Calculating Ottoneu Replacement Level: Starting Pitchers

When evaluating players in Ottoneu, I typically find total points to be misleading. They tell a part of the story, but are they the most important factor in determining a player (or team’s) performance? Is R.A. Dickey at 862 points more valuable than James Paxton at 383 points (assuming similar salaries)? I believe Points Per Game (PPG) and Points Per Inning Pitched (PPIP) to be much more indicative of a player’s value. For this study, I have examined all starting pitchers seasons from 1986 (the year Holds were invented) through 2014.  I only considered pitchers who made 10 or more starts in a season and have ranked these pitchers by PPIP. Using this data set, my goal was to determine replacement level for starting pitchers in terms of PPIP for Ottoneu linear weights (FGPoints).

A couple questions need to be answered:

1. How many innings do Starting Pitchers “pitch”?

Given the 1,500 IP cap that each team is subject to, it is important to consider how many innings (of these 1,500) that starting pitchers will actually account for. However, these innings will be split between the 10 active pitcher spots on your roster (5 SP spots, and 5 RP spots). I calculated the number of innings each RP spot produces with a little guess work. I feel pretty confident that the average team gets between 60 and 70 innings out of each RP slot every year (the top-60 RP in 2014 averaged 58 IP). I split the difference between the two at 65 IP per spot.  This amount can certainly be debated, but the overall impact it will have in the final results is minimal. (Using 60 or 70 IP per RP slot would only result in a .01 increase or decrease in replacement level PPIP). However, with this reasonable guess of how many innings an RP spot produces, we can easily determine the total innings needed from each of the 5 SP spots in a lineup:

SP Per team formula


The key takeaway: Of the 18,000 total innings available in a 12 team Ottoneu league, roughly 14,100 innings need to be “pitched” from starting pitchers, or just over 78% per season.

2. How many SP are owned in a league to produce 14,100 innings?

Ranking all SP in order by PPIP for a given season, I looked at how many SP it took to pass the 14,100 innings needed for the league. This is better expressed with an example. In 2014 Clayton Kershaw was the number 1 ranked SP by PPIP and pitched 198.33 IP.  Jake Arrieta was number 2 and pitched 156.67 IP. Together they pitched 355 innings. I repeated this process (adding in the 3rd, 4th, 5th, etc. ranked SP) until the total innings pitched by the top ranked pitchers passed 14,100 innings. For 2014, it took 86 starting pitchers (min. 10 GS) to pass 14,100 innings, or 7.17 SP per team. Here is the number of SP needed by each team to meet the 1,175 SP team inning estimate, beginning with 2014 as “1YR SP” (“All Time” is 1986-2014):

Formula 2

SP per team

3. What are some historical PPIP values?

Historical PPIP

In 2014 86 pitchers were needed to pass the 14,100 innings needed by all starting pitchers. The replacement level for 2014 (1YR) in the above chart is the PPIP associated with the 86th ranked SP by PPIP. In 2014 this pitcher was Ryan Vogelsong with 4.24 PPIP. If you’re thinking, “I wouldn’t start Ryan Vogelsong!” you’re probably right. It makes sense that a team’s 7th SP would be someone who was only started occasionally, and is likely owned for a replacement price of $1-$3. Several other SP who fit this description include: Henderson Alvarez, Roenis Elias, Chris Tillman, and Jon Niese. Keep in mind, the 2014 replacement level should not be used for all valuations going forward. My personal preference is a weighted average of the last three years (~ 4.20), but this should give everyone an idea of historical replacement level. Feel free to choose what you believe to be best.

One note on the  chart:  Each specific row of the data (2YR, 3YR, 4YR, etc.) is not referring to one specific season.  The row labeled “1YR” is specifically considering 2014. However, the “3YR” row is examining all SP seasons with 10 GS from 2012-2014 (with the innings threshold being 14,100 * 3 seasons). Similarly, the “20YR” row is examining all SP seasons from 1995-2014 (with the innings threshold being 14,100 * 20 seasons).  I did this because 1 year trends (while interesting) likely do not provide much use to the general reader – I don’t imagine anyone reading this is thinking “I want to use 2010’s replacement level for my valuations.”  However, I could easily see replacement level since 2010 being useful (in which case you would look at the 5YR row.) Each “YR” row is simply stating how many years of data are being included.

Combining the last two points:

I am making the assumption that the 7.17 SP owned would be pitched every time they start in order to meet the 1,175 threshold. This does not happen in reality as teams typically own more than the minimum ~7 SP needed. I own more than 7 starting pitchers on each of my squads (around 10 in most cases), but in each case the last 2 or 3 SP considered are typically used as spot starters. While I may prefer some of my spot starters to options available in free agency, comparable options are typically available.  Given that streaming SP is difficult in Ottoneu, it makes sense that more SP than the minimum necessary would be owned by the majority of teams. I do not see this as an indictment of the process of determining replacement level, but rather an acknowledgement that several replacement level type SP will be owned by most teams to serve as spot starters and injury depth. This can be seen in the following graph:

2014 Pitcher PPIP

Replacement level is shown by the horizontal black line that bisects the graph. 86 SP fall below that line.  However, an additional 23 SP fall between 4.24PPIP and 4.20PPIP. In theory, if these 23 SP were owned in a league, 109 SP would owned – or just over ~9 per team.  This meshes pretty well with what I have seen in other leagues – each team owns about 7 SP that they value, while they probably feel they could replace their last 2 or 3 SP pretty easily.

Throughout the rest of the offseason, I will be calculating replacement level for other positions. I hope to produce values based on these replacement level values as well. In the mean time these calculations should serve as a barometer to use when evaluating starting pitchers. It does not tell the whole story but it is a very simple way to determine a starting pitcher’s value when discussing trades and planning your 2015 roster. Feel free to use any of the replacement levels above in your own valuations. If you have any questions about the methodology, feel free to comment.


11 thoughts on “Calculating Ottoneu Replacement Level: Starting Pitchers

  1. chike55 says:

    Great work. I was able to follow along with most of the process, but I’m having a hard time understanding what 4.24 PPIP looks like. (I play roto almost exclusively.) I know what Henderson Alvarez is and I know what Ryan Vogelsong is. They’re both poor pitchers, but in very different ways. Is there maybe a generic statistical profile of a pitcher who is an average of the replacement level pitchers?

    Also, how can I apply this information to my league and my team?

    Again, great work.

    • There really isn’t a generic statistical profile of a 4.24 Pts/IP starting pitcher; two pitchers could have wildly different peripherals (FIP, etc) and post the same Pts/IP. As an example:

      Pitcher A 200.0 IP 222 H 25 HR 53 BB 3 HBP 175 K (3.55 FIP)
      Pitcher B 192.7 IP 166 H 17 HR 71 BB 14 HBP 140 K (4.15 FIP)

      Both of these pitchers scored about 4.19-4.20 Pts/IP (or just under replacement level), but one benefitted from BABIP and some HR/FB luck, and the other was unlucky in both aspects. Pitcher A is Brandon McCarthy who just signed a $12 million/year contract with the Dodgers, and Pitcher B is Edinson Volquez and is still a free agent.

  2. […] 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 […]

  3. fazeorange says:

    Here’s a great critique (by Bill Porter) of the replacement value method outlined above. Once I have more time to review and crunch some numbers I will respond and see if an adjustment to this method might be necessary due to the competition drop off than Bill identifies here. Well worth a read.

    • While I agree with most of what Bill says in that critique, I don’t agree with the idea that those criticisms require an upward adjustment to the replacement level calculation. This is a fascinating issue to discuss, but to me the entire point of thinking about and calculating a replacement level is this: If I have $400 in salary and 40 roster spots to field a competitive team, and the least I can roster a player for is $1, how do I decide who deserves more than a $1 salary? Setting that replacement level as correctly as possible is critical, as knowing how much better Kershaw is than a replacement level pitcher informs the valuation of exactly how much Kershaw is worth in terms of salary. There is probably no right or wrong way to set replacement level, but understanding how that baseline affects valuations across the board is important.

    • Bill, let me start by saying that I believe we are both correct. However, we define completely different things. I believe I am accurately defining “Replacement Level” which will lend itself to future “Points Above Replacement” calculations. While I believe you are correctly defining what a team needs to be competitive in a league. Here is the main difference. The first is that Replacement level is competition independent, whereas the level of performance needed to be competitive that I believe you are defining is competition dependent (in addition to being league dependent).

      When defining replacement level it is important to consider readily available alternatives. A replacement level pitcher should someone who can easily be picked or acquired (or a comparable SP could be picked up). An example can be seen when considering WAR. The acquisition of a zero WAR player is essentially “free” in that it costs teams relatively anything and is easy to acquire. If you are considering readily available alternatives to my 2014 replacement level example of Ryan Vogelsong (4.24PPIP) you could use the following list: Jon Neise (4.25), Rick Porcello (4.24), Bartolo Colon (4.23), Chris Tillman (4.22), and Kyle Gibson (4.20). Now let’s do the same thing for a couple PPIP scenarios which you laid out in your article. (4.29 is really close to the 4.24 I arrived at, so I’m ignoring that).

      4.33: Drew Hutchison (4.34), Danny Salazar (4.33), Tim Hudson (4.31), Nate Eovaldi (4.30).

      4.50: Drew Smyly (4.56), Odrisamer Despaigne (4.54), Hiroki Kuroda (4.50), Charlie Morton (4.49), Jake Odorizzi (4.45), James Shields (4.45)

      4.68: Zack Wheeler (4.69), Ervin Santana (4.64), Kevin Gausman (4.61), Yordano Ventura (4.60)

      4.90: Dallas Keuchel (4.93), Francisco Liriano (4.92), Scott Kazmir (4.90), Julio Teheran (4.90), Sonny Gray (4.90), Lance Lynn (4.87).

      At this point it should become pretty clear that pitchers in these categories of SP really are not readily available in free agency. That being said, these types of pitchers are definitely necessary to field a competitive team. This is similar to the difference between a 2 WAR player being “average” and a 0 WAR player being replacement level. Certainly a team full of replacement level talent would be terrible, but that is the point. In contrast, a team full of 2 WAR players (to use an MLB example) would be pretty good.

  4. Bill Porter says:

    It is fair to say that those type of pitchers aren’t available in free agency, of course – but that’s why I made the point about how Ottoneu has trades that approximate MLB free agency in determining replacement level. Most of those guys are readily available for contending teams — injured players can be dealt quickly to patch holes (example — last year I traded Patrick Corbin at $7, who was to be a cornerstone of my rotation before his TJS) or young players/prospects (the draft prospects come into the Ottoneu pool in midseason, and with the Top 100 prospect lists changing and being released in Jan and July, tons of reshuffling always creates opportunities to use and acquire young assets that a rebuilding team might as well take a flyer on if they’re out of contention and dealing overpriced stars…

    • fazeorange says:

      I appreciate the work @OttoneuTrades has done here to setup the methodology for finding replacement level for starting pitching (it isn’t easy to do and is virtually impossible to find for Ottoneu), and I also appreciate the in-depth review and critique posted by Bill Porter (who does great Ottoneu work on his own site).

      Bill’s initial critique suggests that this method of calculating the 1,175 IP needed from each team doesn’t properly account for those teams that choose to accrue far fewer IP in a given season for various reasons of competition (rebuild, injuries, general non-competitiveness). Bill is correct – in Ottoneu it is common for only 3 or 4 teams to remain competitive all season long (possibly fewer) – so it follows that not every team will be motivated to “throw” all of their SP innings (or RP IP). However, true replacement level calculations are (generally) indifferent to competition level. In other words, the SP replacement level method above already accounts for any level of league competition because replacement level is dependent upon ownership, not contention.

      Using @OttoneuTrades methodology, all SP were first ranked by PPIP (Kershaw ranked #1), until the 86th ranked pitcher (Vogelsong) accounted for the 14,100 IP needed with the league for all SP. The key here is that the team that owns SP’s 1-85 is irrelevant; what is important is that they are owned by any team. To clarify, the real assumption in this method is that all teams, regardless of their position in the standings, will own the best SP available, or the top 85 SP in the league. While my team may be rebuilding, there’s no reason for me to cut an above-replacement level SP – I should always (assumption) be looking to improve my team and add future value if next year is my target – or I can (theoretically) trade an above-replacement level SP for some other asset because contending teams are looking for improvements too. Either way, I think it’s safe to assume that productive players are going to be owned, by who owns them has little impact on what constitutes replacement level for the position. Finally, even if above-replacement level SP are cut, there’s no reason they shouldn’t be added by another team that is actually in contention. Is essence, the 4.24 PPIP represented here by Vogelsong is the replacement level of SP within all of Ottoneu FGPoints, not necessarily the replacement level SP for any specific league or team.

      There are a lot of assumptions here, as Bill correctly pointed out. The key issue, however, is not the level of competition, but the ownership rate. Good (performing, non-injured) players will be owned because every team has a vested interest to improve (not necessarily to win in the current season), thus making replacement level that next player on waivers available to anyone (and probably a contending team looking to backfill an injury). A non-contending (rebuilding) team may never think about picking up Ryan Vogelsong, but as a true contender in a points race, I may consider adding him to “stream” if I think I can get more than 4.25 P/IP for 6 innings against the Padres.

      Whether Ryan Vogelsong embodies the exact value of replacement level in 2014 is less important than the actual replacement level of performance, or 4.24 PPIP using this method. While it is true that large trades can swing the balance of talent owned by various teams at any point in a given season, replacement level, in this case, is intended to be used at the macro (league) level, and not at the specific team or league level. If I were to swing a trade for several good SP at the beginning of a season and end up with 7-9 excellent starting pitchers, I may never be in need of adding SP #86, but that doesn’t mean other owners, regardless of their position in the standings, aren’t also looking for improvement over their current rotation. After all, you never really quite know when a two or three week hot streak is real (Jake Arrieta) or fake (Colon), so good owners are those that can quickly and accurately assess a player profile and are willing to sometimes to a chance that things could actually pay off. This clearly applies to contenders and rebuilders the same, and may be more a function of “activity” (waiver wire skill, player evaluation, etc.) than it is of the standings.

      In general, the replacement level starting pitcher is that pitcher whose value constantly rides the fence between “useful” and “disposable”. In other words, it would be unsurprising if these pitchers were some of the more transactional (added/dropped) players in a given season, most likely at a price of $1-$3 (and often owned by multiple teams throughout the season). Whether or not you added Vogelsong in 2014, consider some of his other contemporaries near the bottom of the replacement level list: Aaron Harang (4.26 PPIP), Henderson Alvarez (4.26), Kyle Lohse (4.26), Doug Fister (4.25), and Jon Niese (4.25). My guess is these guys were yanked on and off the waiver wire all season as possible “spot” starters, the very definition of replacement level. To address Bill’s second point above, I’d be surprised if SP’s like Kazmir and Kuroda were jockeyed up and down with the same regularity as those mentioned near Ryan Vogelsong.

      One final point about replacement level: it’s not a bad thing. In fact, it’s not unreasonable to assume that a winning, championship level team will need to use multiple replacement level starting pitchers in a given season just to maintain their innings and the pace of the competitors. After all, 4.25 PPIP is 850 points over 200 innings, and only 44 SP hit that total points mark in 2014 (or half of the 86). Replacement level pitchers have value.

  5. Bill Porter says:

    I think you’re missing my point — specifically, those higher level SPs are available in a way that nearly approximates being free agents. Of course there may be differences from league to league, but in our league, Ervin Santana was moved straight up for Eddie Butler; Kuroda in a non-elite prospect deal; Doug Fister for a replacement level Erick Aybar. Yes, they are not straight free agents, but pitchers far in excess of the 4.24 P/IP number are available in a near waiver wire manner…

    • fazeorange says:

      I think I understand where you’re coming from, but availability of these above replacement level SP doesn’t change the macro calculation of replacement level because they are never not owned (within the league). In fact, using your examples of Santana, Kuroda, and Fister, it makes sense that they are moved for something small because they are just valuable enough over and above replacement level that a) their current owner doesn’t want to drop them, and b) their new owner is willing to pay at least some (minimal) price above replacement level (something above $1, either in the form of another player or more $). Neither their availability nor their acquisition cost in a trade has any bearing on the actual replacement level of starting pitching owned throughout the league. Who owns them (or what they paid to acquire them) doesn’t matter; all that matters is that they are owned (and not cut) at all. Trade value is indirectly related to replacement value. As an example, I’d rather own a replacement level SS than replacement level SP simply due to position scarcity, but the fact that both are owned at all is what determines the break point of the replacement level calculation.

  6. Bill, I believe I understand what you are saying, but in each specific case you mention, those players were traded for. I think this is a very important qualifier… Would the owner of Butler prefer to acquire Santana without giving up anything? Probably. Would I rather give up someone for Kuroda or just pick him up? The second. Same with Fister, I’d rather just pick him up in Free agency for a dollar (if they were truly replacement level) than I would give up anything to opposing squads. Yes, you are right that trades are available and they are relatively easy to pull once teams fall out of the race, but I do not believe that changes the baseline of replacement level. Each of these players is just slightly above the “replacement level” as indicated by the price paid for each of them in a trade, just as they are slightly above replacement level by PPIP.

    Also, let’s point out that ~4.24 PPIP should sound low. I will not bear itself out in the final league standings, nor should it. I feel very confident that a replacement level team would score well below the typically 12th place team in a league by PPIP. A replacement level team should be absolutely awful (much worse than a typcal 12th place squad). Not to beat a dead horse and keep drawing this back to MLB comparisons, but all studies I’ve seen show a replacement level MLB team would win around ~46 games. Yet how often do we see teams go 46 -116 (or near that level)? The answer is not too often – I’m looking at you 2003 Tigers.

    The point is that it should be extremely difficult for an ottoneu team to be so bad that they are actually performing at a replacement level, just as it should be very rare to see an MLB team performing at replacement level. I believe we are overstating replacement level if we believe that a team performing at this level should be commonly seen in the standings. A replacement level bar in the range of mid 4.2s for SP sounds low… It should. I doubt many of us have seen a team with starting pitching that bad. Just as the 2003 Detroit Tigers do not happen very often.

Discuss! We're listening...

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: