzalez, Trevor Plouffe, Ezequiel Carrera, Lorenzo Cain, Andrelton Simmons, Adam Duvall, Kole Calhoun, Delino DeShields, and Max Kepler. But I don’t expect there to be a clear takeaway from those results because hitters have had less than two months to establish their seasonal baselines. Instead, I think it’s more instructive to look at full seasons, and I’ll be using a 300-plate-appearance minimum to further stabilize player seasonal expectations.

In 2016, Gregory Polanco had the most streaks in baseball with 8. Kris Bryant had the second most with 7. It makes sense that players with excellent power like Bryant would be more prone to hot and cold streaks because home runs produce the biggest change in wOBA of any possible event. After those top two streaky hitters, there were 23 hitters with 6 streaks, including notable fantasy players like Carlos Correa, Chris Davis, Nelson Cruz, Mark Trumbo, Daniel Murphy, and Jason Heyward. Most of the players with fewer streaks were consistently bad, but Joey Votto, Ryan Braun, Yoenis Cespedes, and Dexter Fowler were some notable productive players with only 2 streaks.

For research purposes, more interesting than the players themselves is the typical length of these streaks.

How Long Do Streaks Last?

In general, there is a consistent downward trend, which matches up with the expectations I had. However, it looks like it is pretty common for streaks that last at least 7 days to continue to last up to about 21 days, at which point the frequency really drops off of a cliff. That trend bodes well for the idea of identifying a streak while it is happening, which is something I’ll dig into more in future weeks.





Scott Spratt is a fantasy sports writer for FanGraphs and Pro Football Focus. He is a Sloan Sports Conference Research Paper Competition and FSWA award winner. Feel free to ask him questions on Twitter – @Scott_Spratt

12 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Rotoholicmember
6 years ago

There is a LOT of work to do to prove streakiness isn’t entirely random. Tango Tiger has done a lot of work on hitters that are “locked in” and it has always shown zero predictability, from what I recall. I really doubt you will have any success identifying a streak before it has ended. That is, a player’s performance in game n+1 of an n-game hot streak is measurably higher than his performance over the course of the season. Injuries can affect this, so if a guy is particularly cold it could be that he has a wonky shoulder and we could predict that he will indeed continue to be on a cold streak. But for hot streaks, it’s not really predictable.

It’s also kind of strange to extend the streak longer just because a guy had a few really good games earlier on. So a mediocre hitter could have a 3 HR game and then go 0 for his next 20 PAs over the span of 5 games and still be on a hot streak due to the amount of wOBA he has banked. It really needs to get more granular than that, as it doesn’t pass the sniff test that a guy who goes 0-20 is still considered to be on a hot streak. This would reduce the length of the streaks significantly, be more likely to be predictive of the continuation of the hot streak, possibly eliminate the overlap between hot/cold streaks, and just in general be a much more accurate reflection of what we would consider a streaky hitter.

I really do like this kind of work being done, challenging or testing the ideas that people have about the game and trying to quantify it objectively.

RonnieDobbs
6 years ago
Reply to  Rotoholic

Predictive is one thing, existence is another. It is not a series of coin-flips… I don’t get what there is to quantify or prove. Anyone who watches a team every day can identify a streak or streaky players. In reality, streaks exists because of mental or mechanical factors, which can’t be represented with data points. It is telling of the times that we try to make everything measurable and predictable, but that is not reality. Video games and fantasy sports are fun though!

jfree
6 years ago
Reply to  RonnieDobbs

I don’t understand why this is being downvoted. I personally disagree with the notion that streakiness can’t really or shouldn’t be measured or attempt to be predicted.

But completely agree that there is a strong erroneous tendency to simply attribute ‘modelling failure’ and everything we can’t figure out yet to ‘randomness’ – and pretend that that assertion is meaningful because it looks more quantitative.

RonnieDobbs
6 years ago
Reply to  jfree

I pay attention to streakiness – it is something I try to identify. I think it is a very important and real player attribute. I think it is a huge factor that WAR and all stats miss. Consistency is valuable in real baseball games. When you measure a player by his season counting stats, you miss the context, which is what decides real baseball games. More often than not, those multi home-run 6-hit games come in Coors against poor pitching in a blowout… or similar circumstances. IMO Streaks should be discounted similar to how wOBA discounts extra-base-hits relative to the traditional weights applied to them – for example, a double isn’t really twice as valuable as a single.

jfree
6 years ago
Reply to  RonnieDobbs

My personal belief is that ‘streakiness’ is not the same thing as ‘slumpiness’. iow – I really don’t think the huge single games/series are much more than simply a result of a good matchup at an opportune time. Slumpiness on the other hand imo is very real and not well understood. A combo of player flaws re their plate approach with an inability to therefore succeed via plan B if plan A isn’t working for whatever reason. And that slumpiness can have a very different impact if the player is a big long-term contract v a league minimum with options guy.

RonnieDobbs
6 years ago
Reply to  jfree

A guy who is streaking is a guy who is not slumping. They are two sides of the same coin. A guy gets going well for the same reasons that he struggles. People like to pretend that the cause is something highly visible and observable, but it most likely isn’t. More often than not it is between the ears.

RonnieDobbs
6 years ago
Reply to  jfree

Upvotes and downvotes are not a measure of quality discussion in many cases. Contrary ideas get downvotes and supportive ones get upvotes. I expect downvotes as I am not just a sabermetric fanboy. Obviously, I find the discussion interesting, but I am not just swallowing down other peoples ideas and that doesn’t go over well for a lot of readers. I have noticed that the more casual ‘weekend readers’ come in with floods of downvotes. The core of this community is more interested in a genuine discussion, but that doesn’t describe everyone.

nielsie
6 years ago
Reply to  RonnieDobbs

How do you know who is or isn’t down/up voting you?

RonnieDobbs
6 years ago
Reply to  nielsie

I am just guessing. I don’t actually know anything. I think the comment section self-identifies itself in its up and down votes. Complimentary comments consistent get upvotes even if they don’t say much. A contradictory comment is rarely the top comment. Some readers are just pro-sabermetrics and not so much baseball. It seems like some folks are just intent on downvoting anything traditional.