RSOB Chapter 3: The Broad Assessment Balance Sheet

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

---

For decades, we have been told that the goal in fantasy baseball is to assemble a group of players whose aggregate statistics exceed those of all the other teams in the league. In fact, that is the actual verbiage in the Official Rotisserie Baseball League Constitution.

But we don’t know what statistics our players are going to put up until after they’ve done it. Right? Right?!

Yeah, yeah, yeah, I remember. Still not sure I buy it completely, but I’m listening.

Can we at least agree that we don’t know the exact numbers players are going to put up and the ranges around those projections can be very, very wide?

Sure.

Are you comfortable with the idea that a better approach might be to only plan around the variables that we do know?

I suppose.

Good. We do know each player’s historical skills profile. We have a general sense of each player’s role. And we know the potential risk factors that will ultimately color the numbers.

Our fantasy team is a collection of these skills, roles and risks – each player’s assets and liabilities. But for as long as we’ve been playing this game, we’ve been going into our drafts just trying to accumulate the most projected stats.

Players are more than just a bunch of projected stats.

Take Johnny Cueto. When you draft him, you are not just getting a frontline starter with a three-ish ERA. You’re also getting volatile win totals, the risk of a new ballclub culture in San Francisco and the uncertainty about last season’s Kansas City Stumble. When you draft Giancarlo Stanton, you’re not just getting the potential for 35-plus HR. You’re also getting a wide error bar around those home runs because there is a long history of injury risk.

But aren’t all those variables built into the projections?

Most of us touts attempt to do that, but how do you quantify risk? The adjustments we apply to the projections are often just arbitrary – we’ll lop off a bunch of AB or IP to account for how long we think a hurt player might be out, or we’ll make subjective decisions about the qualitatives. How do you account for the Cueto and Stanton risks? There’s little science behind it yet we’ll be drafting our teams off whatever numbers are on our cheat sheets.

Let’s look at Stanton a little closer.

It is acknowledged that he is one of the best pure power hitters in baseball. However, in six major league seasons, he’s managed to stay healthy for an entire year just twice – in 2011 and 2014. Last year, he hit the DL with a hand injury in June and never came back.

Stanton amassed 539 AB in his healthy 2014 season, a career high. The Forecaster attempts to account for the injury risk by hedging with a 490-AB projection for 2016. There are other sources that take a leap of faith and project a full healthy year, often forecasting even higher AB numbers than he’s ever posted. Wishful thinking, perhaps?

But Stanton is not without risk. Despite positive health reports in December, his historical health track record does not instill confidence that he can get through a full season injury-free. You cannot dismiss the possibility that he might miss some time even if he is perfectly healthy on Opening Day. But you also can’t arbitrarily decide how much of a playing time discount to project. Even if you buy into a 550-AB projection – or a 490 AB hedge – his stat line doesn’t tell you anything about the risk itself.

By combining disparate variables into a single projected stat line, you lose the ability to distinguish the skill from the risk.

We need a way to keep everything separate. We need to be able to present Stanton’s true underlying skills without making assumptions about his risk factors because, well, there is a chance that he does stay healthy all year and we want to see what that might look like. But we also need to present those risk factors so you can draw your own conclusions about how important they are to you, if at all.

As we’ll soon see, Stanton’s underlying skills put him in the same class of players as Bryce Harper, Nolan Arenado and Josh Donaldson, as they should. But risk is what sets him apart. You simply can’t build that into a statistical projection and claim it’s more accurate.

Consider… a balance sheet. That’s something we’ve never done – we’ve never viewed our players and rosters as balance sheets. We may have kept running totals of projections – our assets, sort of – but we rarely kept a record of liabilities. It’s the balance of assets and liabilities – on both a player and team level – that provides a truer view into our team’s potential for success or failure.

Maximizing assets, minimizing liabilities. That’s how we are going to build our rosters. The process is one of planning out your optimal cross-section of skills while deciding up front how much risk you are willing to incur. The players then become just puzzle pieces.

The Broad Assessment Balance Sheet (BABS) is the formal moniker that I’ve dubbed this process. It’s broad because we’ve already determined that “precise” doesn’t work. It’s an assessment – slightly less rigorous than a full-blown analysis because complexity doesn’t buy us enough to make a difference. It’s a balance sheet, because that is what the output of our effort is going to look like. And I want you to become fast friends, so let’s just call her BABS.

(If nothing else, BABS finally gives us a strong female presence in this hobby, at least one who knows her way around a lightsaber.)

So we start with a balance sheet. What do we put into that balance sheet?

Six years ago, I developed a process called the Mayberry Method. It reduced each player to a 7-character code: three characters for skill (on a scale of 0-5), one character for playing time (0-5) and three characters for risk (A-F grades for health, experience and consistency). As much as that was a huge step in the right direction, now I believe that it doesn’t go far enough. It’s still too granular.

Here was my introduction to the Mayberry concept in 2010:

“Tonight, the friendly weather forecaster on my local television station has told me that it is going to be partly cloudy tomorrow with a high of 78 degrees. I suspect the meteorologist’s advanced modeling system spit out that fancy number – 78. I often think, why not 77? Or 79? The truth is, if I were to walk outside right now, I’d feel no difference if it was 77, or 78, or 79.

In fact, it probably requires a good five degrees for me to feel any noticeable difference, and even then, it would be slight. 79 versus 74? 46 versus 41? 97 versus 92? More important, a five degree difference wouldn’t likely make me change my behavior. If I’m not wearing a light jacket at 79, I’m not likely going to do so at 74.

The 10-day forecast is an even more interesting exercise. Besides the fact that I don’t believe they can accurately tell me that it is going to rain a week from Sunday, the list of daily high temperatures seems to be an exercise in excessive precision: 80, 82, 81, 82, 80, 77, 77, 77, 74, 76.

What does this tell me? The first half of the week is going to be warm. The second half of the week is going to be marginally cooler.

In fact, they could just say that the temp will be in the low 80s and I would be perfectly okay with that. High 70s, low 80s, high 80s, low 90s… that’s all I need. They wouldn’t even have to bother with mid-70s or mid-80s because that won’t change what I am going to wear anyway.

What do we gain from the extra precision? We delude ourselves into believing we are gaining accuracy when in fact we are gaining an increased probability of being wrong. We’re just not good enough to predict the temperature to the exact degree on a daily basis. And most important… there’s no great need to be so perfect.”

Now let’s take this a step further.

What if we were to say the only thing that is important is the climate’s affect on what we wear? It doesn’t matter if the temperature is 82 or 95 because in either case, we’re heading outside in shorts and sandals. It needs to get cooler than 65 before we consider donning a light jacket, but 64 versus 54 is nearly irrelevant. And we won’t consider pulling the parka out of the closet until the temps dip into the low 40s.

Now, the range of temperatures that have any actionable consequences becomes quite wide. It’s shorts weather, light jacket weather or parka weather. Any number attached to the thermometer just doesn’t matter.

(Interestingly enough, when I lived in New Hampshire, I felt quite comfortable in shorts when temps were in the 50s. Now in Florida, a jacket comes out when temps are in the low 60s. I suppose that is the climate equivalent of park effects.)

With BABS, each skill – tied to a standard fantasy stat – is going to have an extreme impact on your roster, a significant impact, a moderate impact, or none at all. Power, speed, strikeouts, et al – these are all building blocks. The distinctions between impact levels are based in real skills analysis but in very broad strokes.

So what we will be putting into our balance sheet are descriptors of each player’s skills – and later on, risks – in these broad terms.

Wait, no. Sorry, that doesn’t work for me. Let’s say I have a choice between Charlie Blackmon and Starling Marte – two speedy guys. But Blackmon stole 13 more bases last year. Are you telling me I can’t rank Blackmon ahead of Marte for speed potential?

It’s convenient that you picked these two players (pictured above). Colorado’s Blackmon and Pittsburgh’s Marte both have significant speed skills as compared to the rest of the player pool. Both have batting average skills that are comparable. And both are clean on the Liabilities side. In the eyes of BABS (they are a beautiful shade of blue, if you were wondering), both players are essentially interchangeable commodities. Draft one, draft the other. The odds that one will outperform the other are not significant enough to project with any confidence.

C’mon, really?

Really. You cannot tell me with 100 percent certainty that Blackmon is going to steal more bases than Marte in 2016. You can think that Blackmon has better speed skill, but there are too many variables that need to align for you to guarantee a precise variance in stolen base output between those two players. If Blackmon regresses even a little and Marte improves – not unreasonable possibilities – then the difference between the two is inconsequential and certainly not projectable for your roster-building purposes on Draft Day.

Bottom line – your opinion that Blackmon is going to steal more bases than Marte is heavily steeped in recency bias.

You can put money down that Blackmon will steal more bases than David Ortiz (okay, pretty obvious), and it’s also a reasonably good bet that Blackmon will steal more bases than Brad Miller… but even that is not a 100 percent slam dunk, no matter what their respective skills sets look like NOW. (See what I did there?)

Ha, ha, funny. So how do I decide what to pay for them? If I’m in a snake draft league and they both fall to me, I still need to decide who to pick. Do I flip a coin?

You could. If you need a tie-breaker, you can look for some minor variable outside the balance sheet – Blackmon’s ballpark, Marte’s team, whatever – if you need the comfort of giving one player an edge. But in the end, it won’t likely be enough to make a difference to your team’s eventual success or failure. The error bars are too wide.

Here is another way to look at it. Let’s say you can’t get it out of your head that Blackmon is a better player. Let’s say that someone ahead of you grabs him in a snake draft or outbids you in an auction. If Marte is still available, feel comfortable knowing that you’ll have another shot at landing a Blackmon-esque commodity. And if the cost is lower, you’ve just gained some profit.

So, we’ll be describing each player’s skills profile in broad terms on the Assets side of BABS. The risk variables will be handled likewise on the Liabilities side of the ledger. The primary risk categories are health, experience and ratio downside (batting average and ERA/WHIP), plus a few miscellaneous factors.

To show how that works, let’s add two more players to the Blackmon/Marte mix. A.J. Pollock and Mookie Betts both have the same skills profiles as Blackmon and Marte. Pollock is risk-clean like our original duo. Betts, on the other hand, has a mark on the Liabilities side – he is still short on Major League experience and thus is at risk for more variability in his output. That would rank him slightly below the other three.

In the next chapter, we’ll start providing some structure to BABS. Although we don’t care about figures, you’ll see that she’s still pretty well built.

Sorry, low-hanging fruit.

In Chapter 4, the BABS Player Profiling System…

Follow @RonShandler