Gaining Rationality: Simple Auction Tables
Last week, our own Brad Johnson discussed unpredictable auction values in “Rationality Will Ruin You”. I 100% agree with his premise. Unexpected inflation exists but so does a solution. By charting out the desired and the potential outcomes before an auction, an owner can remove a ton of auction frustration.
Brad examined the elite player section of the draft where the actual auction values are higher than most people’s predicted values. Owners refuse to pay these high values but instead end up spending their resources on near replacement level talent. The following simple chart helps an owner deal with the problem along with uneven hitter/pitcher mix and outlining a personal auction strategy. The process is similar to the one used in Winning Fantasy Baseball by Larry Schechter but with a few more additions. Here’s the procedure.
Step 1. Get auction values
This step could be as simple as using our auction calculator or creating your own projections like our own Mike Podhozer does. To set the pitcher/hitter mix, use the league’s historic mix or just go with standard 70/30 split.
Step 2. Calculate tiered auction values
Divide the auction values between pitchers and hitters. Then, average the first X hitters where X is the number of teams. On average, each owner will have to buy one of the hitters from this tier. Continue the process for each set of hitters. Do the same procedure for pitchers.
These auctions values are the “Perfect Values”. In a world where everyone acts rationally, the final bid values would all be near these. As Brad stated, the world isn’t rational so, more steps are needed but ideal values set a good baseline.
For my Tout Wars team (15-team, 13 Hitters, 9 pitchers), the tables look like these:
| Pitcher | Perfect Values |
|---|---|
| 1 | 24 |
| 2 | 15 |
| 3 | 12 |
| 4 | 9 |
| 5 | 8 |
| 6 | 6 |
| 7 | 5 |
| 8 | 4 |
| 9 | 2 |
| Total | 84 |
| Hitter | Perfect Values |
|---|---|
| 1 | 36 |
| 2 | 25 |
| 3 | 20 |
| 4 | 18 |
| 5 | 15 |
| 6 | 13 |
| 7 | 11 |
| 8 | 10 |
| 9 | 9 |
| 10 | 7 |
| 11 | 5 |
| 12 | 4 |
| 13 | 3 |
| 14 | 2 |
| Total | 176 |
Additionally, spend a couple minutes going through your auction values and designate each tier. My values are on paper, so I draw a line between each tier. Then I mark each tier with its number and average value so I can reference the auctions values and team table. For those who have joined the 21st century, just use a different color and/or font to show the different tiers in a spreadsheet like the following.
Step 3. Add your own insanity
With the perfect values added, owners now add in their own plan. The owner isn’t changing their projections, just how they want to allocate resources. Maybe an owner wants a 75%/25% mix. Or it could be they want to spread the risk around and get as many $15 players as possible. They can implement the structure for their plan now.
One item I would add now is starting and relief pitcher slots. If you plan on buying closers, plan on how to distribute the resources.
For example, I am in a league where an owner goes with the stars-and-scrubs approach of spending all his money on top end players, getting drunk, and then bottom feeding after taking a hour break from the process. The following examples are what I expect his plan to look like (if he has one).
| Pitcher | Perfect Values | My Insanity |
|---|---|---|
| 1 – SP1 | 24 | 30 |
| 2 – SP1 | 15 | 30 |
| 3 – Closer | 12 | 10 |
| 4 – Closer | 9 | 5 |
| 5 | 8 | 1 |
| 6 | 6 | 1 |
| 7 | 5 | 1 |
| 8 | 4 | 1 |
| 9 | 2 | 1 |
| Total | 84 | 80 |
| Hitter | Perfect Values | My Insanity |
|---|---|---|
| 1 – SS | 36 | 30 |
| 2 – 1B | 25 | 30 |
| 3 | 20 | 30 |
| 4 | 18 | 30 |
| 5 | 15 | 30 |
| 6 | 13 | 8 |
| 7 | 11 | 6 |
| 8 | 10 | 5 |
| 9 | 9 | 4 |
| 10 | 7 | 3 |
| 11 | 5 | 1 |
| 12 – OF | 4 | 1 |
| 13 – C | 3 | 1 |
| 14 – C | 2 | 1 |
| Total | 176 | 180 |
Step 4. Add in historical league data (skip to 5 if none exists)
With perfect and personal values are added, now owners add in the historical information. If available, this information in invaluable. Does the league spend a higher per pitcher than most? Do the top 24 hitters have a 30% inflation? These values can point to areas in the draft where an owner needs to spend a few dollars more and when the transition happens to undervalued players.
Here are the historic Tout Wars values added to the tables.
| Pitcher | Perfect Values | Historic Cost | My Insanity |
|---|---|---|---|
| 1 – SP1 | 24 | 25 | 30 |
| 2 – SP1 | 15 | 17 | 30 |
| 3 – Closer | 12 | 13 | 10 |
| 4 – Closer | 9 | 10 | 5 |
| 5 | 8 | 7 | 1 |
| 6 | 6 | 5 | 1 |
| 7 | 5 | 3 | 1 |
| 8 | 4 | 1 | 1 |
| 9 | 2 | 1 | 1 |
| Total | 84 | 84 | 80 |
| Hitter | Perfect Values | Historic Cost | My Insanity |
|---|---|---|---|
| 1 – SS | 36 | 38 | 30 |
| 2 – 1B | 25 | 28 | 30 |
| 3 | 20 | 22 | 30 |
| 4 | 18 | 18 | 30 |
| 5 | 15 | 16 | 30 |
| 6 | 13 | 13 | 8 |
| 7 | 11 | 11 | 6 |
| 8 | 10 | 9 | 5 |
| 9 | 9 | 7 | 4 |
| 10 | 7 | 5 | 3 |
| 11 | 5 | 3 | 1 |
| 12 – OF | 4 | 2 | 1 |
| 13 – C | 3 | 1 | 1 |
| 14 – C | 2 | 1 | 1 |
| Total | 176 | 174 | 180 |
This table can provide needed insight. If the above owner is set on spending $30 for the hitters, he is going to be purchasing players around his ~$27 perfect values because of league inflation. He should spend some extra time researching these hitters to get the best value.
Step 5. Utilize real-time auction results
Two blank columns to add at the end. The first is the player’s auction value and the next is the difference between the “insanity” value and the actual value. If the stars-and-scrubs owner gets a hitter for $40 in Tier 1, they would need to bring down the rest of their values by a total of $10 to maintain the $260 budget.
Here is an example of the final tables with a player added.
| Pitcher | Perfect Values | Historic Cost | My Insanity | Actual Cost | +/- |
|---|---|---|---|---|---|
| 1 – SP1 | 24 | 25 | 30 | ||
| 2 – SP1 | 15 | 17 | 30 | ||
| 3 – Closer | 12 | 13 | 10 | ||
| 4 – Closer | 9 | 10 | 5 | ||
| 5 | 8 | 7 | 1 | ||
| 6 | 6 | 5 | 1 | ||
| 7 | 5 | 3 | 1 | ||
| 8 | 4 | 1 | 1 | ||
| 9 | 2 | 1 | 1 | ||
| Total | 84 | 84 | 80 |
| Hitter | Perfect Values | Historic Cost | My Insanity | Actual Cost | +/- |
|---|---|---|---|---|---|
| 1 – SS | 36 | 38 | 30 | ||
| 2 – 1B Paul Goldschmidt | 25 | 28 | 30 | 40 | -10 |
| 3 | 20 | 22 | 30 → 28 | ||
| 4 | 18 | 18 | 30 → 27 | ||
| 5 | 15 | 16 | 30 → 27 | ||
| 6 | 13 | 13 | 8 → 6 | ||
| 7 | 11 | 11 | 6 | ||
| 8 | 10 | 9 | 5 | ||
| 9 | 9 | 7 | 4 | ||
| 10 | 7 | 5 | 3 | ||
| 11 | 5 | 3 | 1 | ||
| 12 – OF | 4 | 2 | 1 | ||
| 13 – C | 3 | 1 | 1 | ||
| 14 – C | 2 | 1 | 1 | ||
| Total | 176 | 174 | 82 |
Now, the owner is ready for the auction. Besides keeping track of his team, an owner needs to track the individual final auction bids. Look for any trends. For owners in a new league, this aspect is really important. Without the historic values, the inflation amounts will soon become apparent. For the owners with historic data, they can determine if anything fishy is happening like several owners dumping pitching.
For example, here are some final action values. Just by doing a quick eyeball of the results, an owner is likely going to have to spend a couple dollars more to get one of these pitchers.
Finally, if an owner finds the actual costs below his “perfect” pitching projections values, the league’s average split needs to be corrected with the surplus given to hitters instead of pitchers.
By keeping these two simple tables up-to-date during the auction an owner can keep a good handle on inflation and hitter/pitcher splits. They bring needed rationality to any auction.
Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won four FSWA Awards including on for his Mining the News series. He's won Tout Wars three times, LABR twice, and got his first NFBC Main Event win in 2021. Follow him on Twitter @jeffwzimmerman.
I’ve pretty much been doing this exact format since 2012. That was the first year of our 15-manager, 43 player (24 active, 6 inactive PLUS 5 DL and 8 Minor League slots), 7×7 roto-league. Each year, we’re allowed to keep 5 MLB and 5 MiL players. MLB are to be kept at a $5 increase to their previous salary and MiL can be kept at their value from the previous year. I’ve finished, 1st, 2nd, 1st, 3rd and 3rd. I feel bad for the guys who go into an auction without a plan.
The “thumbs down” must be from a guy in my league that I regularly smash. Bravo.
Hostile community.
I’ll add my personal strategy here, which I find to be an incredible help. If there are n teams in your league, create dollar values >$1 for only enough players to fill n-2 rosters.
If it’s a 15 team mixed league with 23 man rosters (14H+9P) and $260 budget per team, create values as if it was a 13 team league. Subtract out $1 for every player that needs to be drafted for the 15 teams ($23/team, leaving $237 of actual spending money, x 15 teams = $3555) and distribute those remaining dollars over the top 299 players. If you do it right, the 299th ranked player will be worth exactly $1. Everybody below him will be worth less than $1, but since $1 is the minimum, assign $1 to each of the players ranked 300-345. In this format, the final 40-50 players probably only go for $1 apiece. You can look at historical auction records to get a better idea, so maybe it’s only the final 20-30 guys that go for $1 depending on how your league is set up, so you need to adjust and create values for a 14 team league in that case. This system automatically takes all of that extra money that you would have otherwise allocated to the scrubs that sell for $1, and gives it to the stars. This more closely mirrors the reality of an auction.
This method closely mirrors the results brought by the aforementioned “Winning Fantasy Baseball by Larry Schechter” but with a lot more reasoning and less arbitrary division of AL and NL only dollar values. This will solve your problems if you find that you end up missing out on the first 50 players during an auction since they’re “overpriced” and being stuck with a bunch of <$10 scrubs that are easily replaceable via FAAB, anyway.
This method doesn't work as well for AL/NL only auctions because the replacement level is so terrible in those leagues, and therefore, there are a lot less usable $1 players at the end of the auction.
“To set the pitcher/hitter mix, use the league’s historic mix or just go with standard 70/30 split.”
Excuse my newb question but can you expand on this a bit? It’s something I never have truly understood and it sounds like I’ve been overvaluing pitching based on what you wrote.
It means the total money spent on batters (70% of the funds) versus pitchers (30% of the funds). It can either mean your particular roster or the league as a whole. In this case, he is talking about how the league’s total funds will be split on batters and pitchers respectively.
My only auction league is an Ottoneu Fangraphs points league. I’ve already kept my entire staff for 2017 and I was like “70/30? I’d never get to that ratio.” I just did the math. My pitchers are 31.25% of my $400 budget. Nice!
I do tend to find a more even split in FGpts. I’ve seen everything from 55/45 to 80/20 splits work with ottoneu. Probably a function of the points format in general rather than anything FGpts specific.
I’d love an article on whether there is an optimal Hitter/Pitcher split for a fantasy league. Just because a league is 80/20 or 66/33 doesn’t mean that is the optimally efficient.
Is this a facet of fantasy that can be exploited? (Either by hoarding pitchers in a league with a large split or going full LIMA in a more equal league)
You can spend as much money on hitting or pitching as you want to. The reason that you calculate your valuation prices based on a 70/30 mix (or your league’s historic mix) is so the you can most efficiently identify good values based upon the prices that are likely to be paid. If your particular strategy is to spend more on pitching, say 40%, then you should do so. But calculate your valuations based on the mix the rest of your league is using as a means to help you acquire the most value for your 40%. Then, at auction, do what you have to approach the 40% target. You may have to pay a little more on some pitchers to get there.
If you don’t identify the H/P split for an auction, you’re gonna have a bad time. It would be impossible to put relevant values on any player and you’re essentially just shooting in the dark and have no idea what players are going for a good price based on your projected performance for them.
If you make values assuming a 60/40 H/P split and the actual split turns out to be 80/20, every single pitcher is going to seem like an amazing bargain. “Wow, I’m getting all my guys for $5 less than my projected value!” You’ll miss out on the actual best bargains on pitchers, which may be $10-$15 less than your projected value. And you’re going to end up with f***-all for hitters; probably an entire lineup of scrubs that nobody else wanted, buying up all the $5 guys for inflated prices in a panic, to avoid being left with $50 at the end of the auction. “Wow, everybody is over-spending by $10 on all of the good hitters. I’ll just wait until prices come down. Surely they’ll eventually have to come down, right? Oops, everybody re-allocated the money I thought they would spend on pitching and are spending it on any half-decent hitter. I wonder if Garrett Jones is still playing.”
Moral of the story: always try to get an idea of the historical H/P split for your league. For auctions especially, but also for drafts. You can do that by assigning dollar values to every player, and also “slot values” for each spot in the draft.
EDIT: To address the part about how to exploit this; it’s not as easy as it seems. There’s really no “best” split, other than the ones that the league as a whole uses. Lots of people argue that hitters are more reliable and pitchers are easier to add via FAAB (starters that come out of nowhere, streamed starters, relievers that get a closing job, etc) which probably drives the split to the historical 70/30 spot. People have tried to make sense of it. Tom Tango wrote an article about how ideally it should be 50/50, for instance, since a 5×5 league has a 50/50 distribution of categories that each count the same. But it all eventually comes down to: whatever the entire league does, you need to follow, when creating values. If you want to go pitcher-heavy for whatever reason, then you should allocate more money from your budget on pitchers while still using dollar values based on their value for the actual 80/20 H/P split, so you can identify bargains. if that makes sense.
Thank you SkinBlues and others that have clarified this. It makes a lot of sense. Much needed info!
Where would be a good place to learn how to assign $ values to slots in a snake draft?
What I do for drafts is assign dollar values to each player as if it was an auction. I do this in excel. You can add another column to the spreadsheet and put a rank by each player, too. So if you have 345 players, or however many, sorted by projected dollar value, the ranks will be 1-345 for those players. You should have a separate column for position as well. For simplicity, change all SP/RP just to “P” when noting their position. You don’t have to, but it makes it easier to write formulas.
In a separate spreadsheet with draft results of old drafts for that league, sort by draft pick number. Create a column for dollar values, and paste the dollar values from your personal rankings into that column. Be careful to paste the values only (right click;paste special;values only), and not the actual formulas. So now you’ve basically applied dollar values to draft position. Team A picked a hitter #1 overall which has a value of maybe $53, Team B picked a hitter second overall which has a value of $49, Team C picked a pitcher 3rd overall which is equivalent to $46, and so on for all 345 picks.
To get the H/P ratio, do as follows. In an empty cell of the spreadsheet containing the historic draft results, with your newly-pasted dollar values column included, you white the formula to calculate the H/P split. Assuming the first row is headings, player names are in column A, dollar values are in column B, ranks are in column C, and positions are in column D, the formula might look like this: =sumif(D2:D346,”P”,B2:B346)/sum(B2:B346)
Which essentially means “sum up all of the dollar values for pitchers and divide it by the sum of dollar values for all players”. You’ll probably get something like 0.30, or 30%. Subtract that number from 100% and you get the hitter split, so in this case 70%. You can do this for a few different sets of draft results for the league, but it can be a lot of work and a bit tedious if you’re not familiar with this type of thing. I’ll admit it’s a bit overkill, but I enjoy this stuff so I don’t see it as work, and it helps reduce the chaos of the live draft/auction to have solid values known in advance that don’t require a lot of tinkering on-the-fly.
I spend countless leisure hours buried in excel spreadsheets for fun so I totally understand and will definitely try this. Thank you! Unfortunately I just thought of another major wrinkle which is figuring out what to do with the keepers.
Can’t help ya there, I don’t do keeper leagues. One guy to talk to would be Todd Zola, he’s on Twitter. He’s with Mastersball and is a valuation guru, basically. I also recommend the Mastersball platinum subscription, even if only for the projections. Todd is good with helping with unique league settings.
do you know why the web auction calculator generates very different values for certain guys compared with the google sheets version (e.g., David Dahl, Rich Hill)? they’re both based on steamer projections and set to FG Pts scoring, so i’m not sure what’s going on.
This is a great article, but the simplest solution is to buy a draft software that calculates the inflation for you as each pick is made because there is so much to keep track of, doing everything in excel or by hand takes my concentration away from the actual auction. I do agree that looking at historical data is extremely important. Like I know my league will overpay massively for closers and upper tier SP
It’s very easy to have Excel calculate in-draft inflation for you as well…
I do this, too. But it can jump around a lot for the first hour or so of an auction; best not to jump to conclusions. Stick to what you expected the split to be for a while unless the trends have swung strongly in one direction or the other.
That’s interesting. I’ve been in both an AL & NL-only leagues (variable keepers) but I’ve never seen a massive overpay for top tier starters.
However, with closers it’s a different story. And since we have 7 x 7 cats with holds, I could never figure that out. I generally punt that category because of the crazy bidding on closers.
I’ve never used draft software. What is a good example?
http://rotochamp.com/
Thanks
This is a great plan for people who play in redraft/auction leagues.
Too bad a lot of people like me play only in keeper leagues, where obviously the best players are franchised, signed to a “contract” or put into some type of category that freezes their cost for at least a couple of years. Until they become too expensive or their “contracts” expire, plans like this are of little value.
And I imagine that in Tout Wars there’s little turnover. Even in the best keeper leagues that I’ve been in there’s anywhere from a 10-15% annual turnover rate. Then there’s no way to know how the new owners tend to allocate their money.
That’s the great thing about full keeper auction leagues. Each owner must determine their own unique values list and the auction ever more closely mirrors MLB free agency: you’re going to overpay. You can fill holes through the auction, but you build a winner through smart ownership and, in my leagues’ case, adept selections in minor league drafting.
Great article! I’ve done a bare bones version of this before, but this will help me refine it. Our auction is online and goes rather quick, so it’ll be hard to follow along real-time and write down each player’s bid.