Fixing the Rockies, a data driven approach: Part 1, diagnosing the problem

Editor’s note: Purple Row’s FanPost section is a place for anyone with a registered account to put together an article to share with the community. A couple weeks ago, community member amuesing1 posted what I have no doubt is the most thorough and extensive FanPost Purple Row has received in its nearly 15 years of existence. Not only that, it’s really, really good!

I’ve edited and broken up the FanPosts so that they can get a wider readership. They’ll appear here and be published every day or two until it’s complete. If you just can’t wait, you can still read the full FanPost here.

For anyone who has ever taken the time to write a Purple Row FanPost: Thank you using the platform and for choosing to share your ideas with this community.

-Eric

★ ★ ★

I don’t have to tell anyone that the Rockies had a bad season. I’m constantly reminded about how they made the playoffs two years in a row for the first time in franchise history, almost a division title, game 163, etc. Then all of a sudden when the Rockies have a bad season, everyone becomes an expert on the subject about how the pitching has been horrible, the bullpen cost too much money, and no critique is complete without mentioning Ian Desmond.

I have a few problems with this.

First, these are all complaints without a solution. Saying the pitching needs to be better isn’t a solution, and saying Desmond hasn’t helped the team also doesn’t provide any actionable information. Even offering a solution like trading Nolan Arenado, as Ken Rosenthal of The Athletic recently did, wouldn’t really solve what kept the Rockies out of the playoffs in 2019, and it wouldn’t lead them back in 2020. As fans we of course want our team to win. But if they don’t, blindly yelling about it doesn’t support the organization or help the players at all. If you have a problem with the team, provide some solutions.

Second, if someone does provide a solution, it is like throwing darts at a board. The starting rotation got hurt; why isn’t the organization spending millions on a veteran starter? Ian Desmond struck out; why don’t they trade him? This is like when someone asks why we can’t just solve poverty. The answer is because the world doesn’t bend to your will. Before you start throwing solutions against a wall and hoping they stick, you need to define the problem. I’m an aerospace engineer by training, and we certainly wouldn’t have put a man on moon by putting a bunch of gun powder under a car seat and hoping for the best. Blowing things up is not a solution.

Throughout this series of articles, I aim to address both of these concerns. It doesn’t take a rocket scientist to see the Rockies need help. Before we blow things out of proportion, we need to define the problem and can’t be afraid to look at things in a different light. Once we’ve done that, we need to use this information to provide actionable changes to help the team.

The Problems Run Deeper Than The Elbow

Why did the Rockies win in 2017 and 2018 and lose in 2019? The entire starting staff got hurt! Well now that that’s answered we can pack it up. 2020 will be great when everyone heals up. Nothing to fix here.

Except even without any stats, we all know that isn’t the only problem that needs fixing, and most of the staff was healthy for most of the season. The pitching staff was the reason the Rockies made the playoff last season, but before we continue, we need to understand why.

If you read enough articles, you’ll also see another theme about the 2017 and 2018 Rockies: over-performance. A friend joked to me that the last word he would use to describe the Rockies is over-performance, but that’s not exactly what we mean here. There is something called a Pythagorean win loss record. This is what we would expect a team’s record to be based on the number of runs they scored vs. the number of runs they allowed. On average, teams will line up with this expectation plus or minus a few games.

If a team over performs their Pythagorean win-loss record, holding all things equal, we would expect them to regress. The same is true if they under-perform. Based on the number of runs the Rockies scored the last 3 seasons, here is what we would expect to see.

The differences can come from many places: luck, statistical noise, or even just winning more close ballgames. The Rockies were 26-15 in 1 run games in 2018. Plenty of good teams have good 1 run records, but in general this is what we mean by over performing. Most teams are about .500 in one-run games, and over- or under-performing in those games is typically a sign that things will correct themselves.

The Rockies were about the same team in 2017 and 2018. A contender for sure, but only just. We certainly were not a division winner of a team. And for context, the Dodger’s Expected win-loss for 2018 was 102-61. Game 163 happened because of a slight over-performance from us and a drastic under performance from the Dodgers. If someone says we expected to contend for the division this year: the truth is we were never even close. In 2017, we performed as expected and that was enough to scrape into the playoffs. That wouldn’t have been good enough for 2018, but it would have in 2019 if things had gone better. This year, we hit our mark exactly.

Looking beyond the Rockies’ performance, we can also look at the projections prior to the season starting. According the FanGraphs, the Rockies were projected to win about 79 games each of the last three seasons. That should tell us something.

Projections are wrong all of the time. If they weren’t sports wouldn’t be exciting. FanGraphs getting even the Paythagorean win-loss record wrong isn’t that uncommon. As the saying goes, “All models are wrong, some are useful”. This model becomes useful if it is well calibrated. Meaning the events it forecasts result at the rate you expect. The only question we are asking is: does FanGraphs have a well calibrated model?

The answer is yes, they are pretty good at it. FanGraphs, FiveThirtyEight, and Vegas sportsbooks for the last 3 years all have a similar consensus: they don’t see much in the Rockies. If we beat the projections, that’s fantastic. But the reason we are looking at these is because the people who are professionals at sports forecasting — the people who will lose money if they miss something or are wrong — don’t see a winning ball club. Sure, there is some luck and magic in baseball and yes the playoffs are notoriously a crap shoot, but if we want the Rockies to win a World Series, it probably isn’t coming because we are lucky and defy the odds. It will come because we are the best.

Who’s to Blame?

But the Rockies were not the best. So who’s to blame for that? Let’s get even simpler and take a bird’s eye view look at the two things baseball boils down to two things: hitting and pitching.

The following table shows how the Rockies as a whole have done on both sides if the ball each year. The first stat is ERA+. This is a fancy way of measuring how good the staff was relative to the rest of the league, and it includes park normalization factors to account for Coors Field (we’re going to need to do this a lot). For ERA+, 100 is average and higher is better. The second is weighted Runs Created Plus. This is a similar concept, but now we are taking all the actions of the offense and assigning them a value based on how good the offense was relative to the rest of the league, also adjusting for park factors. Again, 100 is average, and higher is better. You can think of these stats as “how much better or worse are the pitchers/hitters than the rest of baseball, without getting caught up on Coors Field?”

Last season the pitching staff was 9% better than average, while the hitters were 12% worse than average. Many people around the league still aren’t aware this happened. Many assume we lucked our way into the playoffs with a hitter friendly ballpark. Accounting for park factors, the hitters were actually pretty poor last year, but the pitching staff carried us to the playoffs.

This season, the pitching staff was 5% worse than average. Looking at our record and fan comments on social media, you’d think it would be much worse, but that’s not too bad considering late season injuries and whatever it was that happened to Kyle Freeland. The main problem is not only that we didn’t improve the offense this year, it actually got worse. The pitching was able to make up for the underwhelming offense last season, but the injuries meant they couldn’t carry us this year. This also highlights a very important point. For everyone complaining about the pitching this season, getting rid of Jake McGee doesn’t begin to address the largest problem: the offense!

Diving a little deeper, we start to find the problems with the offense are in fact not just Ian Desmond’s fault, but a more systemic problem altogether.

Top Heavy

Dan Szymborski on FanGraphs wrote an interesting piece on the division of WAR amongst playoff teams in 2018. To summarize: of all playoff teams in 2018, if one were to take away the contributions of the top two hitters and pitchers on each team, 9 of the 10 teams still rank in the top 10 in terms of overall WAR. The Rockies, however, rank 19th. This highlights one thing we already knew about 2019: when our top pitchers got hurt, the team fell apart. It also highlights a much bigger problem: the Rockies are vulnerable.

Nolan Arenado and Trevor Story are excellent ballplayers, but we are screwed if they get hurt for a long period of time. Jon Gray and German Márquez are solid pitchers, but we are screwed if they get hurt. Almost anyone not named the Yankees would be screwed if their entire starting staff went down. We wouldn’t expect any team to make the playoffs in those circumstances.

But there is a middle ground between the entire starting staff stays healthy to throw 150 innings and the whole staff ends up on the IL. Very few teams can sustain many injuries, but the Rockies can’t stand any. Here is where the “buy another starter” argument begins to break down. The Rockies don’t need that one piece to put them over the top. Surprisingly, we already have that. The Rockies need an entire baseball team!

Shown below is a graph showing the WAR produced by players sorted by descending order from a team’s best player to their worst player. The Rockies have the stars just like any other team with our top 4 players, but then it rapidly drops off. The backups and replacements aren’t worse than any other team on average, but our 5th best player to our 25th best player is well below the average major league team. Every team is going to have stars, but no championship will be won if 20 out of 25 players are below average. No single veteran starter, no matter how good, can fix that gap.

This is why if the Rockies are presented with the world’s most unlikely trade for Arenado should take it. We have the stars, but we need a solid foundation. This isn’t the NBA where LeBron can singlehandedly will Cleveland to a championship. In fact those predictions we discussed before are based on how much value each player is projected to add to a team over the course of the season. How many games they play, how many home runs they hit, how many routine fly balls Ian Desmond will drop. Anyone in baseball will be familiar with WAR, but when we say start quantifying players by wins, how do we use that? What is a replacement win, how did Nolan win 5 ballgames (or how did he only win 5 ballgames)? When we say replacement, the calculation is based on what players are freely available to any team. We also expect a replacement level team to win about 48 ballgames due to the variance of the sport and how good the average minor leaguer is to the average major leaguer. This gives us a baseline. The amount of WAR a team produces, should roughly equate to X amount of wins above 48 in a season.

For example the Astros had 55.2 WAR from their team combined last year and won 103 games. 48+55.2=103.2. We also know on average how many wins it takes to get to the playoffs. In the double wild card era, that seems to be about 90 wins, give or take. It can be lower, but once again we aren’t looking to luck our way into the playoffs. Now we have a language to translate value from players into the postseason.

Taking 90 games as our base (beating the Dodgers is considerably harder) we know that as a team we are shooting for 42 WAR. This sounds like a huge number, and it is; building a playoff team isn’t easy. Starting from the player projections we talked about earlier, we can start to make a road map (Ideally, we would have projections that fall in a probabilistic range, but for the purposes of this piece we are just going to look at averages.)

We can pencil Nolan in for 5 WAR. He’s the most consistent player on the team, no need to hope for a 10 WAR season or worry about a 1 WAR performance. Trevor is listed at less than 4. This seems a little outdated, only because his defense this year has been so valuable that he’s been worth over 6 WAR this season (different sites have different calculations for defense).

This is meant to be a back of the envelope calculation, so without diving too deep we are going to assume he hits 5 WAR in 2020. Running through the whole list, we get about 20 WAR. That’s not too bad, we are almost halfway there. Adding the projections from our pitching staff, we are looking at 35 WAR in total. That means we are looking to carve out 7 wins somehow, someway for 2020 and beyond to be a playoff contending team. Ideally, we’d want a few more wins if we can. Predicting WAR isn’t perfect and the actual WAR we have next season doesn’t perfectly predict our record. These are all just indicators that we can use to inform us what we are looking for.

To find these 7 wins we’ll be looking at many things. We need depth around the diamond and we’ll need to take a closer look at our current roster. But before we start diving into the data on the players, we have to address the hardest problem in Major League Baseball, managing Coors Field.

That will be the topic of part two of this series.