RSOB Chapter 4: The BABS Player Profiling System

New to “Ron Shandler’s Other Book”? Read the Introduction. 

---

The foundation of BABS is a basic accounting concept – the balance sheet. On the left side are your Assets; on the right are your Liabilities.

For batters, your Assets are Power, Speed and Batting Average. For pitchers, your assets are Pitching Effectiveness (a proxy for ERA and WHIP), Strikeouts and Saves. Both sides have Playing Time as an Asset as well.

The major items on the Liabilities side are Health and Experience, or actually “lack of” each. For batters, Batting Average can also be a Liability; for pitchers, Pitching Effectiveness is the comparable negative skills offset. There is also a Miscellaneous category for minor variables like moving to a new team, a significant ballpark change, or advancing age. For these variables, you can neither count on them having an effect nor quantify them, though their impact could be considerable.

Assets

Skill and opportunity have always been the two key elements to every projection, and they form the foundation of our Assets. We look for positive contributions in these categories.

Playing time: It all starts here, an element of the forecasting process with a great amount of variability. As such, players will be rated in BABS based on a broad expectation for their potential for playing time:

                   BATTERS             PITCHERS
                -----------------   ---------------
F   Full-time   Approx. 500+ PA     Approx. 180+ IP
M   Mid-time    Approx. 300+ PA     Approx. 100+ IP
P   Part-time   Fewer than 300 PA   Fewer than 100 IP

Most reputable touts go through a meticulous process of fitting plate appearances and innings into the available playing time on each team. That’s an admirable effort and vital for accurate fantasy valuations.

But let’s be honest here; the only players for whom these projections are even close to being on target are full-timers who stay healthy all season. Of the full-timers in last year’s Top 300, there were only 148 who stayed healthy all year, and that included 21 relief pitchers broadly defined as “full-timers.” Beyond the Top 300, the number of full-timers drops sharply. Even if we could deem that there were 200-250 healthy full-timers, that’s still less than 20 percent of the entire player pool.

So I opt to project playing time in broad chunks within which we can account for a good measure of volatility. There are full-timers, mid-timers (mostly platoon types) and part-timers. Beyond that, any quest for precision is mostly a waste of time.

Skill: On the skill side, players are not rated on their potential statistical output. I don’t care whether Gerrit Cole will post an ERA of 2.50, 3.00 or 3.50. There are too many variables to know where that number will land. Instead, players are rated against each other, because that’s how it all comes out anyway. Cole could repeat his 2.60 mark, but that 2.60 is far less valuable in a season where everyone and his wife’s cousin’s housekeeper is posting sub-3.00 ERAs. So players are rated against the population mean for each skill:

Extreme Impact            Top 10% of players with that skill
Significant Impact        Top 25% of players with that skill
Moderate Impact           Top 50% of players with that skill
No projectable impact     Bottom 50% of players

Here are the codes we will use for each player:

Impact Level      Power    Speed    BatAvg    PitchEff    Strikeouts
------------      -----    -----    ------    --------    ----------
Extreme            P+       S+       **         E+          K+
Significant        PW       SP       AV         ER          KK
Moderate           p        s        a          e           k

** There is no extreme level for batting average because that stat has way too much variability. If truth be told, the only player who I’d be comfortable assigning a rating of A+ would be Miguel Cabrera.

Those in the bottom 50% for each skill are assigned no rating. Their contribution is typically not enough to substantively move a team in that category’s standings, or at least not at a level that you can project. In mixed leagues, these players are usually easily replaceable. They might be more important in AL/NL-only leagues, but that does not make them any more projectable. You’re still going to want to target players with at least Moderate skill to move the needle.

The skill ratings for each player represent his true, underlying talent regardless of opportunity for playing time, level of experience or injury history. The latter two variables are accounted for on the Liabilities side of the balance sheet.

For the assessment of each of the skills categories, I return to my roots with the Baseball Forecaster and BaseballHQ.com metrics. For a fuller explanation of these gauges and complete data for every player going back many years, those are the places to go.

Power: I rely mostly on Expected Linear Weighted Power Index here. This combines weighted levels of hard-hit line drives and hard hit fly balls as a percentage of all balls put into play. So this is pure power, not colored by a hitter’s propensity to strike out.

Speed: Here I rely on Statistically Scouted Speed, which looks at run-scoring, triples, infield hits and body mass index. I also look at each runner’s track record of how often he’s been given a green light along with his stolen base success rate.

Batting Average: I use Expected Batting Average here, which looks at a batter’s contact rate and odds that a batted ball will fall for a hit, which is a product of the speed of the ball, distance it is hit and speed of the batter.

Pitching Effectiveness: Here I use Expected Earned Run Average, which approximates ERA while using only situation-independent, skills-based metrics, like strikeouts, walks and ground balls. This is similar to xFIP (Fielding Independent Pitching).

Strikeouts: I combine several metrics for this assessment – strikeout rate, swinging strike rate and first pitch strike rate (which correlates more with walks but provides some nice color).

The Assets section of the pitcher balance sheet also has a column for Saves. This is an opportunity-driven statistic but can be pared down to three levels, similar to what we do in Mayberry:

Significant    SV    Likely to get 30+ saves
Moderate       sv-   Likely to get 10-29 saves

These seem like wide ranges – okay, they are – but we need to cast a wide net in this category. The Significant saves sources are pretty much guaranteed a frontline shot at 9th inning work. The arms classified as Moderate all have some risk associated with them, from uncertain bullpen depth charts to spotty track records in a closing role. By filtering out anyone projected for fewer than 10 saves, we’re essentially saying that those guys are not projectable enough. My advice is always to speculate on relief pitcher skills and be grateful if you back into saves.

Miscellaneous: This is for any positive variables that might have a legitimate impact and are not captured in the other categories. Guess what the key word is in that last sentence?

Legitimate? Positive? The?

Close. It’s “might.” These are variables that need to be on our radar. Most analysts will build them into their statistical projection. I prefer to just identify them and let you know they might be a factor. Or not. It’s your call how important they are.

There is not much I can think of that fits here:

Pk (Positive park effect)
As noted in Chapter 2, park dimensions might have an impact on output, but changes are neither guaranteed nor can be absolutely attributable to a particular change in venue. The only players who will be noted at all are those moving to one of the more extreme hitter parks from one of the more extreme pitcher parks. The hitter parks, based on 3-year data, are Colorado, Baltimore, Houston, Yankee Stadium, Toronto, Cincy and Milwaukee. The pitcher parks are Seattle, Dodger Stadium, CitiField and San Francisco. Any other movement is ignored. Always remember The Nelson Cruz Experience – his counterintuitive improvement moving from Baltimore to Seattle last year – as evidence that this is not foolproof.

Rg (Positive regression)
There are a few players who had really bad 2015 performances, sometimes driven by no more than random statistical volatility. Odds are “last year’s bums” might see some rebound just by virtue of the planets realigning. In any case, it’s important to identify them because this is one of our few opportunities to engage in a full frontal assault against recency bias.

Liabilities

It’s great to roster a bunch of players who you hope will put up big stats. But what separates the winners from the losers is the ability to build risk into the process. Every player provides certain assets but many also have a unique set of liabilities that influence their potential to provide a fair return on your investment.

Here are the ratings we use on the Dark Side of BABS (no storm troopers allowed).

Negative Skill: The core ratio categories in Rotisserie are batting average and ERA/WHIP, and for these, a bad player can do great damage. So, rather than provide a negative rating on the asset side, we have a column on the Dark Side for players with the red lightsabers.

-AV: Bottom 25% of batting average skill
-ER: Bottom 25% of pitching effectiveness skill

Injuries: Every year, this is the one variable that wreaks havoc with our chance at success. Disabled list stays have ranged between 25,000 and 30,000 days lost in each of the past five years, so this is no small variable.

I’ve decided to take a different approach to injuries with BABS. We already know up front that about 40 percent of the top-ranked players are going to spend some time on the DL. We cannot project which players are going to pull up lame at any time, so we have to attach some injury risk to pretty much everyone.

As such, I’ve set a starting point for the health of each player. Everyone has a minimum baseline of a 25% chance to spend some time on the DL. To that, I’ll add greater odds to those players with an injury history (based on days spent on the DL over the past two years) or current health concerns.

The codes will look like this:

INJ
Players who spent more than 50 days on the DL in 2015, spent more than 30 days on the DL in consecutive seasons, or are currently hurt with an uncertain or negative prognosis for 2016. I give them over 50% odds of missing significant time in 2016.

inj-
Players who spent more than 20 days on the DL in 2015 or are currently hurt with a positive prognosis for 2016. I give them 26-50% odds of missing significant time in 2016.

I classify “significant time” as enough missed games that it hurts. If Mike Pelfrey goes down for two weeks with a hangnail and you replace him with Brett Oberholtzer, that’s not significant. And if this is a real move you need to make, you have a lot more problems than worrying about injuries.

Experience: Okay, I’ll say it – Mike Trout is a god. He is among a small class of players who hit the ground running upon promotion and never let up. (Though, if truth be told, his roto earnings for his career have been $54, $47, $38, $35. Just saying.) But most players don’t follow this path.

Patrick Davitt’s research has shown that hitters need at least 800 plate appearances to establish a baseline, or enough experience from which we can legitimately project further growth. Those 800 PAs could mean a big rookie year and a sophomore slump, or a pedestrian first season followed by a growth year, or two consistent performances. But the percentage play is to expect some volatility until that baseline is set.

So as much as we’re ready to anoint Carlos Correa as the next first-ballot Hall of Famer, there is risk, and we need to account for that. I’ve decided to err on the side of caution and increase the benchmark slightly.

On the balance sheet, we’ll identify the young players as such:

                                                      Bat    SP   RP
                                                       PA    IP   IP
                                                    -----   ---  ---
EX   < one full season of MLB experience              500   150   75
e    < two full seasons of MLB experience           1,000   300  150
-    Established players, veterans, has-beens

About 1,000 plate appearances in the Majors – two full seasons – is a good point to determine legitimacy on the batting side. In assigning ratings, I exercised some latitude here, often giving a pass to some outwardly established players who have PAs in the 900s. It’s a little more fuzzy with pitchers, but we’ll go with 150/300 innings for starters. For relievers, we’ll use 75 and 150 innings.

Essentially, anyone who gets an “EX” or an “e” is not yet a fully formed entity.

Finally, given my opinion about age, I don’t give a flying whoop whether a player reaches these thresholds at age 24, or 27, or 31. Experience is experience at the Major League level, regardless of age.

Miscellaneous: These are the negative variables that could have an impact, might not, probably won’t but can, and are definitely not quantifiable unless they are. That’s about as firm a stance as I’m willing to take. But all of these need to be on our radar because, if David Ortiz bats .230 in 2016, we need to be able to come back to BABS and say, “Aha! He’s old!”

You should really change the heading for this section from Miscellaneous to Rationalizations.

Okay, I’ll give you that.

Any of these could be bad, good or have no effect:

Pk (Negative park effect)
As on the Asset side, we can neither guarantee or absolutely attribute performance changes to park dimensions. If Jose Bautista was traded to San Francisco, he would qualify for this code, but you’d think someone with his skill would be able to hit reasonably well anywhere. So take it for what it’s worth.

Nw (New team)
This goes beyond park effects. Many players have an adjustment period when going to a new team, and especially a new league. I tend to give this more weight than others, but it’s just another variable that might have an impact. Only those players with some baseline of MLB performance are noted.

Ag (Advancing age)
Once a player hits 36, anything can happen. Some batters manage to hang on for longer; some pitchers face a steep cliff at 38. All are essentially geezers at this point. No matter how many artificial supplements some of them might be taking to ward off the fear of premature retirement, I won’t be anywhere near the bidding on players like Marlon Byrd or Bartolo Colon.

Rg (Negative regression)
As much as we want to believe that players like Jake Arrieta and Zack Greinke can sustain last year’s performances, the odds are stacked against them. Players noted here are those who posted performances so far above their historical levels in 2015 that it’s tough to justify their sustainability.

I also use this code for players whose track record has been historically volatile, at least from the perspective of their surface stats. So Alcides Escobar will get nicked here, but so will Chris Davis. Recency bias is already pushing Davis up the draft boards this winter, but do we really know whether we are going to see the 45-.260 version or the 25-.200 version?

Beyond that, feel free to add any other miscellaneous Liabilities as you see fit. If you’re worried that the Rockies will trade one of their outfielders out of Coors, then jot a note on the dark side for Carlos Gonzalez, Charlie Blackmon and Corey Dickerson. If you’re worried about potential suspensions for Aroldis Chapman, Jose Reyes and Yasiel Puig, feel free to ding them here too.

I suppose that also means you can change any of the ratings, on either side of BABS. This is your tool and I have no way of knowing what the heck you’re doing anyway.

I’ll start getting into the balancing of assets and liabilities in the next chapter, but there is basic point to remember: The more a player is lacking on the health and experience scales, and the more of these miscellaneous liabilities he has, the greater the risk of him falling short of realizing his assets. I think that goes without saying, but I said it anyway because… this is my book. But it’s your tool.

That’s it.

Hmm, I dunno. It all seems kinda simplistic and based more on opinion than fact.

Simplistic? Well, it’s simple, for sure. That’s the goal, to keep it simple but structured. However, the foundation is still based in real data. The Asset and Liability categories are all driven by data; they are just sorted into broad tiers. The secondary categories are more contextual but no less driven by fact.

So how does it work, in practice?

Players with the same asset ratings are pretty much interchangeable. However, not all players will be exactly the same; some will have more risk factors. In your roster-planning process, you’ll be making decisions as to how much risk you’d be willing to tolerate.

An example, please?

Okay. In nearly all leagues, Cy Young winner Dallas Keuchel will likely get drafted ahead of Sonny Gray (both pictured above). In the National Fantasy Baseball Championship (NFBC) ADPs as of January 1, Keuchel was going #42 while Gray was going #62. This is likely driven by their relative 2015 performances (beware recency bias) and respective team environments (which I discounted back in Chapter 2).

But here’s the thing… on a broad skills basis, both have essentially the same assets. Both have projected ERAs right around 3.00. Over the past three years, both have struck out about 7-8 batters and walked about 2-3 batters per 9 IP. If anything, Gray is more consistent than Keuchel.

So, in evaluating their respective assets, BABS gives them identical ratings. Both have significant effectiveness “ER” and moderate strikeout ability “k”. We’ll begin notating these as (ER,k). If there are Liabilities, they will be shown like this: (Er,k | INJ).

But players cannot be evaluated based on their assets alone. Keuchel also owns a potential Liability – the possibility of significant regression off of 2015’s extreme numbers. If you have to choose between the two, you might opt for Gray since his Liability record is clean and his potential acquisition cost will likely be lower.

You have full control over those decisions. BABS lays out all the facts in front of you.

Hmm. Keuchel/Grey. Blackmon/Marte. What other players are more “interchangeable” than we’d normally perceive?

Tons of them. Within the wide ranges of skills metrics and wide error bars in projections, players are not all that different. So, while they might not outwardly seem similar, Chris Davis, Lucas Duda and Matt Kemp all have the same asset ratings of (P+, a). Gregory Polanco, Angel Pagan and Francisco Lindor are all (s,a), however, each has his own injury and experience liabilities.

Yeah, but I still don’t see those groups as equivalent. Lindor is coming off a .313 season. Polanco has never batted over .260. Duda has never hit more than 30 HRs while Davis has hit more than 45 twice. Not the same.

But from a skills standpoint, the variances are not statistically significant. Duda hit more HRs than Davis in 2014. Lindor has only 390 ABs at the Major League level so you can’t consider his .313 average as a real baseline (because batting average does not stabilize until 910 ABs, remember?).

Okay, then once again, how am I going to be able to rank the players, especially without numbers?

BABS assigns each player to tiers based on:

a. Expected playing time
b. Primary assets (power, batting average, pitching effectiveness) with minimal risk
c. Secondary assets (speed, strikeouts) with increasing risk
d. Decreasing assets with increasing risk

I’ll handle the details for you once we get to the rankings sections.

But first, let’s get into the draft planning process.

Chapter 5: BABS Draft Planning – next Friday, January 29

Follow @RonShandler