02 January 2018

Free Agency Is Still Broken

The free agent market has never been a good place for baseball teams to find talent. Back in 2014, I wrote an article discussing how the free agent market will break your heart because of the high bust rate of even top free agents. Since that article, it seems that things have gotten worse.

The dataset I use doesn't do a good job tracking option years. It's probable I'm missing a few players that had an option. There were 144 total players signed to a free agent deal worth at least $3 million that covered 2016.  Of these 144 players, 45 contributed either zero or negative WAR to their teams (31%). Another 61 contributed between zero and two WAR of value (42%). Only 38 contributed 2+ WAR (27%). As a result, one free agent WAR cost $8.6M in 2016. In total, these 144 players contributed 186.9 WAR (1.3 WAR per player).

The Detroit Tigers were the huge losers in 2016. They spent $115M on Justin Upton, Victor Martinez, Jordan Zimmermann, Mike Pelfrey, Anibal Sanchez, Mark Lowe and Prince Fielder. One caveat of this is that Prince Fielder was traded to the Rangers in 2013, and so Detroit wasn’t fully on the hook for his 2016 salary. If we include Fielder on this list as a player that the Tigers signed as a free agent to a deal that covers 2016, then they received -1.5 fWAR in return. If we exclude him, then they received .3 fWAR. Regardless, nothing will change the fact that they spent a significant amount of cash on players that were worth roughly the same as replacement level players.

The Orioles spent roughly $70-$75M to sign Chris Davis, Matt Wieters, Pedro Alvarez, Darren O’Day, Yovani Gallardo and Ubaldo Jimenez. In return, the Orioles received 6.2 fWAR - $12 million per WAR. This was only slightly worse than average, and illustrates how teams need to spend a significant amount of money in free agency just to get a slight advantage.

The primary exception to this rule in 2016 is the Cubs, who signed Jon Lester, Dexter Fowler, Jack Lackey, Ben Zobrist, Jason Hammel, David Ross, Jason Heyward and Edwin Jackson to free agent contracts covering 2016. Heyward and Jackson were largely busts, but the other five were huge successes. As a result, the Cubs received 26 fWAR from free agents while just spending $117M. Each free agent WAR cost the Cubs $4.5M, suggesting that free agency can be an excellent option for teams if they can hit on multiple players. Their luck largely ran out in 2017 when Lester, Lackey, Zobrist, and Zobrist fell back to earth and their combined horde of free agents signed for the 2017 season only contributed 6 fWAR at a cost of $96 million.

If the numbers for 2016 look bad, then the numbers for 2017 are catastrophic. There were 165 total players signed to a free agent deal worth at least $3 million that covered 2017. Of these players, 53 (32%) produced negative WAR and another 21 (13%) produced zero WAR. 58 (35%) more produced between 0 and 2 WAR exclusive while only 33 (20%) produced 2 or more WAR. As a result, one free agent WAR cost $14M in 2017 and these players contributed 128 WAR (.78 WAR per player).

Detroit again was a big loser, as they spent a fortune on Jason Upton, Mark Lowe, Jordan Zimmermann, Anibal Sanchez, Mike Pelfrey and Victor Martinez and received 2 WAR on their investment. On the bright side, things would have looked even worse if they didn’t trade Fielder to the Rangers previously. It’s worth noting that Detroit appears to have significantly cut payroll for 2018 and is presumably trying to rebuild their team without having to rely as heavily on free agents. The Angels also did pretty badly as their two main free agent splashes, Josh Hamilton and Albert Pujols ended up costing roughly $55M for -2 WAR.

The Dodgers, however, had significant success as they hit on Justin Turner, Kenley Jansen and Rich Hill. Worth noting is that all three of these players were originally Dodgers that ended up resigning with the club in free agency. In other words, they were known quantities and perhaps more akin to extensions than the average signing. As a result, the Dodgers paid $105M for roughly 16 WAR or an excellent price of just $6.5M per win.

The Orioles spent $73M in the free agent market and ended up receiving 1.1 fWAR for their efforts, resulting in a $/WAR of $66.1M. It turns out that their signings of Castillo, O’Day, Davis, Gallardo, Jimenez and Trumbo did not work out well for them – or anyone else for that matter. Just think, the Orioles could have signed Scherzer and Greinke for the amount of cash they spent on those six players and have money left over. Then again, it is well known how risky it is to spend money on starting pitching in free agency.

Minnesota was one of the most successful clubs. They signed three free agents whose contract covered 2017; Ervin Santana, Jason Castro and Ricky Nolasco. Santana ended up being one of the best pitchers in the majors, while Castro was an average catcher and Nolasco was traded to the Angels. Even including Nolasco in this analysis, free agents signed by the Twins produced 8.6 WAR at a cost of just $34 million.  If we exclude Nolasco, then they produced 7.3 WAR at a cost of just over $20 million, as the Twins were able to earn a wildcard spot in the playoffs. 

Organizations simply can’t build a winning team out of free agency. There simply isn’t enough free agent talent available unless a team is able to hit on three or four of the best players. Even in that case, the cost is still prohibitive for all but large market clubs. Instead, teams need to build their foundation with team-controlled players. Teams need to be able to accurately assess their talent and determine which players can help their team over the long run and should receive long extensions.

Teams that aren’t in huge markets should go into the free agent market only sparingly and only when they’re really comfortable about a player’s future performance.  Most free agents only contribute a minimal amount at a significantly high cost. From an owner’s perspective, it has to be tempting to just pocket $70 million in profits rather than spending the cash on free agents that will likely do little to help a team going forward.

This means it does make sense to devote a significant part of payroll to one or two players if a team is sure that they’re good (cough cough Machado). Most free agents contribute minimal production, so spending money to keep elite talents in the fold while focusing internally to fill out a roster makes sense. Realistically, small and middle market teams don’t have an opportunity to have many elite talents at once.

This is extremely relevant to the 2018 Orioles. The Orioles should realize it’s highly unlikely that they’ll be able to add significant talent to their club via free agency, so they need to be able to gauge whether their current team has enough talent to win. Since they don’t, it makes sense to look into which of their current players should be considered part of their future core and which players should be traded as part of a rebuild.

Baseball teams today only win if they’re able to develop team controlled young talent, that’s supported by extended home-grown elite talent and a few smart free agent signings. This means that teams can only be successful if they can gauge their own talent and determine which players they should keep and which should be traded based on the situation of their organization. Have the Orioles learned this lesson?


Jon Shepherd said...

I wonder if since the big contracts often include extra years to keep per year cost down, whether linking contract seasons to overall production makes sense.

Cameron and the main groups using these values tend to look at prorated value given projected performance. That takes into consideration future salary years. A one year snap shot might obscure some of the data and the sheer money paid to top end talent could make that unevenness look worse.

There is also the issue about a contract is not a contract is not a contract. A 1/10 deal for a pitcher is quite different than a single year in a 4/40 deal. Context is a significant issue if we are thinking at a large scale about the usefulness of the FA market.

Matt Perez said...

If big contracts include extra years to keep per year cost down, then presumably players early in their contract are underpaid. If so, players early in their deal should make up for players later in their deal. If you'd like, we could see how players early in their deal perform compared to those later in their deal.

Ultimately, you need something like this to make a baseline even if you don't use it as a projection.

Jon Shepherd said...

But that would assume a static environment which is certainly not true for an individual team and may not be accurate for the league in general.

Anonymous said...

The flip side of this is that a team has a certain amount of money to spend on salaries and they should spend it. If teams are doing their due diligence then they should be able to reasonably predict who might succeed - analytically - based upon age, general health, historic performance curves, medical exams, etc... It won't always work, but it should get you better than average. It's like the stock market - you do your research and make your plays and, if you did your research well, you should succeed more than average. Going with the same metaphor, the team should sell high. If you "hit" on a FA, then you should plan to trade him the next offseason, regardless, to build your farm. If you do this often enough, you might be able to avoid spending on IFAs and still build a farm. The fallacy in all this seems to be holding successful FAs until they are bad. If you don't "hit" then you can look at trading bad contracts to make the best match for team needs (see Kemp/Gonzalez/others). If this is really a market then teams should treat it like a capital market without prejudice or future expectation. If you do it right then you should never have to rebuild as you keep turning talent over (FAs into prospects or into better matching bad contracts). Just like the stock market, you can expect to win on average over a long period of time.

Unknown said...

Jaime Garcia, Vargas, Feldman, etc would not cost a lot. Heck, Brewers got Chosin for 15 mill for 2 years. Sign Bruce for lefty bat, try to dump Trumbo for another bad contract.

Matt Kremnitzer said...

I mean, that sounds easy, but who knows at this point if the O's are even looking to add three starters. Also, that rotation would most likely still be bad.

Matt Perez said...

Jon - I mean, if the environment changes drastically, then that says something also. I'm not proposing using 2017 numbers to necessarily predict 2018. But certainly, reality can't just be ignored.

I guess I'd think of it like this. Suppose I'm hired to manage a portfolio and earn 10% in interest. If I earn 6% in 2017, that doesn't necessarily mean my goal is to earn 6% in 2018. I'm still hired to earn 10% in interest. It could mean, however, that either a lot of things went wrong, 10% interest is an unreasonable ask, or I'm simply incompetent.

But earning 6% interest happened.

Roger - The free agent market isn't like the stock market. The free agent market is a zero sum game (at x $/WAR) while the stock market isn't. In addition, all the participants in the free agent market are extremely sophisticated. It's hubris to believe that one team can easily outperform all the other teams.

Trading free agents after one successful year can be difficult. Most free agents are expensive, so any return you receive will be minimized by that cost. It probably makes more sense to take on salary from failed free agents in return for prospects.

Jon Shepherd said...

What I am trying to articulate is the difference between worth vs. value and the complexity of that. What happened can inform us about what may happen, but we need to describe what happened thoroughly and be open to uncertainty in our assumptions.

In other terms, the club may not have paid 12 MM or so for a win even if that is what they got. Beyond that, contracts are signed in moments of time with the expectation of cost appreciation as well as spreading out cost to maximize the near future. So, I think the discussion is a bit more textured.

So, yes, you can have a receipt based perspective or a prospective perspective and as value appreciates differently than expected, the intent and actually of the efficacy of contract parameters may be altered.

It is complex, which is why those of use who think about it still struggle with figuring out exactly what is the best way forward. I think it really depends upon what question is being asked.

Matt Perez said...

"In other terms, the club may not have paid 12 MM or so for a win even if that is what they got."

What I think you're trying to say is that the club may have intended to pay only $10M for a win. I would respond to that by saying that regardless of intent, a team paid $12M per win.

In this article, I don't care about a teams' intent. I don't care whether a team thought they were spending $3M per WAR. I don't care whether a team thought that Ubaldo Jimenez was going to be an ace that provided 10 WAR a year and laid golden eggs, and therefore was a huge bargain at 4 years and $50M. This is where we assess what happened.

I'm focusing solely on the actual results. If a team paid $12M per WAR, even if they thought they were paying $10M per WAR, then they still ultimately paid $12M per WAR.

I don't necessarily think that teams wanted to pay $14M per WAR in 2017. I don't think the Orioles wanted to pay Ubaldo Jimenez $50M to put up below average numbers. But the fact is that they did.

Teams need to use this knowledge to figure out what they want to do about it. That might be reducing salaries offered to all free agents, or building better models to predict which players will be successful and which to stay away from. Regardless, understanding the actual environment is the first step to building projections for the future.

Jon Shepherd said...

I think that simply is a disagreement in our philosophy. I am interested in what teams are paying for and you are interested in what teams paid for. The actual environment is not exactly applicable. Talent is not an open system. You have limitations. Knowing the cost of intent is effectively what the cost of the market is and you try to find inefficiencies within that market. What you are looking at is a generic way to identify, perhaps, inefficiencies, but I think resolution needs to be increased.

Contract linking applies regardless.