This article is part of our The Z Files series.
One of the allures of auction drafting is everyone can roster every player, so long as they're willing to pay the price. The question is, when is the price too much? When does the cost exceed production, yielding a negative return on investment?
Many likely believe it's as simple as converting the player's projection into projected dollars earned, thus anything over that number is overpaying.
Others will agree, with the caveat they don't need their best players to return profit. All they want is for the top players to provide a stable foundation, confident they'll find ample profit elsewhere.
A few will counter, contending that you're going to get the lower-end profit regardless, but by overpaying at the top, you don't enjoy the full extent of the profit.
Another faction might suggest the same player may be worth something different to each drafter, depending on their overall strategy and team needs at the time the player is nominated.
In my 20-plus years of playing, and discussing this game, I have read and heard iterations of these arguments countless times. However, most seem to overlook an important factor. Regardless of strategy, format, needs, etc., irrespective of whatever qualifier you want to put forth, overpaying isn't a function of paying more than the player is worth because there's no way of accurately establishing what a player is worth.
This goes beyond a simple truism like, "Projections are a range, not a static number." Even if we magically knew exactly how every player
One of the allures of auction drafting is everyone can roster every player, so long as they're willing to pay the price. The question is, when is the price too much? When does the cost exceed production, yielding a negative return on investment?
Many likely believe it's as simple as converting the player's projection into projected dollars earned, thus anything over that number is overpaying.
Others will agree, with the caveat they don't need their best players to return profit. All they want is for the top players to provide a stable foundation, confident they'll find ample profit elsewhere.
A few will counter, contending that you're going to get the lower-end profit regardless, but by overpaying at the top, you don't enjoy the full extent of the profit.
Another faction might suggest the same player may be worth something different to each drafter, depending on their overall strategy and team needs at the time the player is nominated.
In my 20-plus years of playing, and discussing this game, I have read and heard iterations of these arguments countless times. However, most seem to overlook an important factor. Regardless of strategy, format, needs, etc., irrespective of whatever qualifier you want to put forth, overpaying isn't a function of paying more than the player is worth because there's no way of accurately establishing what a player is worth.
This goes beyond a simple truism like, "Projections are a range, not a static number." Even if we magically knew exactly how every player will perform in the impending season, conventional valuation is flawed. The methodology doesn't matter. Be it standings gain points, standard deviations/Z-square, or the percentage value method, none generate a number commensurate with the player's production. If there isn't a number we can compare to the purchase price, how do we know if we're spending too much?
The primary flaw with valuation has been discussed in this space a few times this offseason. Conventionally, the draft-worthy pool consists of exactly enough players to populate a legal roster on each team. The player's worth is computed based on that pool of players. However, unless your league is a pure draft-and-hold with no moves, far more players will end up being active over the course of the season, contributing to the stat base.
Sure, maybe based on previous seasons in your league's format, estimating the season-ending stats is possible, which could then be used as the basis for valuation. The problem here is there will be more players with positive value than can be purchased on draft day, lest money is left on the table. So, now, prices on the better players are inflated in order to spend your entire budget and we're back where we started, with just enough players purchased to be on a legal roster, costing more than they'll produce.
That last sentence brings up a good point. What does the money spent on draft day represent? It isn't commensurate with what the player is expected to earn but rather what it takes to buy him relative to other players. If this is the case, overspending is more about paying more for the player's expected production relative to the other positive contributors, because none will be "worth" their price.
Speaking of other positive contributors, here's another flaw in valuation. Each auction will have its own pool. There will be a ton of overlap, but the endgame of each will consist of different players. That said, this primarily affects just the endgame players. The aggregate stats of this group will be close in each draft, at least close enough that the influence on the top players is mostly round-off.
I prefer not to make a big deal of this now, but there's serious theoretical flaws with SGP (Standings Gain Points) and using standard deviation or Z-square methods. In brief, the huge foible in SGP valuation is you can't gain a point until you've surpassed the category total of the last place team, and that barrier isn't proportional between categories. The easiest example to illustrate this is homers and steals. The SGP of both are fairly close, but it takes many more homers to be second-to-last than it does steals. The effect is some of the SGPs of a speedster go towards getting over the home run barrier, depriving the stolen base specialist of "value". However, because the market often opts to discount steals, the price is a serendipitous match to the incorrect values generated from SGP. The issue with standard deviation or Z-squares is that it may be elegant math, but there's no basis for it serving as a proportional proxy for fantasy production.
For the record, I favor the percentage value method, where players are credited in proportion to their categorical contribution. That is, budget is allocated to each category and distributed in proportion to how each player contributes to the category, after accounting for replacement level. It's not perfect, since the gap between places in the standings isn't linear, but at least each player's production is accounted for on a category-by-category basis and not lumped together like SGP. For me, it's the lesser of three evils.
A little talked about deficiency is comparing ratio stats to counting stats. Regardless of the valuation method, ratios need to be converted to a counting-stat base, with the number of innings or at-bats part of the process. There are a handful of mathematical treatments, all of which make sense, but none of which are perfectly accurate. This goes beyond the scope of this discussion, but the same ratio over the same number of at-bats or innings can help one team while hurting another. It's more than just "anything better will help and worse will hurt". It has to do with the extent by which the player's numbers carry weight relative to the team totals. The most basic example is two fantasy teams with the exact same ERA, but one has 100 more innings pitched. Adding the same pitcher to both teams will result in a different final ERA.
The final valuation shortcoming relates to the hitting/pitching split. In a standard league, each team begins with 260 auction units. Theoretically, 130 should be dedicated to both hitting and pitching. Most teams spend between 165 to 185 of their units on hitting. Some approach or eclipse 200. By itself, this skews the comparative measuring stick. If projections and valuations were perfect, a $35 hitter is really a $26 hitter while a $35 pitcher earns $58. That said, if projections and valuation were perfect, the budget would be distributed equally.
Taken individually, the imperfections aren't a big deal. En masse, they render a number not to be trusted. A $40 player isn't a $40 player. He isn't a $37-$43 player. We don't know what he is. Ergo, how do we know spending $45 is too much or buying for $35 is a bargain? Truthfully, we don't.
The proper approach is spending $40 to buy Christian Yelich's or Jacob deGrom's stats. It isn't a matter of whether either return $40, but rather could you have purchased more by distributing the $40 differently? The overly simplistic example is adding the stats of a $1 player to the $40 player and comparing to the production of a $21 and $20 player.
On paper, the numbers should be equal. However, some experienced auction players can manage their money in a matter that extracts more production from the $21 and $20 players. That said, no matter how experienced you are, there will be instances where fewer stats are purchased for the same 41 units. Some auction players are more adept in the middle game while others are uncomfortable, preferring to bank high-priced solid production and not risk giving it back later. For one, spending $53 on Mike Trout is smart. For someone else, it's foolhardy.
The same line of thinking extends to keeper leagues. In fact, it's paramount to approach auction spending in this manner. It's not a matter of paying the properly inflated price; it's whether you could have bought more stats elsewhere.
In keeper formats, there's more aggregate budget available than talent, at least relative to what would be paid in a redraft scenario. Conventional treatment is to express the excess as a percentage. If there's 25 percent more money available to be spent than projected earnings, the inflation factor is 25 percent. In order to ensure everyone spends all their money, the expected price is multiplied by 1.25. This is the so-called inflated value.
This isn't what happens. A $1 player remains a $1, not a $1.25. A $40 player usually sells for more than their inflated price of $50. Inflation isn't applied linearly. More assets go to the top. While it's possible to model spending by re-running values after every purchase, adjusting the player pool and available money, it's still a model and not representative of what will ensue. The prices may be closer, but they're still a guess. The model can't account for strategic choices such as perceiving it's OK to "overspend" with a great keeper list, or paying what it takes to get a stud to later trade for future assets.
Just as before, it's not as simple as taking a $45 Mookie Betts, subjecting him to 30 percent inflation to land on a $59 price, and thus deciding $56 is a good deal. The correct approach is, "Can I distribute that $56 elsewhere and buy more stats in the aggregate?"
Be it this weekend's Tout Wars auctions or post mortem discussion of your own leagues, don't get caught up in the notion of someone "overspending" on a player. Instead, focus on if/how that budget could have been better spent elsewhere.
