A Diatribe and Confessions Regarding Player Valuations

I must be getting old. Or maybe I’m just learning. But I have a confession to make. I’m starting to change my line of thinking as it relates to fantasy baseball player valuations and team construction.

I’ve been obsessed with the process of finding projections, calculating dollar values, and relying almost exclusively on those inputs to guide my draft process. To make matters worse, after already having those thoughts entrenched in my brain, I then read Larry Schechter’s book, Winning Fantasy Baseball.

My process and beliefs sounded a heck of a lot like the ones Larry uses. And he’s arguably the greatest rotisserie player of all time. “All these chumps not calculating player dollar values out to the penny are doing it wrong“, I thought.

Well, I’m starting to soften on that stance. And here’s why…

More About Myself

I’m new here to Rotographs. And I’m still amazed that I have this outlet.

You see, I’m not really any different than you. Well, it’s highly likely that I am more obsessive about Excel than you (example 1, example 2, example 3). But I wouldn’t really classify myself as a fantasy baseball analyst or expert. It’s highly likely that I’ll never write an in-depth analysis piece on an individual player. That’s not my wheel house. It’s not my cup of tea. Every other writer at Rotographs could run circles around me in that arena.

Up until now, I’ve been strictly a “by the numbers” kind of guy. And I mean firmly in that camp. The “Paul Goldschmidt is worth $36.77 in this league and NOT $37.00″ type.

But I’m starting to feel the battleship slowly starting to change course (is it conceited to liken yourself to a battleship?).

You’re Boring Me… Get to the Point, New Guy

I haven’t missed an episode of the “Sleeper and the Bust” since before Mike Podhorzer stopped running the show. That’s a lot of Eno, Paul, and Jason. I try to listen to any podcast Todd Zola does. I’m nearly done reading Ron Shandler’s Other Book. My hair is graying and I’m a buttoned-down accountant, so half the references at Razzball are lost on me, but I love reading Rudy Gamble’s work. I’m also heavily influenced by Mike Gianella, because I can tell he loves player valuation concepts even more than I do. If there’s anyone that can claim to be a more successful fantasy player than Schechter, it’s Fred Zinkie, who I listen to regularly on the MLB 411 podcast.

So what have I learned from all of them?

There’s not a guy in that group that sticks as strictly to dollar values as I do.

Absolute Dollar Values are a Myth

This is something Todd Zola says consistently. He’s quick to remind people that the importance in rankings or in valuation is in how players compare relative to others. In other words, it’s not important that Goldschmidt’s value calculates to $36.77. It’s important that his value is behind that of Mike Trout and Bryce Harper, but is more than that of Manny Machado or Jose Altuve.

The reasoning behind this is that projections are inaccurate. Do some research on projection accuracy and you’ll surely find something about how “even the best projection systems are only about 70% accurate”. I’ve said that myself. And I don’t even know what it means.

Even though 70% accurate sounds good, that’s AS A WHOLE. It’s not at the individual player level. Here’s the 2015 Steamer projection for Mike Trout:

Is that 70% accurate? Heck, that seems darn near 100% accurate to me. There’s not a fantasy manager around that would be upset with those results. But it’s still off 11 HR and about 10 SB.

How do we even measure accuracy? Is it on a per plate appearance basis? Or is it on the final stat totals? If we’re picking nits, Steamer’s R and RBI totals are a little aggressive when you look on a per plate appearance level.

I’m obviously being ridiculous here. Intentionally. Even when a projection ABSOLUTELY NAILS IT, like it did on Trout, it’s still not “100% correct”.

And the projection systems themselves know this. If you look hard enough, many of them publish a “reliability” grade. They know they can project Mike Trout with a lot more precision than they can Francisco Lindor or Carlos Correa.

Why Projections Can’t Be 100% Accurate

Flip a coin 10 times and what are the odds it comes up heads five times and tails another five? Roll a die six times and what are the odds it lands on each of the sides once? You don’t have to be a statistician (I’m not) to understand that there’s a margin for error even with these extremely simple demonstrations of probability.

Now consider the margin for error in a 162 game season of baseball. The stats a player will put up are dependent upon many more variables than flipping a coin or rolling a die. We’re talking round ball on round bat, wind blowing in or blowing out, and depth of the outfield fences. Is the player going to get hurt? For how long? Will he be the same player when he returns? Where will said player hit in the lineup? How good is his supporting cast hitting in front of him? How good is the supporting case hitting behind him? Will his batting eye improve? Will he see more curve balls this year? Is he now in a division containing Kershaw, Bumgarner, Grienke, Cueto, and Samardzija? Or is he in one facing an aging Hamels, King Felix, and the A’s motley staff? Did the player add a new pitch, lift more weights in the off season, or suffer a personal trauma that could affect his play? Is he aging, gaining strength, on steroids, or playing for a new contract? I’ve got to stop. But you get the idea.

Could those factors lead to an additional RBI, strike out, or home run? Surely.

To surmise that projections can be extremely accurate is a bit ridiculous.

Projections to Dollar Values

We just established that we can’t REALLY rely on the accuracy of projections. But think about what we do with those projections…

After we get that single player projection, we compile it with hundreds and hundreds of other likely inaccurate projections, and then calculate very particular dollar values for each player. We then compile those dollar values into a large list that we use to draft from. And we feel guilt and pressure not to take a player for $1 more than that calculation or not to draft a player even a handful of spots before the list says.

I can’t help but feel like we’re building an extravagant home on a shaky foundation. Those dollar values and rankings are extremely useful. But what do they mean if the foundation they’re built upon gives way? Could it be that we’re just putting lipstick on a pig?

This is Not the Forecasters’ Fault

I’m not here to bash those that develop the projections. My guess is they’d surely admit that we’re using their finished product in a dangerous way.

I’m only here to bash the way that we use these projections.

Don’t get me wrong. I think it would be foolish not to use projections. And foolish not to derive some dollar value or ranking from them. Just like it’s a mistake not to at least look at ADP. These are all important tools at our disposal. I think the mistake is in the blind adherence some of us (myself included) have to them.

Another Example

The projection for Trout we looked at earlier was spot on for R and RBI. But projecting those stats accurately is highly dependent upon the projections of others nearby in the batting lineup. We don’t even know with certainty who those players will be. And the projections for those players are fraught with danger too.

Let’s assume a projection system kicked out the hypothetical result below, but the player’s actual stats for the season were going to be what you see on the second row.

Not bad, right? Looks like another nearly perfect projection. That could be the result of the projections thinking the player would hit seventh and he actually hit sixth, getting 15 or so more plate appearances during the season.

Those small changes are enough to change the player’s projected dollar value nearly $3, jump the player over three other outfielders, AND jump him 17 spots in the overall rankings! 

If you’re curious, the player in this example is David Peralta. That is his actual Fangraphs Depth Chart projection as I write this article. I doctored the “Actual” to show the effect that only subtle changes in a projection have on dollar values and rankings.

What This Means

I mentioned Todd Zola earlier. I’ve heard him allude to the example above several times, but I have never actually seen it. To see it in numbers like this is eye opening to me and I wanted to share it with you.

I used to get furious when I’d hear someone say, “I’ll go an extra buck or two on that guy.”

“You’re wrong! You’re wasting your money. Stick to the projections. Adhere to your dollar values. Maximize value!,” I’d say.

But I’m OK with going that extra buck now! Especially if what you’re really saying is, “I think the player’s projection is four runs, one home run, and two RBI too conservative.”

Do you feel strongly about a player? Do you disagree with a playing time projection and think it’s 25 plate appearances too light? Do you follow a team closely and think a player is going to bat second when everyone else seems to believe he’ll bat seventh? Do you believe a player’s skill progression was real and sustainable but see that projection systems are regressing him too heavily?

Those are all positive adjustments to a player. But the same would hold to with negative adjustments. Do you think an aging player is being given too much credit for the past and that he’s washed up? Are you concerned that poor plate discipline isn’t going to improve but you see the projections incorporating improvement?

If you have positive or negative thoughts about a player, you’re best move would be to alter his projection. But if that’s not in your skill set, just adjust players a handful of dollars and don’t feel guilty about it.

I’d be careful about subjective adjustments to things. But if you have a legitimate reason to disagree with a projection, then by all means, change your ranking for the player. And possibly change it quite significantly.

Another Issue I Have – Reliability

One of the other reasons I’ve heard people say, “I’ll go an extra buck on that guy,” is to pay for reliability. The concept of reliability within projections seems like an area that’s ripe for analysis (I hope to get into this eventually), and the idea of going an extra dollar for reliability seems like a great idea to me. At times.

You see, the other thing someone following a strict valuation approach (like a Schechter, or younger me) will tell you, is that you should always be looking for value in the draft. You want the guy valued at $24 that sells for $21. Or the guy you think is worthy of a fifth round selection that’s still there in round seven. Value, value, value. Rarely, if ever, stray.

The most strict value user will argue that you don’t pay for reliability. The reliability is baked into the player’s projection. So if you think a player is less reliable and there’s a chance he bombs out, that low floor for the player is built into the weighted average probability that make up his single line projection.

But I’m not so sure that makes sense. We’ve all heard the phrase,  “You can’t win the draft in the first round but you can lose it“. And I think there’s some truth to that. If you’re a believer in that, you might assign a premium to Mike Trout and/or discount Carlos Correa and maybe even Bryce Harper. I’m not much of a poker player, but I understand enough to know that I want to leave myself with as many “outs” (no baseball pun intended here) as possible. A safe reliable player offers many different paths to a championship and doesn’t close any doors. A player with a riskier floor might close doors.

Yes. All positive and negatives scenarios are supposedly included when a projection is developed. But just like we gain by breaking baseball stats into smaller components (home runs into HR/FB and FB%), I think we need to consider the benefits of breaking that projection down into its own smaller components. Reliability should be incorporated into our valuations. Right now, it’s not.

Value Versus Team Construction

The last confession I have to make is regarding this struggle we face between trying to maximize the overall value of our team and in trying to build a balanced team capable of competing in all rotisserie categories. Yes, some of us play in NFBC “no trading” leagues. But I’m not even referring to those right now. I’m just talking about internal dialogs like, “The most valuable player on my draft board, that I spent hours developing and tailoring to my league, is Carlos Gonzalez. But I’m short on stolen bases and I’m afraid that commodity will dry up quickly. Should I take Gregory Polanco instead?”

The strict value user would take Cargo. “Value! Why’d you make that spreadsheet if you weren’t going to use it?”

But if one is valued at $15.50 and the other at $15 on that spreadsheet, does that really mean anything? We saw how flimsy those valuations are earlier. That $0.50 difference is literally one stolen base in value. If Cargo steals three instead of four or if Polanco steals 28 instead of his projected 27…

I can’t prove this. But when we go back to this concept of team flexibility and leaving ourselves “outs”, could there be value in having a more balanced team, even at the expense of not maximizing value in the draft? I think so.

I hear the counter-argument. “But we can trade in this league. If I maximize my value now, I’ll just trade away my excess value and get those steals during the season.”

Eh. Does it really work that way? If we’re talking college microeconomics text books, marginal products, and marginal costs, yes. But Adam Smith is not in any of my leagues. What if the guy with surplus steals is the one I stand to pass in the standings? What if I don’t happen to have the stats or positions the manager with surplus steals is looking for? What if the guy with surplus steals stops paying attention in July because he’s out of it? What if the guy with extra steals is a giant Richard and overvalues all of his players?

Fantasy baseball trading is the free and efficient market it’s made out to be. I HATE trading. It’s excruciating at times.

If you’re Fred Zinkie, assuming you can make a deal is a safe bet to make. If you don’t think you’re a good trader or you know it’s difficult to make trades in your league, place a priority on balance over value.

Final Thoughts

I may have been a little rough on those using a strict valuation approach to make decisions. But I used to be one of them. It’s difficult to break ties!

What I’m learning as I age, and hopefully become wiser, is that just like with anything else, the key to success seems to be balance. I’m not talking roto category statistical balance. Balance in the sense that using a “part valuation approach” and “part qualitative decision making approach” may be optimal (am I the only one seeing the “stats versus scouts” similarities here???).

Use projections. Calculate your own dollar values. And when it makes sense, make decisions from that data. But understand the limitations and don’t sweat it if you think your best choice isn’t truly supported by what the spreadsheet says.

Your comments are extremely welcome below. Challenge me. Agree with me. Let’s have a respectful and productive discussion.