Preseason blog series: Welcoming Carl and who is more prone to breakouts and busts: pitchers or hitters?

Carl is excited to join the league this year. He is the father of current league member Dean, and has been getting crushed by Dean in fantasy sports since about 1995, so he should feel right at home in this league.

On a personal front, Carl is married to Jessica (yes, Dean and Carl are both married to a Jessica Nordhielm). Jessica is from the Philippines so Carl tries to get out there every few years.  They live in Tallahassee, Florida where they cheer on the Seminoles despite having no obvious ties besides Tallahassee.  Carl is an avid runner (and very good, though recent injuries have set him back from his peak running days) and up until his recent injuries could claim he was faster than Dean.

From a sports perspective, Carl cheers for Boston teams (which I think all missed the playoffs this year) and occasionally his alma mater UPenn (which really is only good in fencing, chess, and the like).  His favorite team is definitely the Red Sox, so expect him to balance out Max and Arthur’s reaches for Yankees players by reaching for Red Sox players.  Carl also has the distinction of being the first player in the league to have a published fantasy baseball article, having done so quite a few years ago before we all knew what a blog was. 

Up until recently, Carl was running country night clubs in Chicago and Indianapolis (Saddle Up Saloon), but COVID has forced both of those operations to shut down, so he's shifted more of his attention to his business that creates websites for homeowners associations.

Welcome to the league, Carl.

---------

The question has been asked in this blog for years now, what is more projectable: hitting or pitching? We’ve seen in past blogs that a hitter’s productivity is fairly projectable overall, with the fantasy-relevant statistics generally falling within a +/- 30% margin as a total (as in, Mike Trout is projected for 30 home runs so it’s a fairly safe bet that he will hit between 21 and 39…but that number improves to around +/- 20% on a per plate appearance scale (Mike Trout’s 30 HR per 600 PA(0.05 HR/PA) will be between 0.04 and 0.06 HR per PA). Removing Stolen Bases, which are the least projectable stat, improves those numbers to +/- 25% and +/-18%, respectively.

For pitchers, the fantasy-relevant categories have been projectable to a similar +/- 30%, removing wins and saves from this calculation and we improve projections to +/- 25%. The most projectable stats are WHIP, FIP, and strikeouts per 9 innings. These numbers come out mostly the same for a per-inning-pitched analysis because so many of the pitching categories are already based on a per inning pitched metric (ERA, WHIP, FIP, k/9)

But what about breakouts and busts? We’ve seen that projections only take fantasy teams so far. Look at last year, Dean was projected coming out the draft to score 105 roto points, a 15 point edge on the second best-projected team which was Michael at 90 points. Dean didn’t find enough breakouts, his pitching strategy didn’t pan out, and he ended up in 4^th place.

The arbitrary definitions of breakout and bust I will be using here are different for hitters and pitchers. For hitters, I have used a player-rater-style points summation process, and if a hitter outperformed or underperformed their point total generated, then they were a breakout or a bust; or if they ended close to their projected total, they were flat. For pitchers, the player rater summations put too much weight into two fifths of the standard fantasy categories, wins and saves, which skews any projection style analysis, so instead I used % differences between a players projection and their end of year numbers for: Innings Pitched, Strikeouts, ERA, WHIP, FIP, and k/9. Averaging the % differences, I called them a breakout if they were more than 10% better than projected and a bust if they were more than 10% worse than projected.

Let’s show what this looked like for perspective on the conversation. Here is how the league fared on their draft picks last year:

ROTO SCORE	Roto Points Projected	Breakout Draft Picks	Bust Draft Picks	Flat Draft Picks	Roto Points Actual
Brian/Josh	81	6	5	8	89
Arthur	65	3	7	6	80
Paul	73	2	4	7	79.5
Cory	75	2	6	7	55.5
Ryan	64	5	6	4	68
Keith	66	5	5	6	53
Dean	105	2	2	12	85
Dave	71	1	5	6	86
Max	67	3	5	11	58
Michael	90	5	8	4	116

 

(note not all players drafted aren’t accounted for here, some players like Relief Pitchers didn't generate enough data to analyze)

There’s a number of observations from this and things to keep in mind.

First, most of Michael’s breakouts ended up being from players that were dropped by other teams (think, Adalberto Mondesi and Luke Voit), second the Roto Points projected from pitching only accounted for about 75% of the actual league Innings Pitched (4170 IP were projected at the draft, but 5450 IP were tallied by the league last year), therefore there was a ton of streamer IP that aren’t accounted for here. Finally, this shows how important in season management is. There were 4 teams that had at least 5 breakouts on their drafted team (even Sam Nunziata drafted Shane Bieber and Max Fried who broke out)…but the team that drafted the most busts won the league. Having a quick (and correct) trigger on underperforming players is very important.

So, what does this look like for a larger, player-pool-level, analysis?

First, a couple of caveats. Projection gathering is newer for me. So I only have a few years worth of data, and back in 2018 I only had 2 projection systems worth of input (for reference in 2021 there are 5 projection systems on fangraphs plus yahoo’s publicly available now). Also, the 2020 data is only from 60 games, so utilizing any data from that year contains all the small sample size caveats.

That being said, here is how the breakdown of hitting versus pitching has gone. Each of these years I used approximately the top 170 projected hitters for the year and the top 100 pitchers projected for the year.

Hitting	%Breakout	%Bust	%Flat
2018	17.3%	38.1%	44.5%
2019	23.5%	31.8%	36.2%
2020	13.2%	28.1%	54.4%
Pitching	%Breakout	%Bust	%Flat
2018	20.4%	42.0%	37.6%
2019	17.7%	11.8%	44.7%
2020	17.7%	36.4%	35.7%

What we see here is that there tend to be more people projected within the flat (+/- 10%) range of their projected outcomes than breakouts and busts, and second, there tend to be more busts than breakouts. This makes sense, if you only are collecting data from the top 170, you won’t capture those from outside the top 170 that enter it (i.e. capture breakouts from below 170 that are projected poorly), but you will capture those from the top 170 that fall out of it (i.e. capture busts from the top 170 that are projected well). This is also why in-season management is important.

The other observation is that there are fairly equivalent levels of breakouts and busts across hitting and pitching. With these relatively few data points and inconsistent outputs, I’m hesitant to call either hitters or pitchers more prone to a breakout or bust…though the numbers are showing signs that hitting might be slightly more stable…this would be in line with our previous analysis that pitchers tend to have a larger +/- % performance compared to their projection.

So what did we learn about about breakouts and busts and what can we apply for this coming year? First, hitters are slightly more projectable than pitchers right now, but they both breakout and bust at about the same rate. In-season management is every bit as important to win at fantasy as the draft. Finally, you can draft a bunch of draft busts if you cut bait quickly enough and identify breakouts along the way.