Last Year’s Boom to This Year’s Bust: Avoiding the Post-Breakout Player Trap

In the classic poker film Rounders, the main character warns us that “if you can’t spot the sucker in your first half hour at the table, then you are the sucker.” The parallels to fantasy baseball are obvious – no one wants to be the owner who buys last year’s breakout stars at inflated prices. If you find yourself rostering multiple previous season surprise studs, then perhaps you are the sucker. But… are you really?

There’s a clear blueprint for success in fantasy baseball. In general, the owners who accumulate the most value from their rostered players are going to earn the glory of basking in a Yoo-hoo shower at the end of the season. This of course assumes the most valuable roster isn’t wildly unbalanced, like with all the steals and none of the power. Because once you climb into first place in a category, any additional stats amassed are worthless.

So how does one get to the top of the value mountain? By hitting on breakouts. Multiple breakouts. Assuming your league is at least somewhat decent at fantasy baseball and doesn’t allow you to buy all the elite players at hefty discounts, you’re going to need several, if not the majority, of your players to earn vastly more than you paid for them. So we rely on breakouts to achieve the ultimate success.

Looking toward the upcoming season, the question then becomes what do we do with last year’s breakouts? The concept of regression has been hammered home for years now. So when a player surprises with a performance spike, our knee-jerk reaction is to expect that player to give back some, or perhaps even the majority, of his gains. Does this really occur? How much of those gains are given back? What percentage of breakouts hold onto their gains or even make further gains?

So many questions, so little time. Luckily, we have the data necessary to perform the research. I began with historical ADP data going back to 2013 provided by our friends at Razzball and then ran the FanGraphs auction calculator for a 12-team league with standard rosters, composed of 14 starting hitters and nine pitchers. Then, it was off to work.

Out of 4,437 player seasons, there were 1,526 that were part of a three consecutive season trio (2013- 2015, 2014-2016, etc). The plan was to determine whether a breakout season occurred from Year 1 (Y1) to Year 2 (Y2) and then what happened in Year 3 (Y3).

I decided to look solely at dollar values earned, rather than ranking, as it’s more precise. A breakout was defined as a player earning at least $10 more than the previous season. If the player “earned” a negative value, I changed it to $0. That way a player who went from -$15 to -$5 would not end up on the breakout list. Last, I removed all relief pitchers and closers, as values earned by these groups are more a function of opportunity than skill. Of the 1,326 remaining player season trios, 150 qualified as a breakout. Note that, in the interests of time, I didn’t control for playing time and injuries, so there may be some false positives mixed in.

Let’s begin our discussion of the results with the entire pool. Then, we’ll separate the hitters from the pitchers.

First, we’ll compare what the player earned during his breakout Y2 with how he followed up in Y3.

The horizontal axis depicts what the player earned in his breakout Y2 seasons, while the vertical axis indicates what the player earned in the following season, Y3. While the trend line does rise as you would expect, it’s still a big mess of dots in random spots. The r-squared is a lowly 0.116, which does confirm the positive correlation between what a player earned in his breakout season and the following year, but it’s a weak correlation. Note how many players became worthless in Y3, which is likely partly due to playing time changes (whether from injuries or decreased role).

For context, the r-squared of the entire 1,326 player pool from Y2 to Y3 was more than double at 0.278. That means that players aren’t very consistent earners from year to year, but it also means that breakout players are far less likely to earn similar value than non-breakouts. That’s probably no surprise, and in most fantasy leagues, owners aren’t paying in Y3 what the breakout player earned in Y2. There is typically a “do it once more to make me a full believer” discount. Later we will bring ADP into the mix to determine whether owners are overpaying, underpaying, or fairly paying for these breakout players in Y3.

But before we do that, let’s transform the above graph into a table and analyze the numbers using player value buckets.

Player Value in Year 3 Given Value in Year 2
Value in Year 3
Value in Year 2 Percentage of Players $0-$15 $15-$21 $21-$28 $28+
$28+ 16.7% 52.0% 4.0% 20.0% 24.0%
$21-$28 22.7% 52.9% 26.5% 14.7% 5.9%
$15-$21 30.0% 75.6% 8.9% 11.1% 4.4%
$0-15 30.7% 84.8% 8.7% 6.5% 0.0%
TOTAL 100% 69.3% 12.0% 12.0% 6.7%

The numbers on the left correspond to value earned ranges in the breakout season Y2, while the “Value in Year 3” ranges represent value earned ranges in the following season Y3. The “Percentage of Players” column represents the percentage of players in that Y2 value bucket, and the percentages in the bottom row represents the percentage of players in that Y3 value bucket.

Starting at the Total line, we find that overall, a whopping 69.3 percent of breakout players earned less than $15 the following season. Of those who earned at least $28, more than half earned less than $15! That means that half of these big earners in Y2 saw their values cut in half at minimum. Only about a quarter of the big-earning $28+ group earned at least that amount again the following season.

A very small percentage of players earned in the same or higher range in Y3 compared to Y2, proving that the vast majority experience a dramatic decline in value after breaking out.

So we now know how players in various value buckets performed the year after they broke out, but does the magnitude of the breakout matter? Is a player who gained $20 in value during his breakout more likely to hold their gains than a player who gained $10 in value? Let’s look at one more table to answer that question before bringing ADP into the fold.

Player Year 3 Value Change Based on Year 2 Magnitude of Breakout
Value Gained/Lost in Y3
Value Gain Y2 vs. Y1 % of Players -$20 -$20 to -$10 -$10 to -$3 -$3 to $3 $3-$10 $10-$20 $20+
$20+ 24.0% 44.4% 16.7% 22.2% 11.1% 2.8% 2.8% 0.0%
$15-$20 27.3% 14.6% 39.0% 24.4% 12.2% 4.9% 2.4% 2.4%
$12-$15 20.0% 6.7% 43.3% 16.7% 20.0% 10.0% 3.3% 0.0%
$0-12 28.7% 2.3% 48.8% 16.3% 11.6% 16.3% 4.7% 0.0%
TOTAL 100% 16.7% 37.3% 20.0% 13.3% 8.7% 3.3% 0.7%

The values on the left correspond to value gained ranges in the breakout season Y2 versus Y1, while “Value Gained/Lost in Y3” ranges represent value gained/lost ranges in the following season Y3. The “% of Players” column represents the percentage of players in that Y2 value gained bucket, and the percentages in the bottom row represents the percentage of players in that Y3 value gained/lost bucket.

I’m treating the middle Y3-Y2 bucket of -$3 to $3 as the neutral bucket, in which value from Y2 to Y3 is essentially unchanged. That means that a whopping 74 percent of players lost more than $3 of value after their breakouts and 16.7 percent lost more than $20! Only 12.7 percent of players actually took another step forward, gaining additional value from their breakouts.

And the magnitude? Oh, it matters, but in the opposite way of what you probably expected. The bigger the breakout, the bigger the fall, it would seem. Of those who gained at least $20 in value in their breakout, 44.4 percent gave it all back, losing more than $20 the following year. It was almost as if the breakout season never even occurred, as they dropped right back down to where they had been in Y1.

The group that gained the least in their breakouts, those who gained less than $12 in value, fared best the following year, as 21 percent gained even more value in Y3. However, they were just as likely to lose value as the $15-$20 group.

So no one is safe, regardless of how big the breakout was. The big breakouts and the small breakouts are highly likely to give back a significant portion of those gains the following season.

As hinted at earlier, it’s finally time to bring ADP into the analysis. I limited both ADP and EOS (end of season) rank to an arbitrary maximum of 400 to prevent sky high ranks like 1,000 from skewing the averages. Below is a table comparing ADP rank, end of season rank, and EOS dollar value earned for Y1, Y2, and Y3.

ADP & EOS Rank and Value Comparisons
Y1 Y2 Breakout Y3
ADP Rk 217 213 102
EOS Rk 276 55 189
Avg EOS $$ Val $3.77 $20.21 $10.35

In Y1, these players weren’t exactly the apple of fantasy owners’ eyes. They were drafted late, and  ended up earning only a couple of bucks, just like fantasy owners expected. In Y2, owners drafted them in similar fashion, with ADP almost identical. But instead, this group enjoyed a performance spike and their EOS rank settled at an impressive 55, while their average EOS dollar value earned zoomed up to just over $20. That’s a nearly $16.50 gain in value!

So naturally, fantasy owners took notice, nearly pushing them into their top 100 the following season. Observe that this is still a sizable discount to what these players actually earned in Y2. So some regression is most certainly baked into prices. But that assumed regression wasn’t enough! That’s because these players actually earned closer to the 200th overall value than the 100th. And their EOS dollar values earned were nearly cut in half on average.

This table supports the notion that last season’s breakouts should be completely avoided at your fantasy draft. There will always be someone who is a believer, and that will prevent you from buying such a player at the required discount necessary to make the player a good purchase.

Now that we have looked at the player population as a whole, let’s find out whether we should treat hitter breakouts any differently from pitcher breakouts. Out of 150 total breakouts, there were 109 hitters included in the breakout group and 41 pitchers. The first two tables shared above look very similar to the data for just hitters and just pitchers, so we’ll skip right to the ADP vs EOS numbers.

ADP & EOS Rank and Value Comparison (Hitters Only)
Y1 Y2 Breakout Y3
ADP Rk 210 208 102
EOS Rk 277 54 180
Avg EOS $$ Val $3.79 $20.25 $11.22

The hitter’s table paints a similar picture as the entire breakout population table.

And now for the pitchers:

ADP & EOS Rank and Value Comparison (Pitchers Only)
Y1 Y2 Y3
ADP Rk 235 226 101
EOS Rk 273 59 214
Avg EOS $$ Val $3.74 $20.09 $8.01

Finally we see some variance. Though the breakout season was similar, pitchers performed worse than hitters in the season following the breakout. This is consistent with the idea that pitchers are significantly more volatile since their performance is more affected by factors outside of their control, like team defense and bullpen support. If the advice is to avoid last season’s breakouts in general, then that’s doubly true when referring to starting pitchers.

So we have learned that last season’s breakouts make for poor investments. But if we avoid last season’s breakouts, surely we would like to unearth this season’s breakouts. So let’s determine whether any particular position generates more breakouts than the rest.

In addition to determining breakout status, I performed all the same analysis for busts, which was defined as a player losing at least $10 in value from Y1 to Y2. The percentages of both breakouts and busts are included in the position breakdown table.

Breakout and Bust by Position
Position Breakout Bust
C 6.9% 11.1%
1B 20.3% 18.9%
2B 13.1%  5.8%
SS 10.9% 9.4%
3B 9.9% 8.8%
OF 12.8% 15.2%
DH 50.0% 50.0%
SP 9.7% 7.8%
All 11.3% 10.7%

If you’re seeking a breakout candidate, you’ll want to begin your search at first base. Just over 20 percent  qualified as breakouts, which is nearly double the overall average and significantly higher than the second highest position. Of course, betting on first basemen isn’t without risk, as it also comes with the highest bust rate!

Perhaps not surprisingly, catchers are the worst position when looking to uncover a breakout. The majority of them simply don’t get enough playing time to amass fantasy value, thus making it extremely difficult to gain $10 in value. With the bust rate also coming in third highest, it’s clear that catchers are the riskiest draft day investment.

Surprisingly, second base is where it’s at. The group boasts the second highest breakout rate and the lowest bust rate. It’s possible that stolen bases are the reason here, as speed is less prone to slumps.

I’m actually quite surprised that starting pitchers sport the second lowest breakout and bust rates. I’m not sure what the explanation is for such low rates given the volatility of the two ratio categories. However, it does give one pause on loading up on young pitchers in the late rounds with the hopes of hitting on a couple of breakouts. It’s clearly easier said than done.

So what have we learned today? Be very, very careful when investing in last season’s breakouts. They rarely take another step forward and the regression that’s already baked into their draft day cost is usually not enough.


Mike Podhorzer is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year. He produces player projections using his own forecasting system and is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. His projections helped him win the inaugural 2013 Tout Wars mixed draft league. Follow Mike on Twitter @MikePodhorzer and contact him via email.
newest oldest most voted
sgvette
Member
sgvette

So, my picking Villar early and Porcello in 2017 are spectacular examples of this theory?

phaddix
Member
Member
phaddix

Mike – been enjoying your projections. This is a great article. So, by this metric who were the breakouts in 2017 – I.e. those players who posted gains of $10+ and are you betting against all of them or hoping at least a few can retain value.

Paul Sporer
Member
Member

Excellent work, Mike!

stonepie
Member
stonepie

did you look at year 4 by chance?

Aaron
Member
Aaron
Great analysis, Mike! I did a similar analysis, focusing specifically on positions and which positions are the best bet to break-even, return equity, or bust. I found a similar surprise at 2B, particularly among $21+ draft-cost 2B, which ranked as the 2nd most likely to break-even, 2nd most likely to return at least $10 and $20 in equity, and 2nd least likely to bust (1B was first across all of these areas). Surprisingly, SP had the fourth strongest r-squared between draft cost and EOS value (0.24). However, to your point, SPs costing $21+ at the draft weren’t the best bets:… Read more »
Matthew Martin
Member
Member
Matthew Martin

It would be cool to see which specific players bucked the trend e.g., Trout, Altuve, Kershaw, and whether there is any way to predict a player that will buck the trend. If you know, patent the process.

Matthew Martin
Member
Member
Matthew Martin
This kind of freaks me out; I play in a keeper league and the SPs I am keeping are Severino, Ray, Nola, Godley, and Ohtani (who has not had an MLB breakout, but maybe his NPL breakout already happened?). On the other hand, for hitters, I am keeping, among others, Eddie Rosario, A Hicks, and Moncada. Hicks earned -$1.7 (or zero dollars in your methodology) in 2016 and $26.4 in 2017, and Rosario $7.2 in 2016 and $20 in 2017. Per the conclusions of the article, I should not keep them. However, the cost of keeping these two is so… Read more »