The FanGraphs Auction Calculator – Advanced Use
Last winter, FanGraphs rolled out an auction calculator. I wrote two articles on the subject of using and abusing the calculator. First, a tutorial, then my lifehack advice for using it. I’ve been getting numerous questions on Twitter and by email, so what follows is some advice for power users.
Generating Pitcher Values
Like any automated calculator, the tool has quirks and shortcomings. Building values for pitchers is one area where the intuitive nature of the calculator falls apart. You can get an accurate list of pitcher prices, it just takes some massaging of the data.
It asks for you to select the number of starting pitchers, relievers, and generic pitchers. Based on your input, it will then optimize prices subject to your league specific categories. If you input 2 SP, 2 RP, and 3 P, the tool will run projections based on an average of 700 innings per team. Before you pay $85 for that hyper-elite Clayton Kershaw, I’m willing to bet that you actually need something like 1,400 to 1,500 innings.
The trick is to simply list more starters and relievers under the “Pitching Positions” section. You will have found the appropriate balance when the sum total of innings projected for all positively valued pitchers is equal to the sum total of innings projected for your league. So a 12 team, 1,450 IP league should have 17,400 with positive or zero dollar valuations. That’s equal to about 7 SP and 3 RP.
Advanced Use
Based on your personal preferences, you can also tinker with starter and reliever values. I tend to prefer to draft a few high quality starters, a small horde of relievers, and then I’ll just stream the rest of my starts. So while my leaguemates probably aim to draft a full complement of innings, I might aim 250 innings below that cap. That increases the value of elite starters and relievers.
In the tool, simply list more relievers. In the above example, for a 1,450 inning cap league, I might enter 4 SP and 5 RP. Top starter values are buffed by about five to 10 percent. Top reliever values increase by about 40 percent. For example, Kenley Jansen goes from $18 to $25.
The new valuations assume about 14,000 innings will be thrown around the league. Using the tool like this can help you find which players to most aggressively target. In this case, go goo-goo for relievers. It’s still up to you to find the diamonds in the rough.
Non-Auction Calculations
If you’re the sorta of lady or fella who likes to rely solely upon Steamer and/or ZiPS, then the auction calculator still has two nice uses in Snake and Circular style drafts. In short, use the tool to create rankings and tiers.
Here’s a sample page of outfielders. We have the Mike Trout tier, the Bryce Harper tier, the Betts-Stanton-Bryant-McCutchen tier, an 11 player $20-$26 tier, some $18 guys, and a cliff. The calculator is doing two helpful things for you here. You’re getting automated rankings, and you’re getting free advice for avoiding the pitfalls.
This is especially helpful for players in unusual leagues. For the generic 5×5 league, there’s a good chance you prefer X, Y, and Z analysis over the sanitized projections of Steamer. In a custom league, there’s a good chance you don’t have many fantasy experts offering fully relevant advice. You have to adapt what you read. In such a scenario, Steamer generated rankings and tiers can be very useful.
To the Comments
Let’s use the comments for additional questions about the calculator. Remember, when you generate values, the hyperlink becomes your copy of the results. Feel free to share it with me if you see something wacky or unusual.
You can follow me on twitter @BaseballATeam
I continue to receive an ‘unexpected error’ when trying to run the auction calculator. Any idea what could be causing this? I am trying to run a hitters-only, points-league projection
Saves plus holds, please.
Agree on saves plus holds. In the meantime, is there an advanced way of adjusting for this accurately?
To nip this in the bud, we’re not soliciting requests for stats at this time. You can simulate the SV+Hlds cat by arbitrarily increasing the value of RP.
Quality starts for points leagues
How does the keeper cost function work?
When I try and use it for ottoneu 4×4, I get some extreme values no matter what I do. I click the preset scoring for ottoneu points, switch back over to Roto and add the 4×4 categories (R, HR, OBP, SLG, ERA, WHIP, SO, HR).
Mike Trout ends up at $109. Harper is $96. Kershaw is $125, etc.
Tweaking the positions and number of pitchers doesn’t bring things back in line. Is there something I’m doing wrong?
I don’t play much 4v4, but those values don’t really smell wrong. I have them all priced around $85-90 for FGpts.
That’s interesting. Following along with the ottoneu values that have been posted to fangraphs so far, they look like they’ll be much lower (Trout around $70 in FGPTS).
Not that you can assume, but my guess is most people are working off Steamer or Steamer/Zips projections at this point. What creates the large difference in values?
Not to get too deep into how to create values, but is it where you set replacement level?
I’ve got a feeling that I go with the calculator values, I’ll be drafting very differently from the rest of the league, for better or worse.
I run into the same problem. Part of the problem is HR vs HR/9 on the pitching side, but that should be accounted for with the assumed innings limit. Since it’s taking raw HR numbers, starters get negative value because of the bulk number they give up, even though they might contribute overall. But I guess since it’s an equal hit to all starters, it probably evens out.
It also seems to set the replacement level very high. My feeling is that it has to do with the 40 player rosters and limited starting spots that screws up the calculations. It doesn’t track that kershaw is $125 and scherzer is half that much.
You’d think that there would be more guidance for an ottoneu standard setting like 4×4,but it always seems get overlooked.
5×5 values don’t look skewed in the same way (Trout at $75). So maybe…. it’s not crazy? When I tried to create my own values they came out similarly high at the very top, with a steep dropoff. So I assumed I’d done something wrong and discarded them.
I just know that when people who know far more than I do create 4×4 values, they end up more “normal”.
Using ottoneu default settings, only 300 players are given a positive value, instead of 480.
Regarding the Experimental feature, how does the auction calculator calculate the replacement level values?
Brad- lets say i play in a league that uses a random assortment of ML teams like we use all teams except those in the AL West. How can I can exclude those players from the projections?
You can mark them as keepers at a $0 value. It’s time intensive and I’m not positive it would work.
Or you could export the data and use a function or tool to create a new list using specific terms to remove players on teams you don’t include.
That wouldn’t work right. The prices would still be calculated with those players included. So there wouldn’t be the right amount of money allocated to the league and the players wouldn’t be priced against the right pool.
Right. Totally forgot since I’m using this tool for a snake draft league with no dollar values.
Is there a way to use my own projections and not Steamer’s?
Sorry, the tool isn’t set up for that.
Maybe I’m an idiot, but when I run the AC for my 7×7, 15 team mixed league, there are a decent amount of starting pitchers missing from the list. (Off the top of my head, Zimmermann is nowhere). Link to follow so you can please point out where I’m a dumb dumb and screwed up. Thanks!
http://tinyurl.com/jysry9x
It assumes many negative-valued pitchers (like Zimmermann) will be unrosterable in that format. If teams generally roster more than just 4 SP, change that number to something more realistic.
For example, my league “starts” 3 SP, 3 RP, 2 P. But to hit the innings cap, a team will need to roster 6 SP and 5 RP. So I put 6/5/0 into the Auction Calculator instead of 3/3/2.
Remember to deduct the difference from your bench spots too!
Yep, you’re pitcher replacement level sounds like it’s too high. As for Zimmermann specifically, Steamer’s projections knock a few name brand pitchers off the board.
Even with its quirks, the calculator is insanely helpful, so thank you for it, and thanks for fielding some questions.
I play AL Only 4×4 and the auction calculator always spits out crazy results for pitchers, especially swingmen:
http://www.fangraphs.com/auctiontool.aspx?type=pit&proj=steamer&pos=1%2c1%2c1%2c1%2c3%2c1%2c0%2c0%2c0%2c1%2c5%2c3%2c0%2c8&dollars=1000&teams=12&mp=10&msp=5&mrp=5&mb=1&split=50&points=c%7c0%2c1%2c3%2c4%7c0%2c1%2c2%2c3&lg=AL&rep=0&drp=0&players=
Trevor May is just as valuable as Cole Hamels and Vidal Nuno isn’t far behind. Making the adjustments you suggest (I expanded to 7 SPs, giving 11764 IP above replacement, close to the 11600 total IP limit in my league) pushes May and Nuno down to Rick Porcello territory, which at least is in the right direction, but is there anything else I can do to adjust? Thanks for your help!
This is mostly a Steamer issue. It doesn’t like Hamels much. Confusing to see May and Nuno up there though since both project to be holds relievers – and maybe not even that for Nuno.
Clayton Kershaw has a higher auction value than any other player (including the Trout Harper Goldschmidt trifecta) n any format that I have attempted. Does this mean he should be the highest valued player in an auction draft and the number 1 overall pick in snake drafts?
In fact by dollar value the auction calculator seems to be overrating elite starters. Scherzer has a higher auction value than bryce and goldschmidt. Maybe he will return that value but there is surely no way he will be regularly drafted or purchased ahead of harper and goldschmidt. any way to adjust for this?
In the linked tutorial from last year at the top of the article, Brad says: “Once you input those details along with your preferred batter/pitcher split (I chose 70/30)…”, implying that you may want to fill in the ‘Bat Split’ box in the auction calculator with, say, a 60 (for 60%) or 70 (for 70%) to inflate the hitter estimates and deflate the pitcher estimates to match how you want to divide your budget (or how you expect your league to spend its $). Try inputting a couple of different numbers in there and see how it does at matching your expectations for your league (particularly if there is historical data for your league, not just your gut instincts).
Thanks!
You also need to follow the instructions in the article above. In most normal category sets, Kershaw should not be more expensive than Trout even if you’re using the standard 70/30 split. Most likely, you have pitcher replacement level set too high.
I play in an innings cap league using K and BB. From my understanding these essentially become K/9 and BB/9. Using the auction calc
Is this a straight across conversion? Historically only about half the league hits the cap (1250) is there a way to compensate for this?
If only half the league hits the cap, then I’d still treat the category as K and BB. Very unusual that your rivals are taking an obviously suboptimal approach
what do the ‘adjusted’ values represent? some of these values seem way off.
The adjusted column is value with the keepers you entered considered. If, for example, 10 3B are kept for $1 and Manny Machado is the only good available 3B, you might see a really silly price on his head. The adjusted column does require some interpretation and tinkering in my experience, especially with regard to catchers.
Thank you for the great tool. One quick question: I ran two projections–one with AVG, one with OBP, but otherwise identical, to see whose value goes up the most in an OBP league. I noticed that some of the values of other categories changed, too (by 10 to 30 cents)–why is that?
I’d guess the baseline is shifting at the bottom end due to some players being fractionally more valuable than others via OBP. The OBP guys bring different SB, RBI, and Run totals to their position’s bucket than their counterparts who just barely made the cut on Avg.
Sort of a trickle-up effect
I’m guessing it’s more of a trickle down effect, some of the top OBP guys causing shifts in the calculus. It’s all just rounding, I wouldn’t worry about it.
That makes sense– thanks for taking my question!
I would like to input my own projections into this. I’m not sure where to begin though. Is there a way to export it with the steamer projects available to edit? Thanks!
Sorry, but you could reference the methodology used for LPP, export the steamer projections, and build your own value sheet.
first – the calculator is fantastic and i have been using for tiers for a snake draft league. thanks for all your work – unbelievably appreciated.
I am trying to adjust pitchers for a deep 20 team league with a maximum of 29,160 innings but also a requirement that you roster 11 pitchers. it sounds like I should use 6 starters (about 180 innings each = 1,080)and 5 relievers (6o innings each = 360) to get proper values (total innings estimate ((1,080 + 360 =$1,440*20 = 28,800 is as close as I can get). does that make sense or am I looking at this incorrectly? more important, does a roster dictate such as you have to have 11 pitchers change any of the math regarding the innings calculation or use of the calculator.
A roster minimum shouldn’t really change the math. It’s probably optimal to have more than 11 pitchers though.
Three questions:
1. Have you thought about doing Pitching/Hitting Splits? Like 70% weighted for hitters/30% for pitchers? So it would be different @ 60/40 or 65/35. Fantasy Pros has this and I find the splits extremely helpful.
2. What about H2H?
3. It seems weird that under Roto 14 Team league Kershaw is worth $55 while Trout cones in @ $42. No? I don’t see how any pitcher is worth more than a hitter.
1. They’re in there (Bat split)
2. For me, H2H usually just means a smaller hitter/pitcher split. Maybe 60/40 instead of 70/30.
3. You probably need to follow the instructions in the article for properly setting pitcher replacement level. The calculator probably thinks every team is using 800 to 1000 IP in which case Kershaw would be stupidly valuable.
Brad, I play In weekly league where I average about 89 innings per week. This comes out to total 2,100 – 2,200 innings for the season. How many SP and RP would that equal? Going off your approximations above, I’m guessing 11 SP and 4 RP. Is that right?
Sounds roughly right. To do a quick sanity check, just export the pitcher page and sum the IP of all positive valued pitchers. Should total whatever your league uses. I don’t know if you’re unique in reaching 2100 IP.
I’m guessing you have to do a lot of streaming of 2 start SP to get there. So you may actually want to “target” a lower IP amount for building values.
To be more exact. It’s actually a weekly roto matchups league with daily transactions. A good deep rotation usually only needs to stream 1-2 guys a week to hit our weekly 12 starts a week cap. But it’s very necessary to pound innings a weekly basis with 12 starts and high efficiency mrp to stay competitive in all of the counting stats.
I’ll definitely try exporting all of the positive
Value pitchers and see what I get as well.
Thanks!
Any chance the ability to utilize consensus projections can be added?
If you mean ZiPS+Steamer, I think that’s just waiting for some ZiPS updates to player pages.
yes. thank you.
Two questions:
1) Is there anyway we could get the calculator to go out one more decimal place? I play in a $3 league (a hold over from when we actually used loose change when we bought players) and having the hundredths place would be awesome.
2) We play using RF/CF/LF. Is there any calculator setting I could use to adjust for these unique positions?
Thanks!
1. This is a very easy fix on your end. Just make it a $30 league in the calculator then mentally move the decimal place when reading the values. The tenths place becomes hundredths, ones become tenths, and tens become ones.
Or even better, make it a $300 league and you can read the whole values as cents.
Thanks for the tips Brad! Really useful stuff.
One question: Are catchers really that scarce in 1 catcher, 10 team leagues (1 2b, ss, 3b, 1b, Corner IF, Middle IF, 5 OF, 1 util) that their positional value is worth $22.8 (for comparison, under my league’s specifications, this is exactly double the positional dollar value of outfielders, which is $11.4)? It just seems that past the first three there are maybe 15 catchers all with comparable value that the catcher positional dollar value seems a little high.
Thanks for your help!
I agree with you. There are some top guys and then the rest are way too compressed to pay anything for them. It’s basically dart throwing with the Russell Martin dart slightly heavier than the Francisco Cervelli dart.
Brad
I was just looking for a little help based on my specific league. We are an 11 team AL only with 9 pitchers. We have a minimum of 1000 innings required and we are four by four (W, S, ERA, WHIP) for pitching. I have played with 6 sp and 3 rp and 7 sp and 2 rp but I am not sure which one would get better results based the specifics of my league. thanks for the help
nat
I hope someone sees this and points me in the right direction..
Running starter calculations spits out a ton of guys projected to only pitch nine innings, making them useless to me and pushing real, actual starters like Jake Odorizzi out of the results.
How do I make it give me starters are are projected to actually be starters?
Adding IP as a scoring category changed things a little but I’d rather a solution that doesn’t involve valuing a stat my league doesn’t use…
You must have set a very low replacement level. I’m guessing you have something like 3 SP and 2 RP? Follow the directions in the article and it will normalize.
Yeah.. I have 5 SP and 7 RP with no value for generic pitchers. This somewhat reflects my strategy but it didn’t change much from my 4/8 original setting.
I’ll keep playing around with it, maybe changing it to what the league will probably do.. 6/5 or something like that.
I see what you’re saying now. I bumped it up to 9/17 and a lot more useful guys showed up. Still a ton of 9-99 innings guys but it’s better than before.
Hey Brad,
Thanks for the article (and custom Auction Calculator in general).
I have a couple of possible bugs I’ve found with the calculator I wanted to bring to your attention, since these have resulted in skewed auction values… which kinda negates the value of this calculator. Maybe I’m doing something wrong, but I’ve messed with it quite a bit and still can’t get things to look “right”.
1) When I adjust the options for my 6×6, 8 team league, some prominent pitchers just DISAPPEAR. I don’t mean it generates negative values for them, I mean they are not on the list anywhere. For me this includes Sonny Gray, Cole Hamels, Wainwright, Zimmerman, and more. In an answer above you kinda sloughed this off as “streamer projections aren’t kind to Zimmerman”, but I don’t think that’s what happening here.
2) My league uses xbh as a category. However the mXBH values it generates do not at all correlate to the Streamer projections. I’m working in excel to work out the replacement level and multiplier the generator has come up with, and all the “standard” categories are working fine… but xbh (I add up the 2B, 3B, and HR from the exported Streamer proj) are all over the place. For example it gives Mookie Betts a $11.20 mXBH and $5.80 for Trout, even though they’re projected to have 61 and 73 xbh respectively. This has thrown off the values enough that I had to just do those from scratch… but others might just trust them blindly and not realize they’re not accurate.
Add pitchers until the total projected innings pitched for positive pitchers that the calculator spits out is roughly equal to the minimum innings pitched for your league (or greater).
Jeff explains it pretty well in the comments, it’s assuming you need way fewer innings that you actually do and therefore assumes that guys just won’t get drafted and doesn’t include them. For me though it just meant the list was full of sub-99 inning “starters” just taking up space.
For me I needed 9 starters and 17 relievers to get to a mix that worked better. It got rid of a lot of the low-innings starters ctual, etc.).
Stupid phone..
Anyway 9/17 worked for my league but that was only my first run at it. There are still more non-starter starters in the way for my liking. I haven’t yet tried making the minimum games something way too high to see what happens, but I will once I have more time for draft prep.
I appreciate the reply, but that isn’t what is happening. After playing with it a bit it looks like it has something to do with using K/9 as a category instead of SO. Of the top ~50 SP (by any reasonable metric), all of them are there and resonably valued minus like 8 guys.
After reading your posts whatever issue you were having is not what I’m running into. I think if you’re having to adjust things to 9 SP and 17 RP you’ve messed one of the other variables. Whatever it is probably results in a weird order and/or valuations, but unless you have non-standard categories I’d bet every starter is on that list somewhere (for me the list is 500+ pitchers). The list it is generating for me doesn’t have really low values for Gray, Hamels, Richards, etc… it just omits them.
This is a nitpicking question but is it possible to add the actual projections to the output so that I can kill two birds with one stone when exporting this to excel?