clock menu more-arrow no yes

Filed under:

Pinpointing what’s changed with Taylor Rogers

New, 4 comments

Location, Location, Location... A breakdown of what’s behind Rogers’ 2020 inconsistency

Cleveland Indians v Minnesota Twins Photo by Hannah Foslien/Getty Images

This past weekend, Matt wrote about Taylor Rogers’ struggles so far this season. As he explained then, the stats don’t show any one particular cause for his inconsistent performance and many of the numbers conflict with one another. Estimators like FIP (2.07), xFIP (2.84), SIERA (2.74) indicate he’s pitching as well as ever. Statcast measures like average exit velocity (87.0 mph) and whiff rate (26.7%) are near career bests. He’s striking out batters at a rate in line with his career averages and walking his fewest ever. And yet his ERA is over 5.00, he’s blown two saves, and lost another game in a non-save situation. Those fantastic numbers at the top are counterpointed by the fact his hard-hit rate (35.3%) and barrel percentage (11.8%) are up from the past, driving a career-high (and probably unlucky) batting average on balls in play (BABIP) of .424. From a stats point of view, it’s important to remember these are tiny sample sizes – only 11.1 innings and 49 batters faced through Monday. But it’s not what we’ve come to expect from Taylor Rogers and the 2020 season basically is a tiny sample size.

Matt’s post published before the game on Sunday. That game, of course, saw Rogers enter for a save situation with a two-run lead. He would give up a run on two hits to make things interesting before locking down the Twins win and road series victory over Kansas City. That outing seems to perfectly encapsulate Rogers’ season thus far – including a strikeout that made the hitter look bad, a rocket allowed (106 mph exit velocity) on a poorly located 2-strike breaking ball, and a flared RBI single on a breaking ball down and out of the zone that had just a 74.5 mph exit velocity. For Rogers, it was the maddening 2020 combination of dominance, deserved loud contact and extra base damage, and bad luck all in one outing. That combination and the comments on Matt’s post made me want to do some further investigation – is this just bad luck, or has something changed? What’s going on?

Pitch Mix

The first place I looked, beyond the numbers above, is his pitch mix. Has he changed how he is attacking hitters? What I found is that his mix remains generally in line with last year – roughly equally split between his sinking fastball (51.9%) and breaking balls (45.5%). Rogers can be a bit of a handful for pitch tracking and classification algorithms because he throws both a curveball and a slider. The systems have a hard time distinguishing between the two and tend to lump them together or mis-classify one as the other. I’ll just combine them and go with generic breaking balls for this analysis. Regardless, the mix is mainly the same as the past. Nothing to see here.

Pitch Type Results

If the mix is the same, how about how the pitches have performed? Fangraphs calculates stats called pitch type linear weights that measure how well a pitcher has performed when using a certain pitch. There’s a lot of math behind the weights, but at a high level linear weights sum up the changes in run expectancy that occur from pitch to pitch into an estimated run-value above or below average for each pitch type. Here is where I might have found our first clue about Rogers. Last season, both Rogers’ sinking fastball (6.5) and breaking ball (7.4) were well above average. So far this year, the sinking fastball has been below average (-1.3), while the breaking ball remains above (+0.9). These results are also backed up by more traditional measures. Look at how his sinker performance has changed from last season:

Now, remember these are crazy small sample sizes – Statcast reports only 11 batted ball events against Rogers’ sinker this season through Sunday. This could be an anomaly, but it gives us a place to pinpoint when trying to understand his struggles.

Why, all of a sudden, is his sinker being hit so hard? Has something changed?

Statcast Pitch Profiles

Digging further into the Statcast data, I found there are a few very slight changes in Rogers’ sinker. First, his velocity is down ever so slightly to 94.2 mph on average. Last year averaged 94.8 mph – but that was a career high. He’s pitched successfully at this velocity (and lower) in the past. The velocity decrease shouldn’t be that big of a deal by itself. I kept looking. How about the movement profile? Here again, I found a very slight change. His sinker is moving slightly less – both horizontally and vertically – but only by about half an inch. Last year, his sinker averaged 23.7 inches of drop (gravity included) and 17.4 inches of horizontal break. This year those numbers are 23.2 inches and 16.9 inches, respectively. The changes are super small. Could only half an inch each way, combined with slightly lower velocity, lead to worse results? It doesn’t seem very likely. How about the spin profiles of his pitches? I found, by and large, there’s not a dramatic change there either.

So, the pitch mix is mostly the same and the stuff is mostly the same. That leaves us with location. Is there anything discernibly different in how Rogers has located his pitches?

Statcast Plate Discipline

I again went back to Statcast, this time looking for macro changes in plate discipline stats. Is he throwing strikes like he usually does? Overall in 2020, I found Rogers is throwing slightly fewer pitches in the strike zone, as measured by Zone %. Last year his rate was 55.0%. This year so far, it’s 53.1%. Again, not a huge change, but maybe something. That lower zone rate also tracks with a decreased first pitch strike rate of 62.2%, down from 68% last year. These data points imply he’s fallen behind in the count slightly more often than last year. A consequence of falling behind is that batters tend to get more aggressive when they are ahead in the count – often because pitchers will turn to their fastballs to get back into a count. There’s a reason those are called “hitter’s counts.” In Rogers’ case, that baseball truism plays out in the data – he’s seen a higher swing rate overall (49.1%, up from 45.3%) and on pitches in the zone (65.6%, up from 59.3%). Also, worth noting is Rogers has played into this by throwing more sinkers when behind in the count than he has in the past. Over 70% of his pitches when behind in the count this season have been sinkers. Last year that rate was 64%. Perhaps we have our first tangible takeaway – he’s fallen behind more, putting hitters in a position to look for a fastball to drive, and then he’s giving them the chance to hit one. Perhaps that is a factor in the sinker’s poor performance?

More Granular Location Data

That plate discipline data above gave us a macro view of Roger’s strike throwing. Throwing the ball around and over the plate is what describes “control”. Throwing the ball consistently to specific spots within and around the strike zone is what describes “command.” To really understand what’s going on, I needed to understand how Rogers has commanded his pitches. I next investigated more granular data about his locations in and around the zone. Is there a change in the quality of his pitch locations? Statcast tracks pitches thrown on the corners of the plate by Edge %. So far in 2020, Rogers Edge % is down a bit to 40.6% from 43.5%. The flip side of Edge % is Meatball %, which are pitches thrown right in the middle of the zone, belt high. Here again, Rogers has seen a change for the worse – he’s sporting a Meatball % of 11.4%, up from 8.8% last year. As a high-leverage reliever who often works in tight spots, even one meatball has the potential to ruin an outing. So far in 2020, Rogers has thrown 20. To put that in context, Rogers had thrown 175 pitches through Sunday. If he had maintained last year’s rates, we would have expected him to throw 5 more pitches on the corners instead of the middle of the plate. To illustrate the point, here are side by comparisons of Rogers 2019 and 2020 locations:

The heat maps match up with the Edge % and Meatball % data. See how both his sinker and breaking ball most frequent locations have moved up and towards the middle of the zone in 2020? In fact, Rogers’ average sinker location has been higher in the two months of this season than it was in any month of last season save for a tiny 11 pitch sample in October. Look at the right side of the black line in the chart below.

Now I think we’re on to something that explains Rogers struggles this season. Something is causing him to miss up and in the middle of the strike zone. What might that be?

Release Point Data

When I see a pitcher missing locations regularly, I always look to see if something is different mechanically with their delivery. From a publicly available data perspective, the most convenient place to look for something changing is a pitcher’s release point. Thanks to tracking at brooksbaseball.net, I found that Rogers has changed his release point so far in 2020. The chart below is a plot of his horizontal release points the past two seasons. We can clearly see a change of a few inches in the 2020 data points on the right side of the chart.

In short, Rogers’ release point has dropped ever so slightly, causing him to throw around the baseball more. Already a lower-slot slinger, a change like this means Rogers is throwing even more from the side. This is a mechanical flaw that is probably preventing him from driving down through his pitches in the manner necessary to get them down in and below the zone where his sinker and breaking ball are most effective. The release point change likely is a cause of him leaving pitches up and to his arm side (and in the middle of the plate, as the heatmaps showed). At minimum, a change like the one displayed in the chart above would impact his ability to consistently hit his desired spots and also may be the cause of the subtle movement profile changes I mentioned above.

I think this is the leading culprit for his inconsistent performance so far – mechanical inconsistencies that are negatively impacting his command, causing him to fall behind into more hitters counts, his sinker to live higher in the zone where it is less effective and gets hit harder, and to make more frequent location mistakes when he’s ahead and trying to put a hitter away. If you’re interested in a detailed breakdown of Roger’s mechanics and how a change like this could impact the success of his different pitches, Matthew Trueblood wrote a good one over at Twins Daily.

At the macro level, Rogers is as good as ever. He’s still throwing strikes and striking hitters out. But in the game of baseball, and especially on the mound, the margins between success and struggle are very thin. Even a very slight mechanical change like this could be the difference. In the pressure cooker of high-leverage relief, even tiny mistakes can hurt.

The good news for Twins fans is that this is correctable – and Rogers has corrected it before. Look again at the release point chart above – see the spike in May 2019? That spike corresponds with Rogers’ worst performance month of last season, when he posted a 4.10 FIP and only struck out 23.9% of the batters he faced. In no other month last season did he have a FIP above 3.60 or a strikeout rate below 30%. He was able to make the adjustments to right the ship then. Let’s hope he can do it again.


John is a contributor to Twinkie Town with an emphasis on analytics. He is a lifelong Twins fan and former college pitcher.