The Process of JeDi Leadership

By Greg Bird, AngelsWin.com Staff Reporter - 

In his third season as GM Jerry DiPoto is a lame duck leader. He’s gone all in this year but it looks like the second best record in the major leagues will only be good enough for first runner up. At least that gets the Angels into the one-game wildcard playoff. While this is better than the previous few years it still is rather anti-climactic for the second best team in baseball. Judging a GM’s performance and whether he should be retained has to do with both judging his process and his results. Should Arte give Jerry a new contract?

Having the second best regular season record in all of baseball is a good result. DiPoto can’t control who is in his division and can’t really be criticized for another team’s success. Reaching the playoffs is a great judge of baseball success since winning in the playoffs is really about who gets hot at the right time. A GM’s skill is witnessed by a team’s regular season record. In the end it is best to judge the future of a business and its leader mostly on their process and only slightly on their results. A GM’s results often follows from his process. Is DiPoto’s process good? 

In this installment of Inside Edge we will take a look at Jerry’s management process. Admittedly, this article has a bit of conjecture in it. We are not privy to any inside information from front office personnel, but through the numbers we will make a few reasonable assumptions. As always, we will do our best to explain the complicated statistical parts. You are always welcome to skip to the conclusions if you get bogged down in the numbers.

The Facts

What do we know? During the 2013 AngelsWin Spring FanFest DiPoto was asked which stats he prefers for initial analysis of a player. As usual he didn’t answer the question directly but he did provide some useful information. He said that he had been encouraging his scouts, coaches, and analysts to start with OPS+ when looking at hitters. He didn’t say this was his preferred stat but rather how he was managing his team. He elaborated a little about that being something that his staff could wrap their mind around easier.

The more we’ve thought about that here at Inside Edge the more we realized that it says more than anyone initially thought it did about JeDi’s front office process. It tells us he doesn’t necessarily push his own ideas on everyone around him, but he certainly nudges people forward. He doesn’t use the power of his position to enforce use of his favorite statistics, but rather encourages them to use a stat that they’ll be able to easily digest. He understands sabermetrics but he also understands the difficulty many old baseball people have with implementing it in day-to-day operations.

You may wonder, ‘how did you come up with this conclusion?’ Again some of it is conjecture, but some of this we can prove by the numbers. We said that he understands sabermetrics and the difficulty in implementing it system wide. How do we know this by the numbers?

DiPoto chooses OPS+ as the stat he wants to be sure all his baseball people are familiar with and use. Is this the best stat to use to judge a hitters performance? No, it isn’t. Is OPS+ that much worse than those other, better stats? No, it is not that much worse than ‘better stats.’ Are those better stats as easy to understand as OPS+? No, not at all. Let us explain this in detail.

First a Little History

What is the purpose of all of these statistical measures? The simple answer is to determine who on a team is most responsible for the runs the team scores. “Runs are the currency of Baseball.” The team who scores the most runs wins the game. Duh! One of the big questions sabermetricians have always tried to answer is, “Who is really responsible for each of the runs a team scores?” If a team can accurately measure each player’s true contribution in terms of runs then they could make a good decision on who to add and who not to add to their roster each season.

Sabermetricians have been doing this for a long time so which is the ‘best stat’ to evaluate a hitter’s contribution to the team? This has been debated throughout baseball history. Let’s look at a few of the stats that have been used by fans and teams alike throughout the long history of the game.

One of the first ways baseball teams judged the value of a player was batting average. The player who had the highest batting average was given the batting title and considered the best hitter in baseball. He was considered to have provided the most value to his team that year. Then the mythical Triple Crown was created as people realized just getting hits wasn’t enough. Guys needed to create runs for their teams. The initial thought was the guy who knocked in the most runs (RBIs) and who hit the most homeruns was the most valuable guy in the league. 

These methodologies were seen to be problematic as early as the 1950’s. Branch Rickey and Alan Roth were trying to construct a team to maximize its wins by obtaining players who would provide the most value to their team. They ruled out RBIs and batting average as useful metrics and set to determining how best to determine how runs were scored.

In the late 1970’s Pete Palmer created a measure of production by each player called On-base Plus Slugging (OPS). OPS was made famous by the book “The Hidden Game of Baseball” which Palmer wrote with John Thorn in 1984. It was an attempt to use two earlier statistics, on-base percentage and slugging percentage, to create a more descriptive stat that encompassed the total contribution of each player. 

The two newest statistics that have risen to the top in the long process of measuring a hitter’s value are Runs Created (RC) by Bill James (late 1980’s) and Base Runs (BsR) by David Smyth (early 1990’s.) These are more complicated stats but they have been proven to be more accurate at predicting team runs than all others.

Evaluating the Statistics

(Warning: math described from this point forward, main points are in italics for those who don’t like math.)

It is from Palmer’s OPS that we derive the more modern stat of OPS+. A player’s OPS is derived from the simple addition of a hitters on base percentage and his slugging percentage. OPS+ takes the player’s OPS for a season and divides it by the league average OPS for that season. This number is then divided by the overall park factor of the parks he played in that season. Park factors for each field are determined by comparing the home team’s runs scored and the runs they allowed at home versus the runs the team scored on the road and the runs they allowed in road venues. Those two numbers are divided to create a number usually between .8 and 1.5. Once the OPS number has been divided by the park factor of the parks played in that season it is finally multiplied by 100. This sets the number on a hundred point scale where 100 means the player, in a neutral park, had the league average OPS for that year.

OPS+ allows evaluators to compare player’s OPS from different run scoring environments, who hit in different park environments, on equal footing. It allows us to compare apples to apples. It became increasingly obvious over time that the simple weighting of singles, doubles, triples, and homeruns in a player’s slugging percentage wasn’t accurate. It also became clear that simply getting on base wasn’t nearly as valuable as hitting for extra bases. With this in mind many Sabermetricians tried to decipher how to properly weight each offensive event in terms of how many runs it produced. If a new statistic could correctly predict the number of runs a team in the past scored then it could conceivably estimate how many runs a future team could or should score. From this was born more advanced hitting measures. Due to the complex nature of weighting offensive events Runs Created (RC) and Base Runs (BsR) are not nearly as easy to calculate or wrap our minds around as OPS.  

RC in its early state, which we use in this article, takes the total number of hits and walks a player has and multiplies it by their slugging percentage and then divides it by their plate appearances. Bill James’ original formula has been improved upon by adding in other important offensive events like stolen bases, caught stealing, hit-by-pitch, and sacrifices in recent years.

BsR (sometimes abbreviated BR, not FanGraphs’ Bsr which is a base running metric) uses the basic idea “that Runs Scored = baserunners * (% of baserunners that score) + home runs.” Here we use Smyth’s basic formula which is non-homerun hits times total bases plus walks minus hits and homeruns. That is all divided by total base plus walks minus hits and homeruns plus at bats that did not result in a hit. Finally homeruns are added back into the whole thing. These are not intuitive statistics. They are very complicated and not easily explained without a lot of math. They are valuable simply because they accurately estimate runs. 

The question still remains, how well does a player’s OPS describe a player’s personal contribution to a team? Since what every team really cares about is the number of runs an offense scores we will use each of these as team stats and compare those to the number of runs a team scores. Through this we will be able to see how well each stat correlates to the number of runs a team scores. 

In our comparisons we used all team seasons from 1955 to 2013, except the 3 strike years (1972, 1981, and 1994.) We compare each stat as a team stat to team runs scored for that year on a simple graph. The solid line in each graph is the line of best fit line for the data. Finally we find the correlation coefficient, the r value for each, and square it. The resulting value, the r2, represents a percentage of runs each individual statistic accounts for. Thus if there is an r2=.901235 we could say that the stat accounts for 90.1% of the runs a team scores.

To give us a baseline we will start with the antiquated statistic Batting Average (BA), the earliest hitting measure. 

  

Batting average doesn’t correlate very well to runs scored. You can see on the graph it does correlate some but a team with the same BA can have wildly different amounts of run scored. BA only accounts for about 66.62% of the runs a team scores. Another way to say this is teams with the same team BA have team runs scored that vary by as much as 33%. 

Next we will look at OPS. Do teams with similar team OPS score similar amounts of runs?

The graph shows outliers but the spread is much better than batting average. Most of the data is bunched up around the line of best fit. We can conclude that teams with similar team OPS score a similar amount of runs. Team OPS accounts for about 89.16% of the runs a team scores. 

Now we can look at the real question, how much better are the best hitting metrics than OPS? Since OPS+ is rooted in OPS it will carry all the inherent flaws and benefits in OPS. If OPS is considerably worse than RC or BsR than DiPoto’s talent evaluators are beginning their analysis with a flawed assumption. But if OPS’ flaws aren’t that pronounced than maybe he has other reasons for using OPS+ that could supersede using a more advanced metric.

The basic formula for RC is superior to OPS and more closely correlates with team runs. If a team has an RC of 600 then they will probably be somewhere around 600 team runs scored +/- 50 runs or so. RC correlates with team runs 92.16% of the time but that is only 3% better than OPS. If we were to choose the more accurate measure we would be wise to use wRC+ instead OPS+. wRC+ is created by comparing it to league average RC and adjusted for park factors, just like OPS+. In wRC+ 100 is league average runs created in a neutral park. But wRC+ is only slightly better than OPS+ because it is based on a measure that is only slightly better.

BsR is the most accurate of the hitting measures even in its earliest state. Since we are using only Smyth’s first equation we can only assume the more advanced forms are more accurate. In this form BsR accounts for 92.98% of the runs a team scores. I have seen data for more recent forms of the statistic that include more offensive events that say BsR/BR has 98%+ correlation to actual runs scored. There is no BsR+ statistic. In fact it is still hard to even find BsR on the web on Baseball Reference, FanGraphs, or Baseball Prospectus. In 2008 The Hardball Times switched from using RC to BsR (they notate it ‘BR’) for their calculation of Win Shares for players. Due to the lack of availability, and without any associated statistic to adjust for park or league effects, BsR has been relegated to the domain of only hard core baseball sabermetricians. Someday it may be better known, but it is not widely used at this time. 

Conclusions (no more math)

What does DiPoto gain by using OPS+ compared to what he loses? While OPS is inferior to other metrics it is a decent measure of individual production to runs scored. JeDI loses very little in terms of accuracy. 

OPS+ has the distinct advantage of being easier to understand and explain to his front office team. Generally, if people understand something they are more apt to use it and buy into its value. None of us can enforce our will on others successfully. Attempting to do so will only create a toxic environment that is counter-productive to inspiration, teamwork, and productivity.

We can only assume JeDi knows the numbers comparing OPS and other more advanced metrics. His use of OPS+ shows us that on top of being savvy with sabermetrics he is also savvy with managing people. He realizes that older baseball guys have value and instead of replacing all of them with people who agree with him (think Houston) he chooses to try to upgrade them slowly by introducing them to a metric they can easily understand. With this he can merge the wisdom of those who’ve come before with the knowledge that has been developed by the sabermetricians of today. He does all this without having to become an overbearing dictator enforcing his will on others. 

DiPoto can still, within his inner circle, use and evaluate players with more advanced metrics. He didn’t say which stat he prefers to use to evaluate players himself. He only told us which stat he likes to use organization wide. This shows us how savvy of an interview he can give. He answers our questions without telling us anything he wants to keep secret. 

What he has told us is that he understands people. He understands them enough to lead them well. Of course, this is all conjecture. He could be a real jerk to his employees, but one would assume we would see a bit of this in his interviews. He appears affable and open. 

For a GM to be both good with player evaluation and good with people must be a rare combination. This can’t be easy to find (consider the Phillies’ Ruben Amaro.) In our opinion this, coupled with this year’s team success, means that Arte would be foolish not to extend the contract of the JeDi master in Anaheim into the future. We believe his extension will mean continued success for our beloved Halos.

Love to hear what you think!