Fun Times with MLB Trade Data

NOTE: I wrote this just after the 2018 Non-Waiver Trade Deadline but didn’t post it until November.  I figured it was still a worthwhile read, so here it goes…

The final hours approaching the 2018 MLB non-waiver Trade Deadline were characterized by a frenzied cluster of mostly-pedestrian exchanges that were overshadowed by one headline-grabber (Chris Archer to the Pirates). Perhaps the most noteworthy part of the 2018 deadline wasn’t any particular trade, but the sheer volume of exchanges that went down. After tuning into episode 1249 of Effectively Wild last week, I wanted to pay especially close attention to the teams making deals as the deadline approached – not necessarily the players getting traded – but the teams making the trades. On the EW podcast, Ben and Jeff brought in Adam Fisher (formerly of the Braves’ and Mets’ front offices) to discuss the anatomy of trades, which piqued my interest in who traded what, and why they might have done it (outside their obvious needs lining up). Adam seemed to imply that the rapport established between front offices goes a long way in initiating the discussion between teams who ultimately make a deal. Anecdotally this sounded about right; prior to even looking at the data, it seemed to me (an A’s fan) that half the deals made by the A’s the last few years involved the Nationals…I’ve been under the assumption Billy Beane and David Forst are more cordial with Mike Rizzo than they are with most other GMs. So I set out to find who the preferred trade partners of every team are, the teams that make the most deals, the ones who make the least, and a multitude of additional frivolities to captivate the masses.

So in this post, we’re looking at deals made since the 2011 World Series – a somewhat arbitrary cutoff, but it’s an approximation of when all teams were operating with the knowledge of divisional alignment looking like it does today. While the Astros hadn’t yet moved to the AL West, their front office was aware their primary future competition would be comprised of the Angels, A’s, Mariners, and Rangers, and we assume that influenced their trading activity, as well as the teams in their current and future divisions.

One thing of note – when we aggregate trade data and compare team totals, we count each trade per team involved; when the Rangers dealt Cole Hamels to the Cubs, that was a deal for the Rangers, and also a deal for the Cubs. Since we’re looking at each trade from the perspective of a given team, the Hamels deal will be counted as two deals. I recommend not thinking too hard about it.

Trade Frequency & Game Changing GMs

Unsurprisingly, Jerry Dipoto’s Mariners lead the pack by a wide margin with 93 trades – 14 more than the next highest figure: 79 deals made by the A’s. The 14-trade margin between 1st and 2nd place is wider than any margin from consecutively ranked teams. Interestingly enough, Dipoto’s only served as the Mariners’ GM since 2015 after he was let go by the Angels, for whom he didn’t exhibit the same trigger-happy tendencies he has since his arrival in the Pacific Northwest. Nevertheless, in the 2.5 years Dipoto’s been in Seattle, he’s made 60 deals. Compare that to the 32 deals made by the Mariners the previous 4 years, and it’s pretty remarkable – not necessarily for better or worse, but remarkable in how much it stands out.

Just by looking at the chart of total trades since the 2011 World Series, we can see a gradual, pretty linear slope from the A’s down to the Cubs, who complete the top half of MLB. In contrast, the bottom half is distinguished by a steeper slope that begins with the Marlins and ends with the Giants. The visually evident divide between the top and bottom half illustrates just how top heavy trade behavior tends to be; the upper 50% has out-traded the bottom 50% 1,023 to 599 in the last six years.

Jerry Dipoto’s propensity for trading for trading’s sake has been noticed by almost everyone. With that said, Erik Neander of the Tampa Bay Rays has actually traded even more frequently than Dipoto since his promotion after the 2016 season with roughly a trade more per year than the Mariner’s GM. And while Dipoto’s wheeling and dealing has been considerably more frequent than his predecessor, Jack Zduriencik, the Mariners’ change in trade frequency ranks third behind Neander and Mike Girsch of the Cardinals. Girsch has bumped the Cardinals into the middle of the pack as far as trade frequency goes, but that’s mighty leap for a team that was among the most infrequent traders under John Mozeliak.

Probability Chicken Scratch: Who trades (and doesn’t trade) with whom?

In absolute terms, the Rays and Mariners connect on more deals than any other duo in baseball; since the 2011 World Series, they’ve struck an accord 11 times. Once we get below the 5 most common trade partners, this gets far less interesting, as the frequency of ties increases rapidly with every subsequent move down the ranking poll (that is, until we reach 1 trade, at which 117 team pairs are tied…93 pairs are tied at 0).

What’s probably more interesting to investigate is the proportion of all trades made by team X that involve team Y. Unlike the totals, the trade proportions give us a better idea of which teams are more inclined to strike a deal with a particular team when a deal is imminent. For example, since the conclusion of the 2011 season, the Mariners have made 9 deals with the Yankees. Meanwhile, the Nationals made 9 deals with the A’s. The proportions show us that 9.7% of Mariners trades involve the Yankees, whereas 20.5% of Nats trades involve the A’s. So while a Nats-A’s deal has occurred equally as often as an M’s-Yankees deal, it’s roughly twice as likely that any given deal involving the Nationals also involves the A’s. Put another way – the Nats involve the A’s in 1 out of every 5 deals they make, while the M’s deal with the Yankees in 1 out of every 10.

On the flipside, team pairs that never trade with each other may be equally as interesting (if not more so), were it not unremarkably common; 93 team pairs have combined for 0 trades. Hypothetically, however, if we assume trade partners were randomly selected by the initiating team (this obviously isn’t the case), the probability that team Y isn’t involved in any of team X’s trades becomes worse than a coin flip after X’s 20th consecutive trade not involving Y. This means that the further a team goes above 20 trades without involving a specific team, it’s decreasingly likely to be random chance, and more likely that those teams intentionally avoid dealing with each other. That all holds under an assumption we know isn’t true; teams don’t randomly select their trade partners. Instead, trade partnerships are presumably a function of both team’s front office rapport, current records, temporal focus/outlook (long term or short term), present needs, financial flexibility, competing bids, geographic markets, division placement, and other random variables. However, given that these deals aren’t sought out randomly, it stands to reason that exclusion from trades has an even smaller threshold where intentional aversion becomes more likely than not (before the 20th consecutive trade of exclusion). Since we don’t know the probability function of trade agreements, we can approximate the probability under the assumption that trade partners are randomly selected (which we’ve established as a more conservative estimate than whatever the true function would give us).

So who avoids whom? To make our data richer and get a better idea of which teams really don’t like dealing with each other, we’re going back 4 more years and starting at the 2007 World Series. This gives us 700 additional trades to observe, but more importantly, 700 more trades in which 47 pairs of teams failed to complete a deal together. How likely is it that these are coincidences? In every case, it’s pretty unlikely. In some cases, it’s almost impossible. Incorporating each team’s share of all MLB trades since 2007 and applying it to probability, we can easily calculate how probable it is that each pair hasn’t yet made a deal to this point, assuming that trade partners are randomly selected by the team initiating a deal.

The table tells us a lot, but to make it easier, we’ll look at a few examples. Lets take the pair at the very top – the Rays and Jays. They haven’t made a deal together in over a decade. The table also tells us the Rays made 92 deals with everyone who’s not the Blue Jays, while the Blue Jays made 85 deals with everyone who’s not the Rays. The probability of the Rays excluding the Blue Jays for 92 consecutive deals just by random chance (unintentionally) is 0.0785%; if we subtract that from 1, we get the probability of the Rays intentionally excluding the Blue Jays for 92 consecutive trades, or roughly 99.9% (5th column). The Blue Jays are also 99.9% likely to have avoided the Rays (6th column). The probability that either one of the teams is avoiding making a deal with the other is a near certainly at >99.9% (7th column), while the probability of both teams avoiding each other is 99.8% (last column).

At the bottom of the table we find two seldom traders in the Cardinals and Giants, who, among all the pairs, may be the least surprising – at least among the non-divisional and non-market sharing teams. So while it’s still very probable that someone’s avoiding someone, it’s no guarantee. The Cards have a probability of 80.7% of intentionally avoiding the Giants, while the Giants have a 76.9% chance of intentionally avoiding the Cards. The probability of Team 1 avoiding Team 2 isn’t the same as the probability of Team 2 avoiding Team 1 because they don’t share an equal frequency of trading; since the Cardinals have made 5 more trades than the Giants, they’re slightly more likely to be avoiding them than vice versa. When a team trades once in a blue moon, they’re not avoiding a particular team as much as they’re just avoiding trading altogether. What we see with the Giants in comparison to other teams, is general trade aversion; rather than aversion to trading with any specific team, they’re simply less inclined to trade with anyone and everyone. They’re aversion is relatively less discriminating than other teams, though it is still likely they’re avoiding deals with some teams.

A random point – the Dodgers and Tigers never trade with each other and it’s really weird. They’ve had different GMs/philosophical approaches/general front office personnel, and they’ve both made tons of trades with everyone else. But the last time the two teams traded with each other was in 2004…when Cody Ross was sent to the Dodgers for Steve Collyer! The Tigers also don’t trade with the Padres though, so maybe they’re just anti-SoCal NL teams. The Yankees and Royals also haven’t made a deal since 2000, but there’s been less turnover in their respective front offices, so maybe Brian Cashman and Dayton Moore aren’t the best of friends.

The Understandable Side of Trade Aversion

Notice how a lot of the non-trading pairs are either in the same division or same geographic market. This isn’t random chance either. As often as writers and fans label trade “winners” and “losers”, the fact is deals are generally made between teams with at least some expectation of a mutual benefit; GMs with reputations for slighting trade counterparts won’t find themselves in the good graces of their colleagues for long. When a deal is agreed upon, we often see present value being exchanged for future value (proven veterans for prospects). How much a team competes with a given trade partner is undeniably a factor when a trade is consummated, with games being just one factor of competition; leagues, divisions, free agents, fans, employees, and tv ratings can all be at stake depending on the teams involved in a given deal. Adam Fisher mentioned this on Effectively Wild when he explained why there was almost no way that Chris Archer or Jacob DeGrom are ultimately dealt to the Yankees. In the case of Archer, it would represent a deal in which a team on the cusp of competing (the Rays) send a controllable and productive asset to a team that represents more than 10% of their scheduled opponents annually that also competes for the same division. While the benefit may be mutual, you’re still offering a benefit to a team that has an edge in every competitive aspect imaginable, why contribute to it? In the case of DeGrom, you’re competing not for as much on the field as off of it; the benefit of sending one of your biggest stars across town won’t come to fruition for years, while alienated Mets fans find themselves capable of watching winning baseball if they can only convince themselves to jump ship. When you’re the #2 show in town, don’t send your biggest assets to the #1 show; you may not be on-field competitors, but economically, there’s no bigger competition.

Do teams actually prefer sending players outside their league altogether? Well, 52% of deals are interleague deals, meaning over half the trades that go down involve players swapping leagues. The difference between that and random chance is not statistically significant (15 of 29 potential trade partners are in the other league…that’s 52%). Some specific teams, however, may be more averse to intra-league trading than others; the Diamondbacks, Cubs, and Brewers trade with NL less than a third of the time. Curiously, the Reds are highly averse to intra-division trading, yet they exhibit a strong preference for trading within the National League at better than 3 out of every 4 trades.

However, as far as divisions go, we do see statistically significant trade aversion between teams residing in the same division. 8% of all trades are intra-division, while random chance would place intra-division trading at 14%. Some teams are extremely reluctant to deal in their division – most notably the Cubs and Indians, who’ve only made one intra-division deal in nearly 7 years. It was an underappreciated quasi-milestone when the Tigers sent Leonys Martin to the Indians in the final hour before the deadline this season, because it broke a streak of 8 consecutive years of non-divisional trading for the Indians.

While most teams are understandably less trigger happy when it comes to dealing with teams in their own division, three teams out-trade the random chance rate of 14% (it’s really 13.7%) within their division:

  1. The Houston Astros (15.7%)
  2. The New York Mets (15.6%)
  3. The Oakland A’s (16.5%)

There’s not much to glean here since they’re barely above 13.7%, but maybe there’s something to it – the Astros and A’s are often commended for their smart and savvy front offices. Intra-divisional trades are rarely blockbusters, and typically involve deals made at the margins. When so many others seem to prefer stagnation over marginal improvement if it requires dealing with a division rival, it’s worth nothing that improvement may be worthwhile regardless of who else it involves or benefits.

Ten teams find themselves sharing a market with another – the Yankees and Mets, the Dodgers and Angels, the Cubs and White Sox, the Giants and A’s, and the Nationals and Orioles. These teams almost never trade with each other. In fact, since the 2011 World Series, these ten teams have accounted for 569 trades; they’ve made a total of 3 – the Dodgers and Angels made 2, and the Cubs and White Sox made the Quintana deal. Go back 4 more years and they account for 791 trades collectively; they’ve still made just those 3 deals. That’s 0.76% of trades (again, you have to double count each deal so we have it from the perspective of each team involved, so it’s 6/791 instead of 3/791) – random chance would have it at 3.4%, so this is quite significant.

Going back to what inspired this particular body of research, Adam Fisher mentioned on Effectively Wild that it’s difficult to make a trade with a team you compete with for fans. Economists might refer to market-sharing teams as substitute goods, and the slightest shifts in appeal for Team X’s substitute (call it Team Y) changes the relative appeal of Team X to the consumers of baseball within a given market. Let’s say that the Nationals were buyers instead of whatever they actually were this year – and they managed to pry Manny Machado from the Orioles by outbidding the Dodgers. Obviously the Nationals were already the most appealing team between them and the Orioles in 2018, but they’ve just added a premier player for the remainder of the year. The Nationals’ product is even more appealing than it already was, while the Orioles become even less appealing. The Nats just took a considerable chunk of the O’s marginalized fans with that addition; there’s simply the local Manny Machado fans, the O’s fans who root for the Nationals as their NL team (the Nats fans who root for the O’s as their AL team left any thought of Camden Yards awhile back), the pissed off O’s fans looking for any reason or excuse to root for someone or something else, and maybe even those just trying to catch baseball at the highest level. Sure, those people are a small portion of gate revenue and merchandise, and there’s a reasonable chance they’ll be back at some point in the future, but you could’ve avoided those losses by not improving a competing product. By sending Machado to the Dodgers, the Orioles don’t directly improve the in-market competition; the Nationals may be relatively more appealing as a baseball product than they were prior to Machado leaving for LA simply because the Orioles got worse, but Orioles didn’t proactively improve the Nationals at their expense.

Performance Returns from Trades

Now for a little controversy – if we can even call it that. It’s hard to objectively evaluate trade returns for a couple reasons; the first is the obvious time factor, since trades for prospects who are years from the Majors are going to be skewed by a temporal bias until the prospects either debut, are traded, or leave baseball permanently. The second reason is based on the varying objectives between trade partners, which can be difficult to quantify for rentals who help a team win a championship or even boost them into the postseason, especially if they’re traded for prospects who contribute positively at the MLB level…it’s difficult to say a team lost a trade that helped them win the ultimate prize.

But, as it goes, we try to be objective with numbers and take note of the caveats. Table 7 shows us the WAR acquired and lost by each team since the 2011 World Series, and it’s sorted by the net WAR gained per trade. What’s interesting here is that the seldom-trading Giants actually lead all of baseball in net WAR gained per trade, adding nearly 1.2 Wins Above Replacement each deal. In other words, the Giants rarely make trades, but when they do, they get the most out of them – more so than anyone else. The Reds are the antithesis of the Giants’ economical trade behavior, giving away 1.05 WAR per deal. Like the Giants, the Reds rarely engage in trades, but unlike the Giants, they get the least out of them – a whole win worse than what we’d expect from a replacement level player we’d theoretically believe them to already possess.

The last 4 columns are split to show trades in two separate time periods; the 2011 World Series up to the 2014 World Series, and the 2014 World Series up to now. This is so we can get an idea of how much data in the middle columns should be attributed to trades that took place on a date far enough in the past that evaluating the outcome is appropriate, and how much of it is too recent to evaluate. Comparing the data specific to the respective time period only makes a handful of teams look noticeably different than what we can read into with the middle columns. We see a big discrepancy with the Tigers, in the last 3.5 years compared to the 2012-2014 period, which certainly represents a divergence in the on-field product, but this is nearly a swing of 90 WAR! By comparison, the A’s were good and bad in the same periods, and we see a 49.4 swing in Oakland. Perhaps worth noting is that the Tigers’ initial period was entirely Dave Dombrowski, and their latter period was about a quarter Dombrowski and the remainder Al Avila.

If I weren’t more cautious, I might label Dave Dombrowski a far better GM than I’d ever given him credit for; his Red Sox trades have been far more beneficial than the pre-Dombrowski Sox trades. And there’s no denying his Tigers trades look a lot better at the moment than Al Avila’s, even if the jury is still out on Avila’s Tigers moves and Dombrowski’s Sox moves. If anything, I might be willing to admit that Dombrowski is a good trader, and I’d even be cautious to make a deal with him because he’s gotten a lot more than he’s given up – at least since the 2011 World Series.

A final worthwhile look at Table 7 would be to the Gross WAR per Trade column (column 9). This one is interesting because it shows the average WAR changing hands per deal – we can look at it as the teams willing to make the biggest splashes, and perhaps even the teams more inclined to be involved in a “blockbuster” deal. If we draw the line at 4 gross WAR per trade, we see teams that often are involved in big and memorable deals: Arizona, Boston, Cincinnati, Detroit, and Oakland. Table 8 is more of an anecdotal re-hasher than anything else, but it adds some support to the idea that gross WAR per trade is a metric worth of our time.


There’s not a whole lot I’d like to put here, mainly because this was like 4 times longer than I expected it to be. But if anything is to be taken away from this, I’d hope that it would be the following points:

  • Jerry Dipoto trades a lot, but not as much as Erik Neander, the Rays’ GM.
  • The Mariners and Rays trade with each other more often than any other pair of teams.
  • The Nationals have a stronger preference to trade with the A’s than any other team prefers a trade partner; the 2nd-strongest is the Phillies’ preference to trade with the Dodgers.
  • Teams intentionally avoid trading with specific teams; the more at stake between the two teams, the more probable it is they’ll avoid trading with one another.
  • Most teams avoid trading in their division, but not all; about 8.4% of trades are intra-division.
  • Teams in the same market almost never trade with each other; 0.76% of trades made by 2-market teams are intra-market.
  • The Giants make fewer trades than anyone, but also acquire more WAR per trade than anyone.
  • Dave Dombrowski is probably pretty good at making trades.
  • The Dbacks, Red Sox, Reds, Tigers, and A’s have more talent changing hands on average than the rest of MLB.

First Post!

Welcome to my blog, Random Walks.  The name is kind of stupid but it’s a play on a statistics term (or I guess a math term, more generally) but it sounds like it could almost be referring to a baseball term sort of-maybe-ish.  Regardless of what you might think of the name, my intent in name the blog was to reflect my interest in baseball and stats, which is what this blog will be all about (for the most part)!

I spend a lot of time messing around with Fangraphs data, and I occasionally discover something that might be mildly interesting to other people, so I figured I’d blog about those cases, plus a few more.  I have a little bit of advanced training in statistics from when I was working on my Master’s in economics, but I’m not a PhD-holding data scientist by any means.  Still, baseball stats are among my favorite things in the world, so hopefully my insights provide some amount of usefulness to someone who finds this place!