The Obsession With Tiers
We humans love to categorize and label things. Are you an extrovert or introvert? Does your favorite band play rock or pop music? Is that movie you saw last night a drama or romcom? Are pancakes a breakfast food or the best dinner ever? So it’s no surprise that this infatuation with ensuring everything fits into a box has spread to fantasy baseball. But I just don’t get it.
In fact, I think creating fantasy baseball tiers is a silly waste of time that provides no additional value and there are numerous reasons why. Let me explain.
Tiers Are Subjective, Arbitrary, and Random
Let’s begin with a table. The below values are calculated using the default FanGraphs Auction Calculator settings.
| Player | Dollars |
|---|---|
| Chris Sale | $48.90 |
| Max Scherzer | $39.70 |
| Jacob deGrom | $39.00 |
| Justin Verlander | $33.20 |
| Corey Kluber | $27.70 |
| Gerrit Cole | $26.60 |
| Carlos Carrasco | $26.30 |
| Luis Severino | $25.90 |
| Clayton Kershaw | $25.50 |
| Blake Snell | $24.90 |
| James Paxton | $22.00 |
Now, being the tier enthusiast you are, you would like to group these pitchers into tiers. How do you go about it? Does Chris Sale, projected to be worth $9 more than the second highest valued pitcher, deserve to sit in Tier 1, all by himself? Does that make Tier 2 double the size, composed of just Max Scherzer and Jacob deGrom? Do we have another one man tier at 3 with Justin Verlander? Then where do we draw the line from Corey Kluber on down? Do I drop down to $25 to include Clayton Kershaw, but exclude Blake Snell, who happens to sit just 10 cents below my arbitrary cutoff for Tier 4? Why not cut off Tier 4 at $26 and stop at Carlos Carrasco? Does James Paxton begin a new tier, even though SP12 is projected for $20.60, meaning there’s actually a small gap between Paxton and the next pitcher?
If I sent this table to 10 different fantasy owners requesting it be returned with clearly delineated tiers, I would likely get at least five different sets. That owners would create varying tiers when we already know our projected dollar values proves how subjective the exercise is. Succeeding in a draft/auction requires minimizing subjectivity.
Tiers Remove Information, Thereby Reducing Accuracy
We can all agree that the draft/auction plays a significant role in the ultimate success of our fantasy team, right? If an owner somehow awoke one morning with the power to perfectly see the future and learned exactly how every player performed during the upcoming season, he would win the league, right? Of course. That’s because his projections were 100% accurate since they weren’t really projections, but actual results known before the season began.
So if we agree that arriving at the draft/auction with the best projections that have been converted into dollar values provides an enormous advantage, why would we then perform an exercise that removes information and decreases the accuracy of our rankings by reducing our player list into tier groups? How can we make the best decision with less information?
Let’s look at a restaurant menu. First, a standard menu with prices.
| Item | Price |
|---|---|
| Steak | $29.00 |
| Fish | $32.00 |
| Chicken | $18.00 |
| Pasta | $20.00 |
You’re a budget minded diner and therefore pass on the more expensive steak and fish dishes. You decide to opt for the pasta, hoping your $20 gets you a perfectly al dente plate that reminds you of your trips to Italy.
But what if instead this restaurant listed price tiers instead of specific prices?
| Item | Price Tier |
|---|---|
| Steak | $20s |
| Fish | $30s |
| Chicken | $10s |
| Pasta | $20s |
Well hey, the steak is priced in the same tier as the pasta! Pasta isn’t typically expensive, and I love steak, so I’m going to order the steak since being in the same tier suggests a similar price.
Oops, you just spent $29 on your entrée, resulting in a final bill well above your budget. You are not happy and vow to never return to a restaurant that lists price tiers, rather than specific prices. How can you make an informed menu decision without complete information?!
The menu example is a perfect analogy for the use of tiers. Of course, I picked an extreme one as it’s unlikely that you choose to group a $20 player with a $29 player in the same tier. But there are differences in projected dollar values between players in a tier that are washed out when these players are grouped together.
You Can’t Compare Players Across Positions With Tiers
Tiers are generally created across positions. So you have the catcher tier, first base tier, starting pitcher tier, etc. But if you arrive at your snake draft with a bunch of position tiers, how do you know which position to draft for your pick?
Let’s compare an arbitrary Tier 1 for catcher and shortstop:
| Player | Position | Dollars | Player | Position | Dollars |
|---|---|---|---|---|---|
| J.T. Realmuto | C | $23.90 | Francisco Lindor | SS | $30.80 |
| Gary Sanchez | C | $23.80 | Trea Turner | SS | $29.10 |
| Salvador Perez | C | $21.20 | Manny Machado | SS | $28.70 |
| Buster Posey | C | $20.80 |
With only a set of tiers at your disposal, how do you decide whether to draft a catcher or shortstop next if all players are still available? Since all seven players belong to a Tier 1, they should be interchangeable, right? What if only Buster Posey and Manny Machado, representing the last of Tier 1 for each position, are still left on the board? Obviously with dollar values, the choice is simple. But tiers aren’t nearly as specific, as they merely group names together. The assumption when using tiers is that the same tier in each position is roughly equivalent, so all Tier 1s are composed of similar valued players. That’s definitely not the case in this example and likely not in the vast majority of leagues.
So without specific dollar values attached to those player names and only a tier number, how can you possibly make the best decision for your next pick? The answer: you can’t. Don’t make the mistake of thinking you can draft the steak at the same price as the pasta like your tiers are telling you.
Challenge: Convince me that tiers without attached dollar values are worth creating. If you include dollar values with your tiers, convince me that adding a tier number provides additional benefits beyond what those dollar values already provide.
Think you can beat a Tout Wars Champion? This is your chance! Join the “Beat Mike Podhorzer” league, an NFBC 12-team mixed format eligible for the $125,000 grand prize. The online draft takes place on Fri, Mar 8 at 9 PM EST. Full details here.
Mike Podhorzer is the founder of ProjectingX IQ, an advanced fantasy baseball analytics platform that transforms projection data and in-season performance signals into actionable intelligence. He is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.
I have $5 tiers and having them at the draft (and even before…with projected/announced keepers) really helps me identify where big gaps exist. Thus, I am not comparing pasta with steak.
I also tend to place the players in the tiers by where they fit in the tier (i.e. the name of the $20 player is at the top of the tier). Later in the draft, this helps me quickly identify where value exists.
Binning, which is what you are talking about in the article, has a useful bin size in a particular problem. Too coarse (i.e. bins of $20) and you really have the steak/pasta issue. Too fine ($2 bins) and you have not really helped yourself much. For me, $5 bins seem to be the sweet spot.
If Chris Sale is in a bin by himself, he’s earned it.
Why would you use tiers to identify where value is if you already have dollar values?
I imagine that beyond the first couple of rounds or so, the projected dollar values are pretty close together and also come with decently sized (and varyingly sized!) error bars. Makes sense to group them into chunks then, although I agree more information is better, so include the projected dollar values too.
Yeah, that second point about less information is pretty strawman-y. I can’t think of anyone who uses projected values to list players in tiers and then deletes all the projected values only to replace them with $20s $30s $10s.
Your posts are always a great read Mike. Although an amateur, I have independently arrived at the same conclusion as you have over the years for the most part. I do look much more now at where there are larger gaps in projected dollar values using SGP. and make notes on my paper bid sheet (being a bit old fashioned). For example, I note a significant gap at 1B this year following Goldschmidt, Freeman and Rizzo. Other than that sort of modification I am right there with you.
Amen Mike!
I have always hated tiers for both fantasy baseball and fantasy football. It over-simplifies the process and leaves owners shooting themselves in the foot.
Amen. Tiers are a terrible tool to use during a draft. They suffer the same flaw as the way most people treat position scarcity; comparing players to an arbitrary point other than the level of a replacement player. It doesn’t matter where a player ranks compared to the average at his position, or vs the best at his position, or within any tier. What matters is how much better he is vs a player that can be had for free. This is why catchers are the only “scarce” position, and why tiers make no sense.
In a standard 5×5 Roto league you can still draft a losing team even if you draft a team with the highest projected dollar value. You could easily draft a team high in $ value without much in saves or steals, good luck winning. It’s all about drafting a balanced team, you have to look inside the tiers and inside the $ values to see what you are actually reasonably projected at getting stats-wise across the categories.
“Convince me that tiers without attached dollar values are worth creating.”
I know that I won’t be able to convince you. But some benefits would be:
Helping you avoid errors related to false precision.
Simplifying your thought process.
Helping in the mid and late parts of a draft.
Saving time in pre-draft preparation.
Simpler is not necessarily better. Ranking them alphabetically would be even simpler, after all.
And I don’t see how doing extra work inherently helps you in the pre/mid/late part of the draft. If it leads you to take a worse player because it was the last guy in a tier, then that extra work did more harm than good.
I DON’T use tiers, but to play devil’s advocate, I could see two benefits:
1) We know projections are far from accurate, that each player has a significant range of outcomes. Throwing similarly-valued players into one pool explicitly removes our (potentially misplaced) confidence in our ability to distinguish between the $29.10 for Turner and the $28.70 for Machado.
2) As a shortcut, i.e. to avoid needing to have an opinion on every player. “They’re all about the same, I’ll take whichever falls to me”. You could look at this as lazy, or less pejoratively, as a necessity for those who don’t have the time to deep dive the player pool.
1) But Josiah, do you really need a tier to remind you that a $29.10 player is virtually the same as a $28.70 player? The dollar values are all you need, grouping them into a tier doesn’t actually add any value.
2) How do you know they are all about the same if you don’t have an opinion on the player? You need to form that opinion, backed by a projection (whether mental or on paper) converted into a dollar value (whether calculated or ballpark), before you could determine two players are similar.
Yeah, I don’t use ’em. Just like to argue with you 😉
I know that personally, I spend so much time in the pre-season working with my Excel file that I have to be very careful not to over-weight my confidence in the resulting valuations. If someone made the decision to force themselves to acknowledge the volatility via tiers, I wouldn’t argue too hard.
It would work if players were assigned to tiers based on number output+skill. Would help show you if you’re getting value or over-paying (obviously determined by comparing adp within tiers).
All your tier examples come from the top of the auction or draft, where you should be working from individual value. Tiers are more useful at the end of auction where everyone has a value within a couple bucks. A player I like who I’m targeting for a buck or two may be bid up to 2 or 3 dolllars and I’ll pass because I know I can get a player from the same tier or asset class for 1 dollar later.
Who uses tiers without dollar values also attached? I “tier” based on large gaps between players based on my SGP $Values. I want to know when the bottom falls out of a position mostly. Like most have said so far, it’s nice to group similar players together for the mid/later portions of the draft.
If you don’t have any strategy for the beginning of your draft/auction, you are probably using a generic ESPN/Yahoo! cheat sheet anyways. I feel like if you are reading this article, you probably know to put as much information that you can easily decipher onto your custom cheat sheet as possible. (Position/$$ value/ADP/upside, etc).
Lots of people who are doing drafts instead of auctions don’t bother to generate dollar values.
Then they deserve to lose 🙂
Agreed, can I get those people into some of my leagues?
They should. It’s easy. And way more accurate than doing tiers.
I keep those ESPN and Yahoo cheat sheets handy since most of the people in the room will be using them or some equivalent magazine. That way I can more accurately predict how the others might be drafting, as opposed to using it on my own.
That’s too many extra papers for me at the draft. I will typically have whatever site’s dollar value/ADP in a column on my spreadsheet to try and find the bargains. I’ve found that what people spend/when they draft someone is so totally dependent on the Site that is hosting, since most people won’t do the ranking work themselves. The default ranking vs. my rankings is where I tend to extract most of my value. I could probably tell you 2-3 of my last picks in a draft based on the default rankings on a Site.
I didn’t mean that I keep all of them. Just one, mainly the ESPN because it concisely fits like 300 players onto one page, and I’ve noticed that others keep it for that very reason. In fact I’lll just write my own values on that page so I don’t have to cross reference with my own spreadsheets and whatnot the entire time.
“I want to know when the bottom falls out of a position mostly”.
Dollar values will tell you this better than tiers will. When the players are worth <$3 the bottom has fallen out. Or pick whatever point suits you. Replacement level is the same for every position except catcher; reaching on a particular position because a tier is ending is a bad decision.
1) Sure, more is better unless there’s information overload (which some people would have a harder time with than others).
1a) I get the appeal of tiers with a draft, as it’s a simple reminder to focus on drop-off and to avoid buying a player/position early if you can wait a round.
2) I would argue clustering similar-type players (stat-wise) would be more useful than clustering by value. This would focus the owner on building a balanced team which is as important as optimizing your dollars.
3) One of the methods I use to generate dollar values uses simulation (ie simulate teams and standings, repeat, weight each player’s contribution). Converting the “contribution” from each player into tiers is a good way to collapse the uncertainty (by uncertainty I mean the confidence intervals created through simulation).
Before trying to convince anyone of the value of tiers, I think there’s an underlying flaw to each of the three arguments presented here; the fallacy of the false alternative. For me, tiers are not in place of any of the information you have provided. They are in addition to it. For instance, the menu example is nonsense. Tiers don’t remove the “prices;” they simply supplement them. Off point, but the comparison is also faulty because with food you are making a single selection (or two), whereas with a draft you are constantly weighing opportunity costs involving analysis of what you are likely giving up by passing or selecting a particular position.
I find the “subjective/random” argument goofy. OF COURSE we view things differently. That’s why we play. If we didn’t, every draft would follow ADPs and we may as well go home. Every single method of deciding which player to take in a draft “suffers” from the precise “shortcoming” alleged here: we view things differently. Whether or not to group Cole and Kershaw together, or where to put them, given their different risk profiles, is the essence of fantasy baseball. Somewhere, somehow, every single person who drafts will be called upon to make a subjective evaluation of where to take a player. Tiers are no different.
“With only a set of tiers at your disposal, how do you decide whether to draft a catcher or shortstop next if all players are still available? Since all seven players belong to a Tier 1,they should be interchangeable, right?”
Nope. Not at all. Not even close. Again, the fallacy of the false alternative. The argument PRETENDS that you can only use tiers in a vacuum – that you can’t look at dollar value, scarcity, age, injury risk, playing time risk, or anything else in the decision-making process. Of course if five years from now if I have the top catcher (TIER 1!) projected at $9, and the next 15 catchers “(TIER 2s!) at $5, and the top OF at $60, and the second best at $5, I can choose the OFer. The idea that once you draw a few lines on your draft sheet you can no longer use any other tool known to man is silly.
So, the case for tiers: admittedly it’s one little piece of information IN ADDITION to any and every single other piece of information out there. What it does is forces me to evaluate where I think cliff-like drop offs in talent occur, and to use those, with ADPS, to roughly plan my draft.
Example? I personally see a drop off after about eight catchers, and there are two far ahead of the next six. I’m in a single catcher league. That helps me a lot. So, my loose strategy going in is to nail one of the 3rd through 8th best catchers, preferably at the tail end of the group (I’m largely indifferent whom).
In draft, the tiers allow me to more quickly track catchers 3-8 and to time my selection of that position to be as late as reasonably possible without falling off what I see as a cliff dive in talent.
To me, the tiers help even more with closers.
ADP, dollar values, tiers, a running total of your projected stats, BABS, Mayberry scores, whatever you use; they are all simply pieces of information to be used to make decisions. But none of them can survive the charge that they are subjective, or that they replace everything else, or that they somehow prevent you from using the other listed methods. Each of those charges pretends something that isn’t true.
I still don’t understand why you feel you must get catchers 3 – 8 if there’s a dropoff after 8. If catchers 3 – 8 range in value from $20-$14, and then 9 is worth $8, assume everyone goes for value and you’ll pay whatever each is worth. If you pass on the $14 catcher ranked 8th, knowing your best possible catcher now will only be worth $8, so what? You then likely drafted another $14 player at a different position and eventually you’ll draft that $8 catcher, or another catcher, when the round equivalent of $8 players begins or an even later round for a cheaper catcher.
Mike – I think this is actually the best argument you’ve made against tiers. Valuation just describes how much a given set of stats is worth. There’s no inherent benefit to getting a player in one tier versus another, if you’re buying at cost.
Looks like I might have my follow-up article for tomorrow since this is exactly the argument I expected and see it in multiple comments!
“You then likely drafted another $14 player at a different position and eventually you’ll draft that $8 catcher, or another catcher, when the round equivalent of $8 players begins or an even later round for a cheaper catcher.”
No. Not necessarily. In fact, not even likely.
I’m picking, say, 10th in a 14-team league. It’s coming down to me where I’ll be picking twice in 10 picks.
Catcher has three players left in Tier 2: $15, $15, $14. Then $8.
Second base has (say, next four players): $22, 15, 15, 14.
I need both positions. I see that the four teams picking (twice) before I get my next pick all need second basemen, but two have catchers. Tiers, in combination with tracking picks here, with perhaps some ADP consideration left out of this example, help me quickly ID the talent drop off, but in context, and I grab the second baseman here, and the catcher on the way back.
These opportunities – to grab towards the tail end of clustered talent at a given position is precisely what Tiers seek to help in quickly identifying.
Your view – roughly simplified as “with a $28, $26, $25, $25 …” distribution of player values, ignores the fact that a team has to fit those values into specific positions.
Stated most simply, we disagree on whether one can simply “then likely drafted another $14 player at a different position …and eventually …”
My position on this issue can be seen most clearly if you imagine a drop from $20 to minus $20 at a position, and that minus $20 player is the 15th player taken at that position in a 15 team league.. Once you’ve missed the $20 player and dropped off the cliff it’s not as easy as shrugging your shoulders, picking a $20 player at another position, and then making it up later.
The exact same principle applies, although not as starkly, and only here and there, with dramatic drop offs at various positions.
I think I know what you’re trying to explain here and if it is what I think, it’s actually exactly how I draft as well. In fact, I explained this strategy in my 2014 article How to Win Your Snake Draft in Point 3 under It’s Draft Time!
https://fantasy.fangraphs.com/how-to-win-your-snake-draft/
Is that what you’re trying to explain? I obviously agree with the strategy, but it doesn’t require tiers. All I need is all players listed in descending order of dollar value. I color code position to make it easy to see where those gaps are.
So why color code for positions, if all the other information is already listed on your spreadsheet?
Answer: because you’re saving yourself some mental effort at distilling all of that information, in real time, for every single pick.
Tiers are the same thing. The difference being instead of color-coding positions, you’re indicating a place that you’ve predetermined to be a significant value drop. Now you don’t have to look for that during your draft anymore.
Colour coding isn’t adding inaccurate information. Assigning players into tiers is indeed adding inaccurate information. It’s arbitrary.
Exactly. The way most people use tiers is just a less precise alternative to using dollar values. They add nothing, yet they have the potential to be misleading and cause you to overpay on a player simply because he was the last one left in his tier.
ie: If there is only one $20 shortstop left in a tier and the rest are worth $17 or less, and there are five $23 first basemen left in a tier, I’m still taking a first baseman, all else being equal. I don’t care if there’s 5 of them and the tier is loaded, I want one on my team! He’s worth $3 more than the shortstop, and I can take another shortstop much later anyway. If they’re all over-priced by $3+, then I’ll take one in the last round that is by definition worth at least $1. Better than taking the $17 shortstop and instead getting a $1 first baseman from the last round ($24 net value vs $21 net value).
I will read the article you posted, I promised. I only have a second right now, though.
I think there may be a disconnect here, probably at my end. I use “tiers” in precisely the way you describe color coding. In fact, in rotolab, the product I use at the draft, you mark tiers VIA color coding.
I think color coding is susceptible to every argument you made up top here: it’s arbitrary – no two owners would color code the same – and while I don’t agree that EITHER color coding or tiers “removes information” or in any way prevents you from comparing players at different positions (in both cases you can and should have your dollar values there too), I don’t see any difference whatsoever between the two methods in these regards.
I now read the article, thought it was great, and wholeheartedly agree with pretty much every word.
I find that drawing a line, or color coding – to me they accomplish pretty much the same thing (again, shilling for Rotolabs, they actually up the ante by allowing you to effortlessly toggle between color-coded tiers, AND a grid that breaks up players into $5 groupings) – to allow a slightly quicker and more intuitive visual display off those drop offs (for some reason I’m always “missing” those value DH only guys).
I don’t see how color coding, tiers or anything else is any more subjective or arbitrary than anything else we choose to create and bring into the draft, and don’t see how it takes away from any other tool.
Maybe I wasn’t clear? I color code positions, not tiers or tiers within positions. The color coding is obvious – during a snake draft with only a 1 minute clock, need to quickly see what position a player plays so I could ignore those I’ve already filled and focus on those I need.
Ahh, I totally misunderstood.
Why do you feel the *need* to fill catcher in this wrap? Grab that 2B (because obviously) and take the best player available in the wrap. If it’s a catcher, great. If the best player is not one of those catchers, then you’re overdrafting him because you are adhering to your tiers.
I didn’t mean for it to sounds as doctrinaire as it came out. That was a consequence of trying to provide a simplified, two-variable (position) example to illustrate a point.
Yeah, I don’t advocate over drafting, but I believe there are situations in a draft where in a close call between a few options you can reasonably consider drop offs in talent at a position in making your decision.
Let me try to explain in a different way. For game theory/thought experiment purposes, lets assume a league with only starters at each position and no bench. You only need to fill (again) catcher and shortstop -it’s the last two rounds of the draft. It’s your pick, one off the turn, and both you and the guy picking at the turn need a catcher and a shortstop.
You have your best two remaining shortstops at $30 and $29, and your best two catchers at $20 and $2. A simple list of dollar values might cause you to choose the $30 shortstop and you’d likely end up with the $2 catcher ($32 in value) whereas choosing the $20 catcher (because you are note the more severe drop off) nets you $49 in value.
While this is a concocted scenario, it illustrates a game theory point here – that simply choosing the best player does not maximize total value in all cases. That is because you cannot simply play players everywhere. So, if picking a player because he’s a couple of bucks higher than a player at another position ultimately, because of value drop offs at positions, results in you losing more than that difference later, you have not maximized your team’s value.
While the example is a bit loaded just to show the principle, these types of considerations occur up and down a snake draft.
Tiers are for when you want to deviate with your dollar values. If you, when drafting, strictly go by your dollar values, tiers are not for you. If you are an irrational human being that goes off of more than just dollar values, you might prefer tiers.
But like, if you look at those pitchers, I’d say there’s a nice $25-dollar tier from Kluber to Snell. These are pitchers I’d spend about $25 on – which means, yes, I might spend a littler more on Snell than my numbers tell me to and I won’t go as high on Kluber as my numbers tell me I should. But I like Snell, and would overpay a little to have him on my team, where Kluber doesn’t do it for me and I’m not looking to pay full value for him. Ultimately I’d be fine with either of them – they are in the same tier and I’d pay the same for them – but the how (or why) they’d get on my team is a little different.
I mostly agree with you, but you are overlooking the midrange value cliffs that sometimes occur at a position.
Adding a tier number to your dollar values can provide additional value if done correctly. Tier 1 is everyone who should be drafted purely on dollar value. Tier 2 is everyone midrange where variations in positional value matter a lot (e.g. “cliffs” where if you haven’t drafted someone yet, you will be left with a player who is unlikely to perform at a high level). Tier 3 is the undifferentiated “safe” but low-end players likely to be above replacement. Tier 4 is the mass of players likely to be around replacement.
Drafting with just the dollar values is fine for most purposes, but it requires nearly superhuman focus to notice when you are approaching the Tier 2 cliffs for a position, unless you label that group in some way.
Mike, you probably tier that food menu naturally, and with different cutoffs. “Hey, I’m deciding between the 2 expensive dishes,” or “I’m probably going with one of the cheaper items.” It’s the same way with fantasy sports… tiers are a way to communicate large drops in dollar value. “Oh man, there’s only 1 of the top tier 3Bs left, and then it’s a huge drop off. I need to grab the last one from the top tier.” Mainly, grouping distinct options into categories is a way to simplify our communications to others. The entire purpose is to leave out details!
To provide the flipside to the above 3B argument… when there is only 1 player left from a tier, and it’s an auction, then that player can often be bid up above his projected value due to subsequent drop-off in value to the next player. Recognizing when this is likely to happen is helpful.
I’m not sure why you think that tiers would be independent of dollar values, or determined without dollar values. Are you responding to a different article? It seems like you’re arguing against a strawman. But to answer your question, tiers can be helpful — independent of dollar values — because they can help you identify which players are likely to be overbid. In my auctions, the last player remaining in a position’s tier is often bid up beyond the projected price. This foresight has value, independent of your projected auction value.
Tiers are helpful if most of your competitors are also using tiers–they can predict when the cost for a player is likely to spike. The ‘last good player’ premium is almost always responsible for the biggest over-payments.
I prefer to target the second-last player remaining in a tier, rather than wait and hope for a value on the last guy in the tier.
As other folks have noted, it all comes down to how you prefer to organize your information–if you know the player pool well, and how your competitors value players, you can use whatever system you want.
Larry Schechter suggests that in auctions a sort of tier system can be helpful for achieving discounts while maximizing total roster value — e.g., paying full price for Mike Trout and some other elite players, but demanding $10 off to buy someone from a next lower dollar value group, and then $20 off a next lower dollar value group, and so on. This is a different use of tiers, but perhaps relates to your challenge. Maybe it too is too blunt, though.
In your example Tier 1 catchers =/= Tier 1 SS…. so why use numerical system? On my big board I have shades of color (green for tier 1) that go across positions, so in your example JT Realmuto will have the same shading not as Lindor, but as Xander Bogarts who has similar $val. Catcher does not have anyone in tier 1 and will have no one shaded green
Additional,
Snake drafts – ADP tiers let me know when its “go time” on a position and allows for easier planning predraft. Also helps to avoid panic picks when a target gets scooped.
Auctions – $ tiers gives me a price range for a group of players which keeps spending in check
Yep. This. Not every position has to have every tier. There might not be ANY Tier 1 catchers some years.
I have only found tiers helpful in the uber-specific example of picking near the turn but not at the turn of a snake draft and knowing I have a pick upcoming and X amount of picks (6 or less) before my next pick, and am also keeping pretty good track of what positions other players have filled. In that case, if I know I am likely targeting a catcher and a SS in your example above , and there are 4 picks before my next pick, and only 1 catcher left in tier 1 and 3 shortstops left in tier 1 respectively, I feel comfortable going catcher there, even if the dollar value is lower. But other than that, year what’s the point.
Using tiers for an auction draft is mental masturbation.
The way (I believe) most of us come with our pricing is via z values/standard deviations. The pricing after that is more or less a reflection of these standard deviations. You can say that $20, $30, $40 are 2,3,4 standard deviations above replacement level at any particular position.
If you’ve data modeled your reports and are applying a filterable pivot and are using this as just an additional tier in your hierarchy above position… you can more or less create a view that gives you each respective range of standard deviations in an easily viewable pivot.
Basically, this would allow you to group all the best players together. It’s not a necessary step, but I do think it provides useful information on how many remaining players are left that are x standard deviations above replacement.
To reiterate, this is something you’d place within the hierarchy, the values would remain next to each player. Or to say it another way, your assumption that you cannot compare between positions by creating tiers is wrong, the tier is just a level above in the hierarchy.
There’s only one tier you need concern yourself with, and that’s the first tier. The top tier. One tier, two players. Mike Trout, of course, and Wander Javier, the generational prospect currently in the Twins org.
Tiers aren’t perfect, but they’re better than relying on ordinal ranks. I think that is the number one drafting fallacy, and tier sheets help break out of that mindset.
Your menu example would have been fine if you had made 2 tiers with steak&fish in one and pasta&chicken in the other.
I think your first example shows where tiers can provide something superior to dollar values. The key is that projections are a very inexact science. That the difference between Sale and Scherzer is the about same as the difference between Aaron Nola and Andrew Heaney shows this, because that is just a bit nuts. Which pitcher between Sale and Scherzer is actually better this year is hardly skewed strongly in Sale’s favor—it likely comes down to injuries and luck. By using tiers–and going with similarly strength players as relative equals rather than dogmatically following inexact projections can result in better results.
Basically, with many other differences not relevant here, the argument against your anti-tier theory is Shandler’s BABS theory.
Id be worried about my calculations more than tiers if its suggesting that big of gap between Sale and Max…….. Not touching Sale this year and I havent seen a single site put Sale above Max and that gap is criminal to say the least
The Athletic has him ranked #1 over Scherzer. 3 experts had him #1, two had Scherzer #1, one had DeGrom. Steamer also has Sale #1 when you convert their projections. It all comes down to health..
Mike, Great Article!!! I agree that tiers are highly overrated and need way more detail than what is typically written on them. Tiers can be useful in a snake draft to identify times in the draft where there is value to be gained by waiting to select a player. At the beginning of the draft, the talent curve is very sharp and tiers don’t really help much and are pretty useless. As you pointed out when the difference from top to bottom of a tier is $10 then those players aren’t really in the same tier. The best strategy early on is simply to draft the best player available regardless of stats or positions.
Tiers only start to become useful when the talent curve flattens out and you have 4 or 5 players that all have a relatively similar value I.E. all 5 players are within $1 or 2. No matter how good your projections are there is likely to be enough uncertainly so that when 5 players are within a dollar or 2, there is no significant advantage to be gained by selecting any of them b/c if there were they would be worth significantly more. At this point in the draft you need to start focusing on specific team needs, I.E. I need a SS or SBs. Tiers then help to quickly identify the talent pockets left in the draft and allow you to better plan how best to fill out the rest of your team while taking advantage of the remaining talent to be able to select more valuable players while your opponents pass over them because they need to fill other spots or have already filled those.
Think about a situation where you have a 2 team league that needs to select 2 more players each a 2b and a SP. You look at your draft board and see a $7 SP and $6 SP along with an $5 2b and a $1 2b. By setting up tiers you see that see that there are 2 SPs in a tier and only 1 2b in each tier. You would then select the $5 2b letting your opponent choose the SP as you will gain more value from drafting that $5 2b and either pitcher as opposed to selecting the $7 pitcher and hoping the $5 2b comes back to you.
Since the game of fantasy baseball includes way more than 4 players, tiers can help group players to identify the pockets in the draft where value is to be had. I.E. selecting a $10 SP and forgoing that $11 1b because you note that there are 3 other $11 1bs left
Tiers are a way of saying, “I don’t get math so I’m going to approximate value in a very imprecise way.” I guess you can use someone else’s auction values, but I get that that is imprecise in a number of other ways. I use valuations specific to my league and have a clear baseline for how valuable a player is relative to what I need in order to win. But the answer to why others use tiers is it’s easy.
Thanks for a good read. I use tiers with a $ value range. Honestly, there is no reasoning other than when I tried using specific dollar values I became very rigid in my biding. Using your example with pitcher values, I think you could reasonably say that from Kluber down to Snell any one of them could put up $25-$30 worth of stats. It allows my brain to say it’s Ok to grab Snell at $28 if I wanted instead of waiting out “value” that may never come in that price range.
Question: If you leave the specific values with your positions or just in a straight dollar value list do you ever cross what you have set for the value or do you make exceptions up to a certain amount?
I’m a strict value adherence guy. I rarely go above my value, but will do if I screwed up somewhere and am at risk of leaving money on the table. The object of the auction is to accumulate as much value as possible (obviously also balancing categories, it doesn’t make sense to get Dee Gordon, Billy Hamilton, and Mallex Smith even if all came at discounts to your values), and the only way to do that is adhere to your values and scoop up as many undervalued players as possible. You can’t win with a $260 roster unless you get super lucky with outperformers and in-season moves.
Are you not essentially using 1 player tiers? All tiers are doing is rounding your auction values. (Which I want to point out are totally made up numbers that are based on predicting the future).
Let’s take your pitching chart and remove cents. As no draft ever uses anything other than whole dollars.
Chris Sale $49
Max Scherzer $40
Jacob deGrom $39
Justin Verlander $33
Corey Kluber $28
Gerrit Cole $27
Carlos Carrasco $26
Luis Severino $26
Clayton Kershaw $26
Blake Snell $25
James Paxton $22
Not really that helpful, so now lets round to the nearest 5.
Chris Sale $50
Max Scherzer $40
Jacob deGrom $40
Justin Verlander $35
Corey Kluber $30
Gerrit Cole $30
Carlos Carrasco $25
Luis Severino $25
Clayton Kershaw $25
Blake Snell $25
James Paxton $20
That feels like a pretty good tier setup to me based on your imaginary stats. It is just a visual aid to help during a draft. You could go a step further, round to 10 and then have more compact set of tiers. You could go the other way and use the full dollar tier. Or you can go even further and separate out your players to .01. Whatever floats your boat.