2018 Bounceback Players (WIP)

PREAMBLE INTRO

As fantasy baseball players -- and as humans, frankly -- we tend to fall prey to a phenomenon known as recency bias. Essentially, recency bias theory suggest that we're inclined to use our recent experience as the baseline for what will happen in the future. More specific to fantasy baseball, my interpretation is that we overweight the most recent information more than we should. Baseball is a variance-driven game; there are tons of variables that contribute to odd things happening.

Of course, I'm not really telling you anything you don't know. The question is, how can we exploit recency bias to our advantage? I don't have the perfect answer to that question, but I believe I have a good start: who was highly valued a year ago, in 2017 drafts, but isn't today? In other words, who are some of the biggest decliners in ADP year-over-year? And of those, who, perhaps, shouldn't be? Or, which players are victims of recency bias and are worth a bet this season?

APPROACH

If you're a normal person, unlike me, and find the sausage-making process boring, please feel free to skip this section and dive directly into the players. I did want to briefly touch on how I'm devising this breakout list though, which is a triangulation of several different data points:

First, I'm using 2017 ADP minus 2018 ADP as the starting point, and sorting by the largest decliners. Next, because I largely play in auction leagues, I've assigned an implied auction value based on my league's historical data ($300 budget, 22 players drafted). I've also included additional auction value data points: 2018 median projected value and 2015-2017 historical values. These data points are especially important for two reasons: 1) To understand the 2018 cost, or "risk" needed to deploy in order to capture value from a bounceback and help answer the question, "is the discount large enough?"; and, 2) The level of upside that may exist. Sure, if a player fell 100 spots from 400 in 2017 to 500, and their highest production to date has been $3, there could be a bounceback in there ... but is it worth our attention? Because of this, I only looked at players going 350 or earlier by 2018 ADP. (ADP data is as of 3/10).

Anyway, that's our starting point. I'll also be leveraging:

1. xStats: Calculates what a player should be producing based on exit velocity and launch angle
2. 2017 Injuries: Identified which players were playing through injury in 2017, particularly useful for less-publicized "nagging" injuries that are unlikely to be chronic
3. Spring Training News: Any news out of spring training? Changes in swing, new pitches, health, improved approach, different lineup spot, etc.
4. Projected Batting Slot: Likely lineup slot, per Roster Resource

That should do it! With that, let's dive into the best bounceback bets for 2018.

2018 ADP 200+

Carlos Gonzalez, OF, COL
ADP: 2017: 63 | 2018: 295 (Diff: -232) | Proj Batting Slot: 3
2018 Values: ADP Implied: $1 | Projection: $3 ($1-6)
Historical Values: 2015: $28 | 2016: $22 | 2017: $1

CarGo is the poster boy for this season's bounceback analysis, being drafted 63rd overall last season and producing an absolute dismal season. While the 295 ADP is artificially low as he didn't have a team, since he re-signed with the Rockies on 3/10, his NFBC ADP has pushed up closer to 258. Still, that implies a $1-2 player. Projections alone think he's worth $3 this season, and he's produced seasons of $22 and $28 just prior to 2017. He's also still in Colorado and the bandbox that is Coors Field, and FanGraphs' Jeff Sullivan thinks he'll play everyday. Roster Resource projects him to bat third in a good lineup. What happened last season though!? CarGo was reportedly dealing with insomnia, which he purportedly fixed late in the season. To wit, his xStats in September were .313/.413/.622 with 5 xHR (33 HR per 600 PA pace).

Verdict: Still just 32 and playing regularly in Coors, at $1-3 he's a strong buy late in drafts for me.

Hanley Ramirez, DH, BOS
ADP: 2017: 76 | 2018: 301 (Diff: -225) | Proj Batting Slot: 3
2018 Values: ADP Implied: $1 | Projection: -$9
Historical Values: 2015: $3 | 2016: $26 | 2017: -$6

Admittedly more of a dart throw than CarGo, Hanley has still shown near $30 upside as little as one year ago. Additionally in his favor, his xStats last year suggest he should have been better: .271/.346/.479 and 27 xHR per 600 PA (553 PA). That's not bad! There are a few knocks against him, however. First, he's 34 (two years older than CarGo) and has shown this non-existent floor throughout his career. Second, due to an interesting clause in his contract, the Red Sox actually have a bit of a disincentive to play him. If he reaches 497 PA this season, an option for $22M in 2019 is automatically triggered. It's not hard to envision a scenario where the Red Sox want to keep him "fresh" while avoiding paying $22M to a 35-year-old and injury-prone DH. He also underwent "minor" offseason shoulder surgery to repair issues he was dealing with last season -- on one hand, this could be good news that the issue is fixed but on the other, he seems to always be dealing with these kinds of issues.

Verdict: Worth a $1 or reserve pick, although I like CarGo more.

Matt Kemp, OF, LAD
ADP: 2017: 95 | 2018: 322 (Diff: -227) | Proj Batting Slot: 3
2018 Values: ADP Implied: $1 | Projection: -$32 (ZiPS: $8)
Historical Values: 2015: $22 | 2016: $23 | 2017: -$1

Kemp feels like a hybrid of the CarGo and Hanley stories above: he's shown more consistent annual value similar to CarGo, but he also suffers from the a similar potential playing time issue as Hanley. Reports out of spring training are glowing for Kemp, who's reportedly lost 40 pounds this offseason and has produced well in spring training, leading to him being penciled in as the Dodgers starting LF and 3rd in the lineup. Kemp only had 467 PA and was essentially a $0 player, but looking deeper into his xStats suggests he was still very productive -- his per 600 PA stats were .291/.333/.500 xTriple Slash, 27 xHR, 60 runs, and 82 RBI. You don't expect him to steal anymore, but if he's getting the PA, you do expect him to produce better runs and RBI numbers. With triple slash upside still clear there, and not terribly old at 33, you can squint and still see Kemp being a productive player. And yet, I can't get it out of my head that it was seemingly a joke and purely financially-driven acquisition the Dodgers made by acquiring him. Are they just showcasing him for a future trade? Also, the Dodgers are flush with viable options across the diamond -- between Joc Pederson, Andrew Toles, Trayce Thompson, Kike Hernandez Alex Verdugo, among others -- enough to reduce his counting stat upside. Additionally, YMMV but I play in local SoCal leagues and you typically have to pay an Angels or Dodgers "tax" for the local players, driving their cost up.


Verdict: Worth a $1-3 bid or early reserve pick -- I like Kemp more than Hanley but less than CarGo.

Todd Frazier, 3B, NYM
ADP: 2017: 77 | 2018: 256 (Diff: -179) | Proj Batting Slot: 4
2018 Values: ADP Implied: $1 | Projection: $2 ($1-5)
Historical Values: 2015: $24 | 2016: $21 | 2017: $1

Frazier appears to have gotten a tad unlucky last year, rather than a skills drop off -- he actually registered a career-high 14.4% walk rate en route to a career-high .344 OBP. 

Equity Analysis: Where To Place Your Bets (WIP)

OVERVIEW

Last week I explored the common threads across league-winners -- that is, is there something repeatable and useful about what league champions are doing? One of the key questions that surfaced from the analysis was, "are there certain positions that are better bets than others?" For instance, conventional wisdom thinks pitching and catchers tend to be the most volatile positions, and thus might warrant less draft capital. Are they right? That's what I wanted to dive in deeper to here. Furthermore, why does this matter? Well, Rudy Gamble of Razzball, among others, have conducted research on years of past champion data. They found that across all champions, they tend to earn value from their top picks. That makes sense intuitively -- if you don't get production from your top picks, then, well, it's going to be hard to win. But my question then becomes, how do we know who's a good top pick to bet on? That's some of what I'd like to get into here.

APPROACH

To conduct this analysis, I first wanted to understand how draft value correlates with end-of-season value. I used the last three years of one of my home league's data (traditional 5x5 mixed, 12-13 teams) to assess the relationship between draft day cost, end-of-season value, and the resulting equity. While we're very interested in which positions returned the best equity, I wanted to also look into which positions tend to see the highest bust rates.

CORRELATION ANALYSIS

In starting this analysis, I first looked at the correlations between each position and their end-of-season value. Here were the results, sorted by highest r-squared to lowest:

Position	Overall-EOS
2B	0.396
1B	0.347
DH	0.294
SP	0.243
3B	0.216
C	0.159
OF	0.150
RP	0.058
SS	0.043
Hitting	0.225
Pitching	0.224

Starting at the top, second base and first base had the strongest relationship between draft-day cost and end-of-season value. Based on this analysis, one could reasonably hypothesize that these might be good positions to invest larger dollars into. Going back to Gamble's analysis (I can't find the link -- someone please let me know if they have it!), he states league-winners see very little loss among their top picks, avoiding injury, maintaining production, etc. That isn't to say they return excess value, per se, just that they returned value aligned with their cost. That's how I'm thinking about 2B and 1B here -- if Gamble is arguing that one step to winning your league is not blowing your initial picks, then I think 2B and 1B could be great places to start. More on this later.

On the other hand, consistent with some analysis Rob Silver had been discussing, closers' draft-day cost has very little correlation with their end-of-season value, with only SS being worse. To Silver's point, I found that among 89 RPs drafted during this time frame (15-17), only 10 broke even or returned surplus value (equity). That is, only 11% of RP at least broke even. Now, that isn't to say that you should ignore closers. In fact, in last week's analysis, we found that saves correlate with overall pitching points by a 0.30 r-squared. Rather, the key is to sink as little as possible into the position while still positioning yourself for saves. When we look at RP who lost no worse than $3, the number of pitchers that qualify jumps to 22. Their average draft salary is $7, ranging anywhere from several $1 closers to a $24 Kenley Jansen in 2017. The bottom line is, draft closers at prices where you think they can break-even. The smart money says it's hard to predict who that will be, so minimize your exposure. Who will be this year's Greg Holland, Raisel Igelsias, and Fernando Rodney? On a side note, I think SS may be skewed by some horrific Troy Tulowitzki and Hanley Ramirez seasons, where they were drafted like top picks but shit the bed, largely as a result of injuries. I don't believe this is representative of the new wave of SS we're seeing in Correa, Lindor, Turner, and Seager, who were all keepers/freezes during the majority of this analysis.

Lastly, I found the fact that hitting and pitching had largely the same relationship with their respective EOS values. Pitching is often thought of as much more volatile, but in this analysis, the numbers suggest they aren't as different as we often think. Specifically, SP draft cost had a stronger relationship with EOS value than 3B, C, and OF. Speaking of C, they are often thought of as the most volatile hitting position, which is mostly consistent with what this analysis shows, even if OF and SS fall behind.

DEEPER DIVE: EQUITY AND BUST RATES

While it's interesting to note the correlation between what you pay for a position on draft day and what they finish the season as, there are additional aspects to consider, such as price, break-even rates, equity rates, and bust rates, which all might differ based on how much you pay for a particular position. 

Consistent with the correlation analysis above, it appears 2B and 1B have the lowest bust rates while C and RP have the highest. In this case, bust rate is defined as a player realizing only 10% of their draft-day cost (i.e., a player who cost $10 but only earned $1 or less would qualify as a bust in this case). More details here:

Position Analysis, All, 2015-2017

Position	Drafted	> B/E	> $10 Equity	> $20 Equity	Bust
1B	69	43%	26%	9%	39%
2B	58	43%	19%	10%	36%
3B	57	33%	19%	12%	40%
C	78	24%	12%	3%	64%
OF	199	35%	18%	7%	46%
SS	56	21%	11%	5%	48%
SP	213	26%	10%	4%	48%
RP	89	11%	1%	0%	61%
Batting	517	34%	18%	7%	46%
Pitching	302	22%	8%	3%	52%

Across all positions, it seems like 1B, 2B, and 3B are most likely to break-even, return surplus value (equity), and avoid being a bust. While we should be careful about applying predictive value to this descriptive analysis, these are useful pieces of data to construct our rosters. In the last three seasons, these positions have had players that have performed better than other positions. Perhaps we can apply to this season and build our teams and larger purchases around these positions. After all, Gamble argued that league-winners top picks didn't bust. Are these the areas to invest to put ourselves in position to get value from our top picks?

Interestingly, and on the contrary from the correlation analysis above, batting and pitching as a whole look very different. While they both had virtually the same strength of relationship between their draft-day cost and EOS value, hitters really shined in this analysis, outperforming pitchers in their ability to break-even, produce meaningful equity, and avoid busting. Maybe this justifies why most of the industry is focused on hitting-to-pitching at a 67/33 rate.

Consistent with the correlation analysis above, it appears C and RP are unlikely to break-even, produce equity, and are more likelier to bust. I think these positions are best treated very thoughtfully, with lower investments warranted -- you need to think about how to profits here, but perhaps the thinking should be how can I achieve value at minimum cost, while focusing more on other more bettable positions. Only 11% of RP broke even and nearly no one returned meaningful equity for Christ sakes!

While this analysis is great, it's only so useful from this higher-level. To really understand which areas of each positions are likely to be better investments, we need to dive deeper into the data and double-click into buckets of investment. In this case, I looked at the same things above but from the perspective of differing draft costs: $1-10, $11-20, and $21+ for each position.

STARTING POINT: WHERE TO SPEND THE BIG MONEY

Going back to Rudy Gamble's point, league winners earn meaningful value from their top picks, even if at just a break-even basis. After all, if you're "hitting" on $1 Aaron Judge's and/or $1 Luis Severino's and adding those to a productive core of break-even larger investment players, you're suddenly looking at a much more dangerous team than if those Judge or Severino hits are merely making up for costly early investments. That begs the question: where should we be investing if we want to return value? I can't answer that question without being able to predict the future [insert obligatory comment of "at which point, I wouldn't be writing this"], but perhaps we can look at past data to give us some clues:

Position Analysis, $21+ Draft-Day Cost, 2015-2017

Position	Total	Total %	> B/E	> $10 Equity	> $20 Equity	Bust
1B	25	36%	60%	32%	16%	12%
2B	15	26%	53%	27%	13%	20%
3B	13	23%	31%	15%	15%	31%
C	4	5%	50%	25%	25%	50%
OF	50	25%	30%	16%	6%	40%
SS	10	18%	10%	0%	0%	80%
SP	53	25%	28%	13%	6%	45%
RP	2	2%	0%	0%	0%	0%
Batting	117	23%	38%	20%	10%	34%
Pitching	55	18%	27%	13%	5%	44%

In the table above, we're only looking at players that cost at least $21 on draft day. You'll notice that C and RP have particularly small samples (only 4 and 2 players drafted, respectively). Also of note, in this case, I'm calling anyone a bust who didn't return at least 50% of their value (i.e., if a player drafted for $30 doesn't return at least $15, they are deemed a bust). For what it's worth, the bust rates on a relative basis didn't materially change even as they were adjusted to, say, minimum 30% returned.

Similar to the analysis above, 1B and 2B in particular stand out as having been solid investments, seeing by far the largest break-even and equity rates, while seeing the lowest bust rates. That's exactly what you're looking for in your early purchases -- how can I build a winning foundation to enable later big "hits" the ability to carry my team as a league-winner? Clouding this analysis -- particularly on the 2B side -- is the potential freezing issue. Altuve was frozen for $20 and $25 in consecutive years, and that wasn't controlled for, leading to potentially artificially high equity rates, and artificially low bust rates. Even so, had Altuve been drafted at market rates, he still likely wouldn't have negatively impacted these numbers. Maybe there's something about this recent crop of 1B and 2B that makes them more reliable. 1B has traditionally been known as a hitter's position, so it's not particularly surprising to see them as more reliable here. On the other hand, I was bit more surprised at 2B.

Looking at the other end of the spectrum, despite the extremely small sample sizes, C and RP perform as expected -- C looks like a 50/50 shot at these prices to return value or bust, while neither of the two RPs drafted at these rates returned value or busted. So it goes. I don't want to invest at these levels at either position, I'll look to other positions at these prices, and pounce on the best values at these positions. SS also shows up extremely poorly here, with 8 of 10 players drafted qualifying as busts. Again, I'm hesitant to apply this thought process to the 2018 wave of SS , but something to file away.

In between, looking at the two most drafted positions in this range, OF and SP appear to be dangerous positions here. Only 30% and 28% break-even, while 40% and 45% bust for OF and SP, respectively. Let's see if this improves at lower costs.

WHERE TO SWING FOR UPSIDE

The $11-20 bucket represents an interesting strategy -- it's a material investment, but only 4-6% of budget. This feels like a great place to make strategic bets on upside. The bust rates are unsurprisingly higher, while the equity rates -- specifically for batters -- aren't that far off from the previous higher spend group.

Position Analysis, $11-20 Draft-Day Cost, 2015-2017

Position	Total	Total %	> B/E	> $10 Equity	> $20 Equity	Bust
1B	24	35%	38%	25%	4%	46%
2B	14	24%	36%	21%	7%	36%
3B	22	39%	41%	27%	14%	45%
C	28	36%	25%	18%	4%	54%
OF	67	34%	40%	21%	10%	43%
SS	23	41%	17%	0%	0%	48%
SP	65	31%	26%	11%	3%	55%
RP	46	52%	4%	0%	0%	61%
Batting	178	34%	34%	19%	7%	46%
Pitching	111	37%	17%	6%	2%	58%

Position Analysis, $1-10 Draft-Day Cost, 2015-2017

Position	Total	Total %	> B/E	> $10 Equity	> $20 Equity	Bust
1B	20	29%	30%	20%	5%	70%
2B	29	50%	41%	14%	10%	55%
3B	23	40%	26%	13%	9%	57%
C	46	59%	22%	7%	0%	76%
OF	85	43%	33%	16%	5%	61%
SS	23	41%	30%	26%	13%	61%
SP	97	46%	24%	8%	3%	64%
RP	41	46%	20%	2%	0%	71%
Batting	226	44%	31%	15%	6%	64%
Pitching	138	46%	22%	7%	2%	66%

CLOSING THOUGHTS (TBD)

Everything fluid and dependent on projected values and draft-day cost, but if were drafting based on the last three years:

Best Investments

1B, 2B, 3B

Would be great if they were in their prime (26-30) and on good teams

Take Your Shots

OF, 

Minimize Loss of Capturing Production