Baseball ProGUESTus: Trade WARP: Evaluating MLB General Managers

Most of our writers didn't enter the world sporting an @baseballprospectus.com address; with a few exceptions, they started out somewhere else. In an effort to up your reading pleasure while tipping our caps to some of the most illuminating work being done elsewhere on the internet, we'll be yielding the stage once a week to the best and brightest baseball writers, researchers, and thinkers from outside of the BP umbrella. If you'd like to nominate a guest contributor (including yourself), please drop us a line.

Tim Malone is an ex-engineer/ex-Angeleno with degrees from UCLA and UW in Bioengineering and Biostatistics who lives in the Seattle area with his family. His likes include coconut, peanut butter, tools, and conversations with strangers. His dislikes include runny eggs, bad champagne, mindless repetition, and conversations with strangers. The return of Major League Baseball to Seattle warms his soul.

The emergence of the Wins Above Replacement Player framework as a measure of player performance has greatly facilitated the evaluation of many activities of major-league general managers, including trading. WARP appears to be a stable surrogate for player value, and WARP and its salary cost ($/WARP in millions) serve as a means of evaluating transactions in retrospect.

Some instances of WARP-based trade valuation exist in the MLB blogosphere. In 2008, Dave Cameron of USS Mariner did a prospective analysis of the notorious Erik Bedard for Adam Jones et al. trade implicitly based on WAR/WARP, which proved depressingly prescient. Recently, the trades of Frank Wren were analyzed from a net-WAR perspective. Global evaluation of GMs has been addressed by Baseball Prospectus’ Shawn Hoffman and Rebecca Glass, among others.

WARP via trade (trWARP) used here is defined as the cumulative player WARP accrued by a franchise (or its partner) consequent to the trade, but attributable to the year of the trade. More specifically, trWARP accrues as of the trade date and continues until the player attains free agency, is traded again, or retires/is released. Thus, a GM’s “WARP report card” illustrated here will not be complete until the last player traded under his regime attains one of these states. (WARP accrued in arbitration years subsequent to the trade is included in trWARP. A subsequent analysis might exclude WARP from arbitration years, which would almost certainly yield more extreme $/WARP values.)

Trade WARP is one component of the three-legged GM stool, along with WARP from free agent (faWARP) and draft (drWARP) activities.

• Unlike the draft or free agent cases, a trade can be directly evaluated by summing up the net WARP exchanged (Received – Given). In a chosen time period, or exclusively within a given cohort, aggregate WARP given equals WARP received. This zero-sum property enables double entry checking and analysis of net WARP exchanged within these closed systems. All else being equal, positive net trWARP is good. Also, trading between MLB GMs is a more “fair” competition than free agency moves, which favor rich franchises, or than drafting, which favors recently unsuccessful ones. Unless otherwise specified, all WARP described in this article is trWARP.
• Similar to trWARP, faWARP can be evaluated immediately after the transaction, but free agent market efficiencies would certainly yield fewer $/WARP outliers, and faWARP relative to a period standard should be used, as opposed to the implicitly zero-referenced trWARP.
• Although draft WARP efficiency differentials undoubtedly persist between franchises/GMs, analysis is complicated in that the drWARP report card is delayed by the prospect incubation period.

Data Sources & Assumptions

• Transaction Data: Retrosheet
• Annual Player WARP: Baseball Prospectus
• Annual Player Salary: Baseball-Reference
• Name Convention Reconciliation: Lahman
• GM Tenures: Input from public sources (i.e., Wikipedia)
• Team Payroll Data (since 2000) from Cot’s Contracts

This analysis includes trades between January 1, 1990 and the end of the 2012 season, and focuses on GMs with three or more years with their current teams as of the end of the 2012 season. Although WARP in this period is essentially time-invariant, the salary cost per WARP ($/WARP in millions) is not, and has been increasing about five percent per year over the past 10 years, to its current level of approximately $5M-$5.5M for free agent transactions. In this article, unless otherwise specified, the $/WARP values for each GM have been adjusted to 2012 levels, assuming the five-percent annual increase. Analysis of salary data is further complicated by arbitration-driven contract resets in which free agency is avoided. As previously mentioned, I am unaware of a database containing individual player contract status beyond the past few years (Cot’s has individual MLB player contracts from 2006).

Results and Discussion

Table 1. Active GMs (as of 11/01/2012), sorted by net consequent trWARP/Year (descending)

[1] As of end of 2012 season

[2] The GM’s Franchise

[3] All of the Franchise GM’s Trading Partners

[4] $Millions/trWARP—adjusted to 2012 salary levels

[5] $Millions/Yr over tenure—adjusted to 2012 salary levels

Net Trading Economy and Efficiency

Andrew Friedman is alone in the upper-right quadrant, at the top in net $/WARP and near the leader (Dombrowski) in net WARP/yr, who pays out $2.5M more for WARP than his partners. Moore, Cashman, and especially Huntington occupy the opposite quadrant. In particular, Huntington and Moore aren’t doing their respective small-market franchises any favors in losing the $/WARP exchange. The patterns of Anthopoulos and Hill, as net WARP losers but $/WARP winners, however, are consistent with a budget-constrained franchise being unable or unwilling to retain more expensive talent.

Transaction Efficiency and Frequency

There is a weak inverse relationship between transaction frequency and net WARP/yr that is not statistically significant (p=.08) at the p=.05 level. This is consistent with the intuitive notion that skilled traders are more selective deal-makers. This cohort’s average number of transactions is 6.4 per year, with the Rockies’ Dan O’Dowd the most active at 8.9 per year and the Cardinals’ John Mozeliak the least active at 3.2 per year.

In their first years, these GMs average 9.3 transactions, compared to 6.1 in subsequent years. This early roster reconfiguration is even more striking, considering that the beginning of the first year is an acclimatization period in which assessments should precede personnel moves.

Payroll Constraints

Not surprisingly, trading economy is influenced by team payroll, as the more affluent teams are more willing/able to take on more expensive trWARP. The trend is visually indicated in Figure 4 above. For the entire eligible cohort of 20 GMs, the regression’s p-value is .09, but if the “outlier” Huntington is removed, the fit is significant at p<.02. Conversely, Figure 5 indicates that there is no statistical relationship between the cost of outbound WARP and team payroll. On both Figures 4 and 5, the average $/WARP ($3.1M) and average 2012-adjusted payroll ($98M) are shown. The $3.1M per WARP acquired via trade is significantly cheaper than the accepted value of open market faWARP (over $5M).

Active GMs are better traders than the GMs they replaced
The current cohort of GMs (including those with less than three years with their current franchise) netted +75.6 trWARP—calculated as received minus given. By the zero-sum property, all previous cohorts evaluated since 1990 amassed -75.6 trWARP. Perhaps the current cohort is more “saber aware” than its predecessors. Only four of the previous cohort’s 30 GMs (Dipoto, Alderson, Hoyer, and Jocketty) were employed as GMs at the end of 2012, and none has been an exceptional trader. The cohort immediately preceding accrued -42.2 trWARP. This result might be misleading—if Mariners GM Bill Bavasi’s misadventures (-113.5) are excluded, the previous cohort netted +71.3 trWARP. It should also be noted that of the net +75.6 trWARP amassed by the current cohort, 69 percent, or +52.0 trWARP, was attributable to its trades with Bavasi (more later).

trWARP Depletion and Replenishment
How quickly can GMs reasonably expect to accumulate WARP via trade? Of current GMs with more than two years' experience, three are very high performers in terms of trWARP/Year—Dombrowski (DET) +5.3, Amaro (PHI) +5.3. and Friedman (TBA) +4.7. All else being equal, a trade-savvy GM inheriting a mediocre roster can expect a playoff contender no sooner than a two-year time frame via trade activities. Rounding out the bottom of the list: Huntington (PIT) -5.8, Moore (KCA) -5.0, Cashman (NYA) -3.2, Hill (MIA) -2.7, and Anthopoulos (TOR) -2.6.

Individual GM Snapshots

The numbers listed after each GM follow Table 1

• (trWARP/Yr Franchise:$M/WARP Franchise) vs. (trWARP/Yr Partners: $M/WARP Partners).
• $M/WARP and payroll values are 2012-adjusted, assuming a 5% annual increase.

The Good

Tigers — Dave Dombrowski (+9.3:$4.3M) vs (+4.0:$1.8M)

No GM has been a more consistent trWARP winner than Detroit’s Dave Dombrowski. Over an 11-year span, he has averaged +5.3 WARP/year, with multiple years of capturing significant +WARP while parting with little in return. In 2004, he was +22.0 WARP (+21.8 Franchise vs -0.2 Partners) courtesy of Carlos Guillen via Bill Bavasi. In 2005, he was +15.2 (+15.3 vs 0.1), largely due to Placido Polanco at +11.1. In 2007, +9.8 (+26.1 vs +16.3), scoring huge on Miguel Cabrera’s +26.6, while parting with Jair Jurrjens’ +10.9. In 2009, his trades yielded +8.8 (+19.2 vs +10.4), netting Austin Jackson (+11.2) and Max Scherzer (+6.6) while parting with Curtis Granderson (+11.3). It is interesting, however, that Dombrowski’s $/WARP differential (-$2.5M) is among the cohort’s worst. His partners have benefited from very inexpensive trWARP ($1.8M)—only Jocketty ($1.5M) and Daniels ($1.7M) have parted with cheaper WARP. The Tigers have made the playoffs three times during Dombrowski’s 11-year stint, and twice over the last five. The average Tigers payroll of $115M exceeds the MLB average of $98M, and allows Dombrowski to pay $2.5M more per win than his trading partners.

Phillies — Ruben Amaro, Jr. (+6.5:$3.5M) vs (+1.2:$4.2M)

Amaro’s trWARP story is dominated by the acquisition of Roy Halladay from Toronto in 2010, yielding +11.6 WARP over three years. Including his injury-truncated 2012, Halladay’s WARP came in around the free agent market cost (2012 dollars) of $58.3M or $5.0M/WARP. In addition, Roy Oswalt contributed +3.9 WARP and Hunter Pence +3.1. Trading partners might think twice before consummating trades with the current Phillies GM, as they acquire a bottom-of-the-cohort +1.2 WARP/yr (at $4.2M) compared to Amaro’s +6.6 WARP/yr (at $3.5M). The Phillies have been a big-payroll organization during Amaro’s four-year tenure, averaging $160M, and a successful one, averaging 93 wins and reaching the playoffs in three of the last four years.

Rays — Andrew Friedman (+8.7:$1.3M) vs (+4.0:$3.7M)

From Figure 1, Friedman stands alone in combining trading efficiency (+4.7 WARP/yr) and economy (+$2.4M net $/WARP). Friedman’s first two active trade years, 2006 and 2007, were outstanding. In 2006 he was +20.1 (+24.7 vs Partners +4.6), due largely to the contributions of Ben Zobrist at +18.5. The following season, 2007, was also productive, yielding +10.7 WARP(+19.1 vs +8.4), netting Jason Bartlett (+9.1) and Matt Garza (+3.7) along with Grant Balfour (+4.5). Friedman’s trading partners haven’t fared well from a $/WARP standpoint, paying $3.7M per win vs Friedman’s $1.3M. Friedman has averaged 92 wins over the last five years of his seven-year reign, making the playoffs in 2010 and 2011, all with a bottom-three MLB payroll averaging only $56M.

Braves — Frank Wren (+8.1:$2.6M) vs (+4.9:$3.3M)

Overall, Frank Wren has been a solid trader over his five-year tenure with the Braves, beginning in 2007. That first year yielded +16.2 WARP (16.7 vs. 0.5), featuring Jair Jurrjens via Dave Dombrowski (+10.1 WARP at $10.5M), and Omar Infante from the Cubs’ Jim Hendry (+4.3 at $6.3M). Since 2007, however, he has only broken even: +24.1 vs +24.4. The Braves have averaged 86 wins and $101M over the last five years, making the playoffs in 2010.

Athletics – Billy Beane (+11.2:$2.6M) vs (+12.8:$3.1M)
Giants – Brian Sabean (+8.1:$2.6M) vs (+5.5:$3.1M)

While Billy Beane is nearly a wash in terms of net trWARP over his 15-year stint, Mr. Moneyball predictably shines on the bang-for-buck side, paying $2.6/WARP while trading away WARP averaging $3.1M in 2012 dollars. Beane’s long-time counterpart across the bay is Brian Sabean, who began his tenure after the 1996 season, a year before Beane. Sabean has the edge in net WARP/year (Beane at -1.6, Sabean +2.5), but in terms of WARP cost, Beane and Sabean are clones, each paying $2.6M per win compared to their $3.1M for partners. Despite different payroll resources, both teams have been successful. The A’s have made the playoffs six times in 15 years, with a well-below-league-average payroll of $72M over that span. The Giants have won two of the last three World Series and made the playoffs four other times in 16 years, averaging $115M in payroll.

The Bad

Pirates — Neal Huntington (+2.9:$5.3M) vs (+8.7:$2.5M)

Huntington is a payroll-constrained, small-market operator; even so, his trade exploits have not helped the franchise. He has overpaid for the Pirates’ 2.9 WARP/yr received at $5.3M/WARP, and parted with 8.7 WARP/Yr costing a below-market $2.5M. He was bad in his first really active year, 2008, (+1.0 vs +29.1) parting with Jose Bautista’s +21.3 and Jason Bay’s +7.2. On the Pirates' side of the ledger, he was done in by the $26.5M 2012 combined salaries of A.J. Burnett and Wandy Rodriguez, resulting from 2011 activity. These two combined for only +4.2 WARP. Pittsburgh’s dead-last $53M payroll has yielded an average of 67 wins over Huntington’s tenure, with predictably playoff-free results (until 2013, perhaps).

Royals — Dayton Moore (+2.0:$4.4M) vs (+7.0:$2.3M)

Under Moore, the small-market Royals, like Huntington’s Pirates, could be doing better on the trading front. Losing 5.0 WARP/year doesn’t help, nor does paying $2.1M more per WARP than your trading partners. To be fair, Moore’s 2012-adjusted average payroll of $69M is 23rd of 30 in his cohort, which both forces and precludes many trades. On the negative side, in 2008, he parted with Jorge de la Rosa’s 6.6 WARP at $8.5M, and more painfully, in 2010, with Alberto Callaspo’s 7.4 WARP costing only $5.2M. In 2010, the Royals couldn’t hold on to Zack Greinke’s 7.6 WARP, which cost the Brewers $27.6M, and in 2011, Melky Cabrera and his 5.4 2012 WARP went to the Giants for only $6M.

The Ugly

Mariners — Bill Bavasi (+0.6.1:$20.0M) vs (+25.3:$2.1M)

Although he's not a member of the class of current GMs, some of you might be curious about how the perceived poster boy/deceased equine for GM ineptitude fares here. The Seattle Times’ Larry Stone evaluated the trades of Bill Bavasi qualitatively and concluded that “…a lot of deals that backfired, and not much quality in return for the Mariners.” No big surprises here—the numbers support his conclusions. I double-checked the WARP and $ numbers above for transposition errors, and yes, Bavasi hemorrhaged a net -24.7 WARP/year, -113.5 WARP total over four-and-a-half years of unchained franchise destruction. Although frustrated Mariners fans tend to focus on the well-chronicled talent basket that Bavasi traded away (Asdrubal Cabrera, Shin-Soo Choo, Matt Thornton, Carlos Guillen, Randy Winn, Adam Jones, et al.), the absolute dearth of WARP received is perhaps more striking. In exchange for 116.5 departed WARP, Bavasi received a negligible 3.0 WARP, the total payroll cost of which was $60M in 2012 dollars. This equates to a preposterous cost of $20.0M/WARP to the Mariners. Compounding the area’s offseason grey gloom, most of this 116.5 WARP willingly transferred to his trading partners was significantly more cost-controlled than the $3.1M average value of the current cohort—$2.1M/WARP in 2012 dollars. In actual numbers, Bavasi’s average payroll over his 2004-2008 stint was $96M—well above the MLB mean of $83M over that period. Despite these financial resources, the M’s averaged 72 wins, leaving Safeco to the seagulls after September.

You knew the job was dangerous when you took it

Mariners — Jack Zduriencik (+5.0:$5.0M) vs (+6.5:$2.3M)

In addition to drafting and free agent misadventures, it is well accepted that Jack Zduriencik was put in a deep talent hole by his predecessor’s trade escapades. In Jack Z’s first four years, his trading hasn’t aided the M’s recovery, having lost 1.5 WARP/year. Like Frank Wren, he started on a very positive note, and like Wren, he hasn’t followed up that initial success. Since the blockbuster three-way deal that landed Franklin Gutierrez, Jason Vargas, et al. (+13.0 WARP) for J.J. Putz, et al. (-1.0 WARP), Zduriencik’s trades have lost nearly 20 WARP (7.1 in, 27.0 out). Compared to Wren, Jack Z pays out $2.4M more per win ($5.0M vs $2.6M) while partners pay $1M less ($2.3M vs $3.3M). It should be noted that Zduriencik has often traded for defensively oriented players, whose contributions are valued differently by the three major WAR(P) variants. In the Zduriencik era the Mariners’ payroll has averaged $99M—essentially league average. In the near term, the Mariners’ recovery hinges on out-trading and out-drafting their competition.

Good and Bad

Marlins — Michael Hill (+5.9:$1.7M) vs (+8.6:$4.6M)

Michael Hill’s tenure with the Marlins has been interesting. Although he has shed 2.7 net WARP/Year, his WARP cost of $1.7M is a fraction of his partners’ $4.6M. The boom-and-bust Marlins are difficult to analyze consistently, with either great teams (World Series wins in 1997 and 2003) or uncompetitive ones (73 win average excluding those two years). Over the last five years, the Marlins have averaged 78 wins with a paltry $62M average payroll. Hill’s performance makes more sense when viewed through the lens of a payroll-constrained organization.

Blue Jays — Alex Anthopoulos (+7.3:$1.9M) vs (+10.0,$4.8M)

Like Michael Hill, Alex Anthopoulos has done well on salary and not so well on talent in his three years. His trades have resulted in a deficit of 2.7 WARP/year, but like Hill, his WARP has been inexpensive—costing only $1.9M, compared to his partners’ $4.8M. A good fraction of Anthopoulos’s action came in the 2009 blockbuster with the Phillies and Mariners, which has so far yielded little and cost the Jays Roy Halladay’s aforementioned +11.6 WARP. A week later, the Jays and M’s swapped Brandons (League for Morrow), a deal in which Toronto came out +8 WARP ahead (+8.1 versus +0.1). Many have speculated that these two trades were in fact a single trade separated by seven days. The Blue Jays haven’t made the playoffs in recent history, and have averaged an $85M payroll and 80 wins over the last five years, which hasn’t allowed them to compete effectively in the AL East, with its rich and/or smart GMs. Their 2013 payroll, however, has spiked, with disappointing results.

In a League of His Own

Yankees — Brian Cashman (+8.3:$4.6M) vs (+11.4:$2.9M)

Cashman has not been a successful trader, losing 3.2 WARP a year over the same 15-year period as Billy Beane. The two began their stints within three months of each other —Beane in November of 1997 and Cashman in February of 1998. Unlike Beane, however, over that lengthy span, Cashman paid an above-market $4.6M for his WARP versus $2.9M paid by his trading partners and $2.6M for Beane. Since 2000, the average MLB payroll has been $104M and the Yankees’ has been $229M, so Cashman can and has made up trading deficits via free agency. It’s nice to be Daddy WARbucks.

Conclusion
WARP via trade is an important determinant of GM performance and subsequent franchise success. By necessity, this article (over)-attributes all trading activities during a GM’s reign to that GM, as opposed to the entire front office/ownership. Also, and obviously, each franchise perceives a unique comparative advantage. Trade WARP in a vacuum is a blunt tool for GM evaluation, as WARP and its dollar cost are valued differently by contenders and rebuilders and rich and poor franchises. At a minimum, I would have confidence in distinguishing the extremely strong traders (Dombrowski, Friedman, Amaro) from the very weak ones (Bavasi, Cashman, Moore, Huntington). Positional valuations and franchise-specific gaps/redundancies are also factors. Additionally and often unfortunately, upper-management directives and the GM’s own contract status can apparently have a significant impact. Some of these factors might provide for fascinating future studies.