After reviewing the list there are some things that certainly stand out. How could Barry Zito have earned $20 million more than Curt Schilling? The answer is obviously that The Giants significantly overpaid for Barry Zito’s services. It’s relatively easy to see where the bad value lies, but how can we go through this list and find where the best value lies?

In order to make sense of this data, we need to find a common measure. A standard way of measuring success, if you will. When we have players like John Smoltz and Mariano Rivera, we can’t simply look at Wins because the Saves would be excluded. Likewise we also can’t simply look at innings pitched, because a closer is inherently going to through less innings than a starter throughout the course of a year.

The most important metric for a baseball team is Wins. To quote Jonah Hill‘s character in *Moneyball*, “Your goal shouldn’t be to buy players. Your goal should be to buy wins.” Over the course of the last five years in baseball, a team has averaged 60 Save opportunities per season. In a 162 game season, that is roughly 37% of the games. So let’s use that as a starting point. A Save is worth 37% of a Win.

To validate, let’s compare how much these players are paid per Win to how much a closer is paid per Save. The average cost per Win (excluding Smoltz and Rivera) is $684K per Win. In contrast, Rivera was paid $260K per Save. If we do the math ($260K/$684K) we see that a Rivera was paid about 38% of what the average cost per Win was for these top earning players. The two numbers closely match, so the assumption seems to hold up. With our assumption in mind, I’ve created a Wins + Weighted Saves metric – which adds the Wins total to a weighted total of Saves.

TOTAL WINS + (TOTAL SAVES * 37%) = **WINS PLUS WEIGHTED SAVES**

With this new W+WSV metric, let’s compare how these richest pitchers rank in terms of value. When you look at the data based on the Cost per W+WSV, you can see players like Tom Glavine, Roger Clemens, and Greg Maddux provided reasonable value compared to the bottom of the list. We also see that pitchers like Smoltz and Rivera who have a large amount of Saves in their careers still provide some good value based on their lifetime earnings.

There are certainly some players who don’t appear to provide value for their salaries, like AJ Burnett or Barry Zito, but one player sticks out like a sore thumb – Johan Santana. It’s unfortunate how Santana’s career had fallen so quickly for the New York Mets. Personally I was a huge fan of Santana when he played for the Twins, and I was there in 2004 when he broke the single season strikeout record. Still, the numbers don’t lie and it would cost nearly 3x as much to get a win from Santana than it would Tom Glavine. That is the definition of overpaid.

]]>Sources:Salaries provided by worthly.com; stats provided by baseball-reference.com

There is no doubt that Teddy Bridgewater has upside, and his rookie performance was one of the better seasons by an NFL rookie. But when we compare him with other quarterbacks who had impressive rookie campaigns, how does Bridgewater stack up? More importantly, how can we use his rookie statistics to *predict *his future success?

In order to understand if Bridgewater will be successful, we need to have a way to measure success. The great quarterbacks like Peyton Manning and Tom Brady WIN. Winning regular season games, getting into the playoffs, and winning a Super Bowl are the ultimate measures of success. Let’s establish that winning percentage is the singular most important way of measuring a quarterback’s performance.

Now that we know what goodness looks like, we need a frame of reference for comparison. As the comparison set, let’s look at 30 active quarterbacks who started a significant amount of games in the last year. I am purposefully staying away from the mess in Cleveland just for the sake of sanity, and adding the past two Vikings’ starters. In the chart below, I have grouped quarterbacks into tiers based on their win percentage.

The chart tells us a few things. First, the highlighted *blue group* are quarterbacks who have made it to the Super Bowl. In order to make it to the show, the quarterback must consistently be over .500 on the year. Secondly, Teddy Bridgewater’s first season puts him on par with average quarterbacks like Jay Cutler and Cam Newton winning an average of 8 games per season. It shouldn’t take much to push him to the next level of a consistent winner. However, we also have seen players like RGIII regress after their rookie campaigns. Since we only have one season to examine, we need to dive deeper and see if there are specific things we can focus on that will help link Bridgewater’s performance with long term success.

My first thought is to look at how some of the other quarterbacks on this list performed during their first year of significant action. The reasoning behind this is that good quarterbacks likely start out stronger than an average or sub par quarterback.

The problem with this view is that bad teams have a better draft position, and therefore good quarterback prospects tend to go to bad teams. Football is a team sport, and while the quarterback is generally accepted to be the most important position, that player cannot win without a supporting cast. If you just look at how Peyton Manning did in his first year, you’ll see that first year performance is not a good indicator of how a quarterback’s career will eventual be. In fact, both Geno Smith and Drew Brees had the same first year winning percentage as Teddy Bridgewater.

The pie graph above indicates the likelihood of improvement over the first year, measured by the number of games compared to the first year. Based on the sample data, there is a 53.4% chance that a quarterback will either not improve, or go backwards after their first year. On the converse, there’s a 16.7% chance we could have an elite quarterback who will improve by 4+ games per season.

Instead of just focusing on the first season winning percentage, we need to look at other data to see if they more strongly relate to success. I pulled ten different season one metrics for the quarterbacks in our data set, and determined the correlation coefficient with career winning percentage. The greater the distance from zero, the stronger the positive or negative correlation.

While season one win % has a weak correlation, season one touchdown percent (TD/Attempts) has a slightly stronger correlation. I was surprised that completion percent had a lower correlation, given how strong Bridgewater is in that area.

When we look at Teddy Bridgewater’s TD %, the picture isn’t the most exciting. Bridgewater ranks near the bottom third, and is behind such greats as Matt Cassell and Christian Ponder. On the positive side, he still had a better rookie season than Eli Manning, Drew Brees, and Joe Flacco – all of whom had won a Super Bowl.

I think there is reason for optimism. As a rookie, Bridgewater has shown he can be a capable starter winning half his games with a limited supporting cast. He improved throughout the year, completing an average of over 70% of his passes during the last five games of the season. However, checking down to an open receiver is less important than the ability to score touchdowns – an area where he certainly needs to improve. I’m not ready to anoint him franchise savior, the data just isn’t strong enough yet. Next year will be critical, and should be very telling about his future.

]]>

Where I have a problem is *how* ESPN used data. I’m going to copy an excerpt from the article:

On the surface this seems like a legitimate data point. More quarterbacks being drafted early means it is a more important position. Using the same data ESPN used from Pro Football Reference, we can plot the QBs drafted by decade* [figure 1].* In the graph we can clearly see that as of the 2000’s, more quarterbacks drafted than any time in NFL history. Furthermore, the trend for the current decade will likely have *even more* quarterbacks drafted.

But let’s look at the framing of the data: From 1960-1979 compared to 2000-2013. There is one big series of events in this timeframe (the last 50 years) that has affected the need for quarterbacks, and it’s not importance.

**NFL Expansion**

Since 1960 there have been **ten more teams added** to the NFL. Simple logic dictates that if you increase the amount of teams, you will have an increase in the number of quarterbacks as well. Let’s take a look at framing the number of quarterbacks drafted compared to the number of NFL teams in that given decade [figure 2]. In this example, we can plainly see the correlation between the number of teams in the NFL, and the number of quarterbacks drafted in the first two rounds of the draft.

**A New Measure is Needed**

Instead of just looking at the pure number of quarterbacks drafted, let’s look at a different metric. I propose looking at the percent of quarterbacks as compared to all players in the draft, or *QB Draft %*. By viewing the data in this manner, we will eliminate the number of teams in the NFL as a factor in the analysis. Looking at the QB Draft % by decade *[figure 3]*, we can see that while the importance of QBs has risen in the last three decades, the ’60s appear to have had more emphasis on passers. Furthermore, if we apply a linear trend line* [figure 4]*, we actually see a slight *decrease* in the QB Draft %. This contradicts the point ESPN made.

**What Does The Future Hold?**

While I don’t have data (yet) to prove it, I believe the next ten years are going to be a boon to quarterbacks coming out of college. With the new collective bargaining agreement in place that contains a rookie salary cap, teams can afford to take more risk in drafting a marginal QB in the early rounds. General Managers don’t have to fear making a Jamarcus Russell sized mistake guaranteeing $32 million, and instead can get by with an E.J. Manuel sized $10 million commitment. The success of Andrew Luck, Russell Wilson, and Cam Newton illustrates that teams can get wins without breaking the bank.

]]>