Which Twins team was best, by the numbers

This past week, across the SB Nation club sites we’ve been looking at the Best Teams to Never Win a Championship. Here at Twinkietown, our folks have put forward their cases.

TJ went with the quarantined 2020 Twins.

Cooper made the case for the 2010 Twins.

Zach argued for the 2004 club.

You probably have your own thoughts about which club had its fortunes swung by unlucky breaks, bad calls, or poorly timed injuries. For me personally, I wonder most about the 2006 team. Justin Morneau was AL MVP. Johan Santana won his second Cy Young. The ALDS matchup wasn’t against the Yankees. What if Francisco Liriano’s elbow hadn’t blown out? Besides putting forward personal opinion, what I like to do is use data and historical context to better understand what happened. The numbers, if applied properly, can help us think more clearly about our opinions and convictions. That’s what I want to do today with this question. What does the data say about which Twins team was the best to never win a championship? Is there a clear-cut answer?

---

The Contenders

I’m going to limit this analysis to the Twins clubs in Minnesota, from 1961 – today. I like Walter Johnson, Sam Rice, and Joe Cronin as much as the next Twins fan, but I think most of us are more interested in the clubs from the Minnesota version of the franchise than the Washington version.

In the 59 completed seasons playing home games in Minnesota, the franchise has:

• A combined winning percentage of 0.497
• Qualified for the playoffs thirteen times
• Won three pennants (1965, 1987, 1991)
• Won two World Series championships (1987, 1991)

I’m going to make an assertion that a team that did not qualify for the playoffs cannot be considered for “best to never win a championship.” So, we’ll focus on the 13 playoff teams. Naturally, we must exclude the World Series champion teams from this exercise, leaving us with 11 teams to look at more closely. They are below, sorted by regular season winning percentage:

A couple of things before we go further. That’s a lot of playoff misery at the hands of the Yankees. I’m sorry – I lived it too. In general, it’s a lot of sweeps and lopsided playoff series losses, but that’s not what the rest of this will be about.

In terms of regular season success, the 1965 club led by Zoilo Versailles and Tony Oliva comes out on top, winning 63% of their games. That group went on to win the AL pennant and lost a tightly contested seven game World Series to the Dodgers. If you’re interested in more content about that World Series, Matt is attempting to simulate the series on Out of the Park 21 and chronicling the series as it goes.

Last year’s Twins club, that of the Bomba Squad renown, checks in 2^nd by winning just over 62% of their regular season contests before quietly exiting the playoffs in the divisional round. The 1969 and 1970 teams led by Harmon Killebrew, Rod Carew, and Jim Perry come in next – both were swept at the hands of the historically good Orioles in league championship series. Then checking in fifth and sixth are a couple of Gardenhire era clubs, the aforementioned 2006 and the Get to Know ‘Em Twins club of 2002.

As you probably guessed, there are better ways to evaluate “best” beyond simple regular season winning percentage. For the rest of this post, I’ll try to use advanced analytics to compare these clubs to determine which truly was “best.”

Method #1: Pythagorean Win-Loss

One way that we can assess how good a team’s true performance was, that strips out some of the variance caused by luck in win-loss records, is to simply analyze runs scored against runs allowed. Ultimately, the whole point of baseball is to score more runs than you allow. And using them against each other is a method that is resilient to shifts in the way the game is played over time. It is a viable method for comparing teams from the low run scoring 1960s with the happy-fun-ball homer driven 2010s. Using Bill James’ Pythagorean Win-Loss formula, which I used last week to analyze the proposed Grapefruit League south division, we can quickly get down to a single expected winning percentage number for each team. First, a quick reminder on the generic formula:

Thanks to the great work by the people at Baseball Reference, we have the data needed to quickly run these numbers for each of the eleven clubs. A quick caveat – the actual Pythagorean calculation is updated each season, usually by slightly modifying the exponent from the standard 2. Thus, the expected win percentage in the table below is slightly different than what you would get if you simply plugged the runs scored and runs allowed numbers into the formula above. Here are results from baseball-reference, sorted by expected win percentage:

By this measure, the 1965 club again comes out on top with a Pythagorean expected winning percentage of 0.614. Interestingly, ten of the eleven teams’ actual results outperformed their expected results. The lone team not to was 1969 which checks in second on this list with an expected winning percentage of 0.610. By this measure you could argue that the 1969 team actually performed a bit better than their final record. Curiously, those early Gardenhire era clubs seemed to make a habit of outperforming their underlying numbers – the three largest differences between actual win percentage and expected win percentage belong to the 2002-2004 teams.

Next, let’s pull the thread on runs scored and runs allowed a little further.

Method #2: Weighted Runs Created+ (wRC+) and Earned Runs Average- (ERA-)

There are a couple of newer statistics that were developed with historical context in mind. Weighted Runs Created (wRC) attempts to quantify a player’s total offensive value, measured by runs, and synthesize that into one metric that can be quickly compared to other players or the rest of the league. This was later expanded to Weighted Runs Created Plus (wRC+), which includes adjustments for different ballparks. Without going into the nitty gritty mathematical details, you need to know that a wRC+ of 100 is always league average and every point above or below 100 is one percentage point better or worse than league average. For example, Max Kepler’s wRC+ in 2019 was 121 – meaning his total offensive value was 21 percent better than league average in 2019. This metric is exactly the kind of thing we need to compare teams and players across eras and ballparks.

On the defensive side the same concept was used to develop Earned Runs Average Minus (ERA-). We’re all familiar with Earned Run Average (ERA) as the predominant mainstream metric used to evaluate pitcher performance. ERA- expands on the traditional ERA measure to account for different parks, leagues, and seasons – making it easier to compare pitchers across eras, regardless of the underlying run environment. Again 100 is always league average and every point above or below 100 is one percentage point better or worse than league average for that given season. As lower is almost always better for pitching stats you want to be under 100 with this measure. To illustrate, Jake Odorizzi put together an ERA- of 75 last season – meaning his performance was 25 percent better than league average.

Thanks to Fangraphs, we have these statistics for all teams back to 1901, allowing us to quickly compare our list of eleven Twins clubs. First the raw data:

Unsurprisingly, most of the table is made up of good numbers that ranked highly in the American League at the time. The 1965 club led the league in both offense and pitching. The 1969 group was 2^nd in both to the Orioles. The 2019 put together the best offensive number on this list yet ranked just 3^rd in the American League by this measure.

To make it easier to see and assess the different teams with these numbers, I plotted the wRC+ and ERA- figures for each club on an x-y matrix. Remember, league average for the two measures is 100 and the scales are reversed – lower is better for ERA-. To assess the strength of the full team, I think it’s important to consider the offensive and defensive metrics together. This visual lets us do that, and the teams that are strong in both areas are easy to identify in the upper right area of the plot below:

There in the very top right is last year’s Twins. While the Bomba Squad offense garnered much of the media attention in 2019, the success of the pitching staff flew a little under the radar. However, the pitching staff accumulated a very strong ERA- of 90 that checked in 5^th in AL. When paired with that excellent 116 wRC+, we can clearly see the 2019 club was very balanced and strong both offensively and at preventing runs. The 1969 team is the next best by this evaluation method, with both offense and pitching comfortably above league average. These two teams seem to be clearly separated from the pack by this evaluation.

But wRC+ and ERA- only measures performance value to the team on one side of the ball. Let’s take one last dive with a more comprehensive measure of value.

Method #3: Wins Above Replacement

Wins Above Replacement (WAR) is the most comprehensive statistic we have today to measure the complete value of a player’s contribution. It is not a perfect statistic (none are), but it does a pretty good job of valuing everything a player does on the field to help his team score or prevent runs. In the context of our question about which Twins team is the “best” to never win a championship, our analysis must include a look at WAR. WAR is also a great tool for evaluating across eras, because it is compared to a replacement level baseline within each season and includes adjustments for context, league, and parks. The value accumulated above that baseline can be easily compared.

Again, we can rely on Fangraphs to supply the data we need. The measurement is done by pitchers and position players separately, so we’ll need to add those together to determine the total Wins Above Replacement for a team. For the eleven clubs we are evaluating the picture looks like this:

Again, we find the 2019 and 1969 teams separated from the rest. These two teams were roughly even in their position player WAR, and the 2019 club edges out the total with a little bit more pitching value. How they went about accumulating these totals differs though. The 1969 team was led by a seven-win, 49 homer, MVP season from 33 year old Harmon Killebrew. On the pitching side, Jim Perry and Dave Boswell both won 20 games en route to posting better than 5-WAR campaigns. In contrast, last years’ group accumulated its WAR total through its incredible depth. There were no monster MVP campaigns, but a full twenty players posted greater than 1.0 WAR seasons for the 2019 Twins (led by Max Kepler and Jose Berrios with 4.4 each), and another handful contributed something between 0.0 and 1.0 WAR. By comparison, the 1969 team had just 14 players with more than 1.0 WAR.

Conclusion

Is there a clear-cut answer to our question? We have a few different teams in the running by the different measures. From the data, it seems clear the choice is between the 1965, 1969, and 2019 teams and I think each may have a case to be made. When we do a simple aggregation of the results across the different evaluation methods the 1969 team has the strongest case to make for “best team never to win a championship.” This is built on the strength of finishing 2^nd among all Twins playoff teams in each of these analytic measures.

The 1969 group was stocked with stars including two eventual Hall of Famers in Rod Carew and Harmon Killebrew. It had the misfortune of running into a 109-win Baltimore Orioles club in a condensed playoff format. From the data perspective, the 1969 Twins are the best team never to have won a championship.