[CORRECTION: I incorrectly state in the article below that Trey Baughn had written the article that I reference regarding replacement level — in fact, it has been brought to my attention that @ottoneutrades is the author. Completely my mistake, I apologize to @ottoneutrades and @fazeorange!]
Ottoneu offseason is in full swing – we are approximately half way between the end of arbitration and the cut down deadline (where bad contracts can be released without incurring the 50% penalty). Our league is in a bit of a lull – from what I have seen, owners have picked through the rosters of active owners and identified the players they want and cut deals as appropriate.
Our league’s commissioner, Trey Baughn (@fazeorange) (who also produces a podcast that it worth following if you aren’t already) recently posted some thoughts on the message board regarding replacement level at SP. Specifically, he identified 4.24 P/IP as the approximate replacement level for starting pitching. Perhaps more interesting than the actual number (which I will discuss below) is the methodology in arriving at that calculation. Here is Trey’s post on how he arrived at that number. It is a very rigorous analysis that he has applied over multiple years, trying to hone in on exactly what teams expect to derive from their starting rotation during the course of a competitive season.
The full post is extremely worthy of your time and an excellent discussion of the topic – but to summarize at a high level, Trey works through the numbers and demonstrates that, given Ottoneu scoring systems and lineup restrictions, a team will target 1,175 IP from its staff (and 300 from its bullpen) – which at a gut level seems to be exactly right, given the limit of 5 RPs per day that can be started. The analysis then moves to translating that number to starting pitchers – 50 years ago, you might reasonably get 1,175 innings from 5 starters, but in this age of bullpen specialization and 5 man rotations, the analysis pegs the SP number at just above 7 for an Ottoneu team. Like the total inning number, this does confirm a gut feel – Ottoneu owners that go into a season contending for a championship certainly can’t hope to ride 4-5 starters throughout the season given the dearth of 200 IP workhorses.
So far so good. But this is the point where @fazeorange and I diverge as the assumptions that form the basis of the analysis prove incorrect. Taking the 1,175 inning calculation and applying it to standard 12 team Ottoneu leagues requires 14,100 IPs from starting pitchers – which in 2014 meant that 86 SPs filled those 14,100 innings leaving Ryan Vogelsong as our “replacement level” pitcher, at 4.24 P/IP.
The problem is twofold. First and foremost, this assumption assumes that there are 12 teams trying to throw 1500 innings and compete for a championship. Given Ottoneu’s year round structure, in-season loans and arbitration processes, this never happens. In our league, 5 teams threw fewer than 1300 innings. At any time, there are a number of teams trying to compete and another number rebuilding for the next year or a few years down the road. Like real baseball, typically championship teams have accumulated salaries that will need to be traded or cut, and while savvy owners should be able to build toward a window that stays open for multiple years, the format makes it difficult to stay on top unless an outside event (i.e., primarily less than competitive other owners) intervenes.
Using Trey’s methodology of 1,175 innings per team, below is a table that shows how replacement level changes as you reduce the assumption regarding the number of teams that are trying to throw their full complement of innings:
|Total IP (approx)||14100||12925||11750||10575||9400||8225||7050||5875||4700||3525|
|# of SPs||86||79||72||64||55||49||42||34||27||20|
|P/IP (Ps Above Replacement||5.01||5.08||5.14||5.23||5.32||5.39||5.47||5.57||5.66||5.80|
As you would expect, replacement level rises as the number of “competitive” teams decreases. Over our league’s years, teams each Spring can be roughly broken into one of three groups – those teams clearly well positioned to compete for a championship, teams clearly in some stage of a rebuild not attempting to field a competitive roster for that upcoming season, and a third group of teams that are likely deluding themselves into thinking that they can compete (which I personally think is the worst position to be in – trapped in a middle ground that will never win and never commit to a full roster reconstruct, dooming the team to perpetual 5th-8th place finishes). Also interesting is the “P/IP (Ps Above Replacement)” line – this is the average production per inning for those pitchers above the applicable replacement line – i.e., what good starting rotations are producing.
As you can see from the table, if you assume only 3 teams are full rebuilding (which in my mind is low, but probably constitutes the number in our league last year that knew they were not in the running for 2014 at the end of January), replacement level rises to 4.50 using this methodology. If you include those teams that are likely kidding themselves, replacement level rises to nearly 5/IP.
The second issue is temporal – as most Ottoneu owners have likely witnessed, “replacement level” changes over time. Now, to be fair, this concept is different from a traditional definition of replacement level as that level of play you could fairly assume could be acquired off the waiver wire. However, the idea is that, especially in Ottoneu Linear Weights/Points formats, teams begin dropping out of competition almost immediately after the season begins (whether due to injury, mis-assessment of the strength of the “going for it” teams, or their own rosters). Because Ottoneu rules permit eating salary in trades, owners with productive but expensive starters start dumping them for whatever they can convince themselves might be useful in future years. Again looking at our league in 2014, here are a few of the early transactions:
(1) Mid March – rebuilding team dumps Mike Trout for a prospect package centered on Gerrit Cole and Javier Baez
(3) On April 7th, Hiroki Kuroda and Rafael Soriano were moved for two two prospects.
(4) On April 12th, the team that had added Mike Trout also added Madison Bumgarner and Jose Bautista for 3 prospects
(5) On May 2nd, the team with Trout/Bumgarner/Bautista added Evan Longoria and Ian Desmond for two younger but upside players
So, before the season was a month old, one team added Mike Trout, Evan Longoria, Madison Bumgarner, Jose Bautista and Ian Desmond. Not coincidentally, this team (@fazeorange’s team) won the 2014 championship. The point here is not to criticize – he won fair and square. In addition, a useful rebuilding strategy for some teams is to buy overpriced assets at auction with the intent to auction them prior to or early in the season, eating their salary but acquiring more useful future assets. But it does illustrate the point here that players in Ottoneu are “available” (via salary dump trades) in a similar, though not exactly analogous, way to replacement players at the MLB level are available.
Brining this discussion back to replacement level – given the nature of Ottoneu’s format with a year round system and rebuilding teams, I think the numbers show that defining replacement level at the 4.24 P/IP underestimates the impact of the rebuilds and the availability of talent over time. Looking at the list of 2014 SP scoring (min 10 Starts), if we assume 9 teams and the table above, our replacement level starter is Hiroki Kuroda; with 7 teams, Scott Kazmir becomes replacement level. While neither are aces, their production levels are significantly better than that of a Ryan Vogelsong, and anecdotally neither would seem to command a king’s ransom to acquire. As I look at competitive 2015 teams, I would be starting here for my replacement level.