Revisiting the O.P.P. Study

Free agency -- exciting isn't? It's a special kind of business (via Keith Allison).

Free agency — exciting isn’t? It’s a special kind of business (via Keith Allison).

When teams need to decide on re-signing a free agent, they have more detailed information about the player than the other teams in the league. This knowledge would especially extend to the player’s medical history. Teams that have seen the player every day would know if the player takes care of himself and if injuries affect his play. By looking at the one and two-year free agent contracts players sign and disabled list information, I found that a re-signing team doesn’t seem to choose healthier players.

To find this information, I am going to piggyback off the work Matt Swartz did for the 2012 The Hardball Times Annual where he found: “Teams re-signing their own players (whom they know well) fare far better than teams who sign deals for “Other People’s Players.”

He looked at several factors when making that statement including production, salary and playing time. Matt was kind enough to send me his updated free agent information (signings from 2006 to current) to see if teams have an advantage re-signing players when they know more about the players’ medical histories. Instead of using playing time to determine health, I am going to look at the average days on the disabled list.

I broke down my findings into several categories. First if the player re-signed with his previous team, I looked only at players who declared as free agents. Then, I used the length of free agent contract (in years). Next, I divided by the type of player (hitter vs. starting pitcher vs. relief pitcher). Finally, I wanted to look at the player from the beginning to end of the contract. There arises an obvious problem there, as many contracts that will expire at the end of 2014 or later can’t be used.

The major issue I immediately ran into was sample size as the contract lengths increased. I went as far as I could until the sample sizes got too small. By using the above criteria, I imposed a limit to examine only one- and two-year contracts. When I got to three-year contracts, I ended up with only six pitchers and five hitters who re-signed with their original team and then became free agents again.

When I split up the starters and relievers for the two-year samples, I end up with only 12 starters who didn’t re-sign. I sort of feel dirty about using such a low number. I wouldn’t draw or make any definite conclusions from this small amount of data, but I included the data anyway for the sake of being thorough.

One factor to keep in mind as we delve into the data is that these are older players. If the player was able to reach free agency under the age of 30 and had any ability, he would likely be given more than a one- or two-year deal. This is a sampling of players who have been around the diamond a time or two.


Matt’s take on hitters: Thus, we can conclude that in terms of both performance and playing time [Matt’s proxy for health], there is little difference among hitters who signed with new teams versus hitters who re-signed with their previous teams in their abilities to beat their Oliver projections.

For all the hitters, I am looking at the percentage of players who ended up on the DL and how many days the players missed on average. Additionally, I put the players’ talent into perspective by looking at the median values for the previous season’s data and the season for each year of the contract. For stats, I looked at plate appearances (PA), weighted on-base average (wOBA) and wins above replacement (WAR). Finally, I used the Marcel projections to look at the players’ projected wOBA and PA. I used the projections to determine baseline talent levels.

One-Year Contracts

Changes in Health & Performance for One-Year Hitter Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR PA wOBA PA wOBA
Before New 300 33 35% 20 0.5 0.3 310 0.312 426 0.321
Same 121 33 37% 18 0.5 0.5 304 0.313 387 0.316
Signed New 300 34 32% 15 0.4 0.1 225 0.304 380 0.313
Same 121 34 42% 26 0.5 0.1 230 0.299 372 0.311

The DL Days per player was similar for both groups the season before signing the one-year deal. In the next season, the re-signers saw the number increase, while the players on new teams saw their numbers drop. It looks like opposing teams have better understanding of the players’ health. The difference can’t be explained in production, as the both groups hitter’s actual (and projected) wOBA were almost identical.

The only real conclusion for this sample is there is none. I have a couple of hypotheses though.

The first is based on familiarity. The home teams likes the comfort of knowing the player previously. The teams that re-signed their players ended up paying a median value of $1.35 million for their rental compared to the $1 million value for hitters on new teams. The teams that re-sign their players seem to want these players more and were even willing to pay more for them. The teams signing players from other teams may not pay the extra dollar for health and production for near-replacement level production.

The second is based on macho male gung-ho nature. “I am new to this team. I need to impress them and show them I am one tough guy. My ankle hurts, but my new team needs to think that they need me. They will see I am a man.”

Two-Year Contracts

Changes in Health & Performance for Two-Year Hitter Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR PA wOBA PA wOBA
Before New 58 32.5 41% 12 0.5 1.3 444 0.324 445 0.315
Same 38 33.0 21% 13 0.3 1.0 405 0.335 428 0.315
Signed New 58 33.5 53% 29 0.7 0.5 364 0.308 460 0.315
Same 38 34.0 53% 27 0.7 0.5 298 0.308 442 0.316
2nd Yr New 58 34.5 47% 27 0.6 0.3 278 0.300 426 0.311
Same 38 35.0 47% 23 0.6 0.5 245 0.314 390 0.314

Just looking at my original assumption of teams knowing their players’ health best again doesn’t seem to be true for the first season. The combined average DL days for all players more than doubled from 13 days to 28 days, with almost no difference between the two groups. In the second year of the contract, the players who re-signed did see less days on the DL, but not to the levels before the contract. It seems like a small amount may be known about the re-signed hitters that can predict better health in their second season.

Whether the hitters go to new teams or head back to their original teams, not much of a difference can be seen in the DL days for these players. Teams that have seen the players before don’t know (or care) if the players are healthier compared to those players who go to a new team. The only possible difference comes in the second year of a two-year contract where hitters who re-signed are a bit healthier.

I need to go off on a tangent on some of the data. The concept of players going for that final contract looks to be true with these hitters. They were up for a contract and they deserved it (or at least they thought they did) and they went and got it. But nothing could be more false. Let’s look at the projections and results for the one-year and two-year contracts for the season before and after the contracts were signed:

Comparing Projections and Results

Contract Length Projected wOBA (same team/new team) Actual wOBA (same team/new team)
1 year .316 / .321 .313 / .312
2 year .315 / .315 .335 / .325

On average, the players who over-performed their expectations (and even had lower projections to begin with) got the bigger, longer contracts. The player with two-year contracts then began to regress to their mean (true talent level). The next season their projections put them around a .315 wOBA, but they ended up only producing at a .308 wOBA mark.

The same is true for the DL days missed. These two-year contract players on average missed 13 days before the contract was signed and 28 after. This pair averages to 19 days missed per person per year of the two-year contract. The average days missed for the one-year players was 20 days. The players who got the two-year deals picked a good time to have a better-than-average year.


Matt’s take on pitchers: Interestingly, it was not injuries that seemed to be the secret that private information revealed, because innings pitched [Matt’s proxy for health] were similarly short of expectations for re-signed pitchers as newly signed pitchers. However, this may be that newly signed pitchers should have been rested more and were less forthcoming about injuries, so these two factors are more inter-twined than they might seem.

With pitchers, I looked at the same data as hitters, except I use innings pitched (IP) instead of plate appearances, and strikeout percentage minus walk percentage (K%-BB%) instead of wOBA.

Starters, One-Year Contracts

Changes in Health & Performance for One-Year SP Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR IP K%-BB% IP K%-BB%
Before New 104 32 46% 31 0.7 0.8 130 6.7% 135 7.4%
Same 47 33 51% 45 0.7 1.3 147 7.6% 149 7.4%
Signed New 104 33 43% 31 0.5 0.5 76 7.0% 119 7.3%
Same 47 34 55% 49 0.7 0.7 113 6.4% 124 7.5%

The same trend exists with starters as it did with hitters. Both groups kept their number of DL days the same before and after they signed their one-year deal. The difference is the re-signing players spent more time on the DL before and after the contract, with the difference being around 15 days. The difference can be explained by the pitchers’ production.

Between the two groups, the teams re-signing their own pitchers pick pitchers who performed better the season before signing (7.6% K%-BB% vs 6.7% K%-BB%). They even put their money at work by paying a median salary of $2.5 million  for the year of service. Players going to a new team cost $1 million. In the season they signed to pitch, the roles switched, with the pitchers on a new team pitching better than the re-signed pitcher (7.0% K%-BB% vs 6.4% K%-BB%).

Starters, Two-Year Contracts

Changes in Health & Performance for Two-Year SP Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR IP K%-BB% IP K%-BB%
Before New 12 32.5 25% 11 0.3 2.3 188 7.5% 159 7.3%
Same 20 31.5 50% 47 0.6 2.0 190 9.9% 175 9.5%
Signed New 12 33.5 50% 30 0.5 1.6 154 6.0% 170 7.4%
Same 20 32.5 55% 47 0.8 0.9 150 8.5% 171 10.2%
2nd Yr New 12 34.5 50% 37 0.8 0.7 132 5.9% 158 7.3%
Same 20 33.5 35% 45 0.6 1.6 171 8.7% 157 8.9%

Don’t read too much into these data, as the sample size is a bit limited, but basically the results are similar. Starters who re-signed were more likely to spend time on the DL before the contract and after. Looking at their production and projections, these 20 pitchers were significantly better pitchers. Also, the high average numbers of days lost stayed constant after the contract was signed.

The starters who went to another team spent a miniscule 11 days on the DL per pitcher. This value is completely abnormal for pitchers over 30 years of age. The average days lost tripled to 30 days lost in the contract’s first season and increased to 37 days in the second season.

Relievers, One-Year Contracts

Changes in Health & Performance for One-Year RP Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR IP K%-BB% IP K%-BB%
Before New 167 34.0 39% 24 0.4 0.1 54 7.1% 61 7.8%
Same 61 35.5 43% 31 0.6 0.2 56 7.9% 60 8.0%
Signed New 167 35.0 35% 21 0.4 0.0 47 6.8% 57 7.6%
Same 61 36.5 61% 37 0.7 0.0 42 6.7% 58 7.7%

The exact same occurrences that happened with starters happened with relievers. Teams are more likely to re-sign their more productive, often injured players who end up regressing quite a bit.

Relievers, Two-Year Contracts

Changes in Health & Performance for Two-Year RP Contracts DL infoStatsProj. stats

Year Team # Signed Age DL% DL Days/plyr Trips/plyr WAR IP K%-BB% IP K%-BB%
Before New 37 33 41% 14 0.5 0.5 55 12.8% 56 9.4%
Same 15 33 13% 7 0.2 0.5 58 11.7% 64 8.6%
Signed New 37 34 49% 30 0.6 0.3 56 11.0% 59 10.0%
Same 15 34 47% 39 0.7 0.1 50 8.3% 60 9.4%
2nd Yr New 37 35 27% 20 0.3 0.1 46 8.8% 59 10.0%
Same 15 35 27% 21 0.3 0.1 43 12.9% 56 8.9%

Again, don’t read too much into the data, as the sample size isn’t what I would like. This is the first case of the players re-signing with a team who spent less days on the DL per player (seven days) the season before they signed than those players who didn’t re-sign (14 days). It didn’t really matter though, as the next season the number of days on the DL exploded to 39 days for the re-signed players and 30 days for the relievers on a new team. In the final season of the contracts, the relievers spent less time on the DL, while perhaps pitching through a little pain because they wanted a new contract.

Summing Up

I didn’t find any obvious trends for teams that re-signed players to one- or two-year contracts having any better knowledge about the number of days the player may miss on the DL. An exception exists. In the second year for re-signed hitters, the average number of DL days drops by five, while the number for those on re-signed teams stays constant. Now with pitchers, re-signing teams are willing to take on more injury risk to sign higher quality pitchers. As more contract data become available, larger contracts can be examined. As for now, with small contracts, a team’s previous knowledge of a player’s heath doesn’t allow an advantage in signing healthier players.

While I didn’t find anything groundbreaking, it’s important still explore our ideas to see if they are true or not. Sometimes an idea is verified. Other times we come up empty. It is not about proving your narrative right or someone else’s narrative wrong; what we want is to get closer to the truth. With so few groundbreaking studies left to do, our research necessarily becomes more granular, and that will inevitably lead to more fruitless studies — or rather, studies that bear less fruit than we would like. Now we know that better knowledge on a player’s health is useful in offering short-term contracts, but teams don’t hold it as a trump card necessarily. Time to move on to the next question with an unknown answer.

Print Friendly
 Share on Facebook1Tweet about this on Twitter10Share on Google+0Share on Reddit0Email this to someone
« Previous: When the Sparrow Sings – Sixth Inning
Next: Lost in the Sun: The Physics of Ballpark Orientation »


  1. Jason said...

    I could definitely see a difference showing up in longer contracts. A team’s knowledge and concern over the health of a free agent would be expected to result in that player, when he remains with his original team, only getting a 1 or 2 year deal. I’d be curious what the ‘so far ‘ data is for all contracts 3+ years.

    • Jeff Zimmerman said...

      I am not sure what to do with the longer contracts. I was thinking of doing the first 3 seasons for all the longer contracts. While the results would be a little biased (like probable jump in playing time in the contract’s last season), it may give some better answers.

  2. Matt Swartz said...

    Just to chime in here– this doesn’t change any of the old conclusions:
    (a) Other People’s Players still get paid substantially more $ per WAR
    (b) Other People’s Pitchers still perform relatively worse compared to their projections on longer deals

    This shows that there is no clear difference for short deals on DL days, which is an important contribution Jeff has made to understanding the broader OPP effect.

  3. 21_22 said...

    what about looking only at players that spent time on the DL in the before season? i would think a first injury or injury independent of a prior injury would occur at random meaning there would not really be any better information. however, once an injury actually occurs, then the team has data unavailable to other teams.

    also, with small samples, averages may not be the most representative measure of the distribution. for instance, what happens if you look specifically at the frequency of long or short DL stints?

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>