Ottoneu 201: 2017 SP Replacement Levels

The subject of calculating replacement levels is a touchy one in ottoneu. Most people who play probably have some opinion on how this process should occur and, if you partake in any of the various ottoneu communities, you have probably witnessed some of the debates from both sides.

I tend to enjoy the process of determining replacement levels and, as I have gone through this process in the past, I wanted to take some time to divulge on my process as we near 2017 draft season. Before we do, we need to acknowledge some methodology or this will quickly turn into a debate of semantics. There are two ways that replacement levels can be defined. This is either as a specific point per game (P/G) or point per inning total (P/IP) or as the nth player ranked at a position. For example, I could say, replacement level for SP is about 4.00 P/IP, or I could say that replacement level for SP is about the 90th SP.

From my research in the past, I tend to lean heavily in the “nth player ranked” camp. I will unpack why that is, but for the purpose of this post, when I reference a replacement level (or change in replacement level) I am referring to nth ranked starting pitcher based on whatever projection system you decide to use in creating your own values. This could be Steamer, Zips, Pecota, or Pod Projections. Or it could be as simple as plugging in you own opinions of how a player will perform for the upcoming season.

Next, we need to look at how many innings can actually be pitched by each team’s starting pitchers. In ottoneu points leagues, each team has an innings cap of 1,500. However, a portion of this total will be allocated to relief pitchers. With 5 relief pitcher slots, most teams can accrue 250-325 innings from RP over the course of the season (50-65 innings per RP lineup slot). The remaining 1,175 to 1,250 will be pitched by starting pitchers.

While teams may certainly rebuild, either not rostering RP or not meeting innings caps, most teams will roster a full lineup out of the draft. While teams may decide their best course of action is to not meet the caps as the season progresses, it is not common for teams to plan to do this prior to actually drafting their  team. Because of this, I do not believe the impact of rebuilding teams on these replacement levels to be large. The long term goal is to use these levels to set draft prices. In total, we are looking for something between 14,100 and 15,000 SP innings for the league.

In 2014, I calculated replacement levels for every position across the ottoneu universe. This was a lot of work, but I did this to help calculate better dollar values as I prepared for my upcoming auctions. At the time, looked at 1, 2, 3, 4, 5, 10, 20, and “All-Time.” (The “All-Time” data set is actually 28 years, as 1986 was the first year holds was tracked). So, for example, year 1 included 2014, the 2 year data set included 2013-2014, while the 3 year data set included 2012-2014. In doing this, I found that replacement levels over multiple year data sets were very sticky. Certainly over any one individual season, the appropriate replacement levels could produce more variation, but over multi-year periods, the historical replacement levels from 2014 were pretty consistent.

As you can see, I have included the Replacement Level P/IP for completeness, but what I would focus on is the nth player ranked methodology in the yellow column. More on this to follow.

Some notes on assumptions. To determine these totals, I removed every pitcher who did not start 10 games in each individual season and ranked each data set by P/IP. Then I looked at how many SP it would take to meet the needed by SP for the league (as determined above). At the bottom I have included a 3 year weighted average, where the most recent data at the time was given the highest weight. The less essential P/IP number corresponds the the P/IP produced by nth ranked SP.

Recently leagues have needed about 85-95 SP to meet their allotted innings caps, or 7-8 SP per team. While I didn’t show this range in my initial calculation, I will be showing a range in the future to account for varying RP usage assumptions (if you really don’t believe rebuilding teams will pitch RP innings, feel free to use the high end of my range provided. Some criticism could be made that historical data should not be used to determine current year replacement levels, and if you were determining replacement level based off of a set P/IP, then I would agree. However, you can see that the number of SP needed within a league to meet their allotted innings caps is relatively stable (changing by ~3 when looking at 1 to 5 year data sets).

The most frequent question I receive on this topic centers around what current replacement levels would be. 2014 data still leaves us excluding both the 2015 and 2016 seasons and much can be made about changing league environments impacting replacement levels. My blanket response tends to be that there isn’t much justification to think that replacement levels, as a whole, have dramatically shifted. However, I need to acknowledge a major caveat and a point of clarification/misunderstanding. When I say replacement levels have not shifted, I am speaking to the number of players needed across leagues to meet a game cap. It is my fault for not clarifying this enough in the past. Certainly, if you view replacement level as a specific P/G or P/IP, we can expect wide variation year to year, and smaller variations with multi-year data sets.

This is the main reason why using an Nth player ranked methodology to produce a corresponding P/IP (and later P/G) metric is important. Many people make their own alterations to projection systems (we can argue the merits of this separately). Due to the different methods of determining expected 2017 player performance, it wouldn’t be feasible for anyone to produce a P/IP or P/G metric that was a fit for every iteration of expected 2017 performance. However, I can confidently give you a less context dependent number. You should use the Nth player ranked methodology to determine your own P/IP based off your values. For completeness, I have updated replacement levels for the most recent data available.

For reference, the 1 year replacement level for 2014 was the 85-88 ranked SP. For 2015, the replacement levels were the 88-92 ranked SP. Ranges are good! Unlike in 2014, I have included a High and Low assumption which accounts for the varying assumptions you could make on how many innings relief pitchers will account for within a league. If specifically looking at P/IP, you will note a drop from the 2014 levels. However, we are still looking at around 85-95 SP per league. If you wanted to be more specific, I would use the 3 year average, and increase that by 1 per year on the high and low end if starters continue the trend of pitching less innings.

As you build your dollar values for 2017, you can expect replacement level to be near the 90th SP. Using whichever projection system you prefer, I would use the corresponding player (or range of players) as your baseline . Next Friday, I will look at relief pitcher, following up with position players in the next few weeks. As always, if you have any questions on the process feel free to leave a comment.