21-02-24 Mailbag
Do you have a question that would be best answered by me and benefit everyone? Submit it on our Contact Page and put MAILBAG in the Subject Line. I’ll select the best ones to respond to in each of the four Game Plan installments.
— Ron
I don’t understand how to convert BABS ratings into draft rankings for the FISH List (or to effectively use BABS). I understand that the asset categories represent players with similar skills, but I also believe that the tenth player listed in one asset category in many cases will not be preferable to the first player in the next lower category. How does one assign a target draft position using BABS ratings? Help!
Regarding the FISH List, there is nothing scientific here. The BABS rankings used for the FISH List are just a plain old vanilla sort of the BABS list. That’s highly inaccurate as a true ranking, so we’re not going to call them rankings at all. They are now called “markers” that FISH uses to calculate offsets to the ADPs. It still works. Mostly. More here.
As for your larger question, that requires a larger answer, and my best response would be to point you to Book 2 of “The BABS Project 3.0” — specifically, Chapter 9, page 40. The link to the PDF is on the home page.
A review of the FISH data and a comparison to normal cheat cheats seems to indicate that on the offensive side, ADP overvalues while on the Pitching side the over value/under value ratio is much closer. In other words, BABS “hidden gems” are much more likely on the pitching side than on the hitting side. True?
This is true, but this really has always been the case, generally. Particularly in a Rotisserie context, the cost of hitters is driven by big counting stats since they represent four of the five categories. There are a handful of pitchers at the top of the pyramid, but there are far more further down with excellent ERA/WHIP levels who have sneaky value.
I like to think of it this way… A .300 hitter is going to find his playing time into a major role. A 2.00 ERA pitcher is also going to find his playing time, but it might only be in the middle of a bullpen. He’ll be more hidden there, but possibly no less valuable.
An important one from the reader forums: What impact, if any, does FISH have in auction drafts? BABS has always provided “value” opportunities relative to ADP. Does FISH add anything to auction strategy?
The thought exercise underlying the FISH concept is already inherent in auction bidding. If the market believes a player is $20 and you think he’s worth $24, then you naturally will bid at least $1 more than $20. The area that troubles me – and why I have not written about it yet – is that this line of thinking believes there are fixed bid amounts for players. But no player is worth any exact amount, and bidding $21 when you think a player is worth $24 is just part of the normal process. This is unlike snake drafts where there are specific spots you have to target to get players. There is more thinking to be done here.
Another one from the reader forums: When planning to limit liabilities as per BABS, it’s easy to see that INJ is riskier than inj-; but is INJ a greater liability than EX? If they’re equal, are they really equal, or is one more equal than the other? (thank you George Orwell)
Good question! They are the same, but different. Both liabilities affect the variability around a player’s expectation. INJ is a risk factor that affects playing time, primarily. EX is a risk factor that affects skills, primarily. Is one more risky than the other? Given that INJ is mostly downside risk and EX could be upside or downside, you might say that EX is a lesser risk. BABS thinks that risk is risk, but you can build your roster according to your own interpretation.
Drew Smyly’s K asset was upgraded in 2020 based on 26 IP. This is about a third of the already shortened season. I feel like a warning about small sample size is appropriate whenever a change in asset class is based on extremely limited data. Shouldn’ t it?
Maybe. Smyly was a unique case because he was more of a dominant pitcher before he lost two years to TJS. His terrible 2019 season was more of a “working my way back to you” year, but 2020 showed signs of his former self, albeit in the small sample size. A 2.5 mph increase in his fastball was particularly notable in pushing his BABS rating to (KK). That rating does not come without big fat liabilities, so we do have the red flags out.
Has the short season Central Division bias been baked into this year’s ratings? This wouldn’t be a liability for Darvish or Flaherty, but the weak NL Central offense should perhaps be held against guys like Woodruff, Bauer, and Musgrove who had their best seasons in 2020.
The short answer is yes. Here is the longer answer… You can track the trends in a player’s annual BABS levels and draw rough conclusions, like we do in the Player History Scans, but it’s not a perfectly linear system. The reason is that league-wide skill levels are not stagnant; they move each year. If a player’s strikeout rate is 25% one year and 27% the next, his BABS rating might go down if league-wide K% levels rise disproportionately. So, yes, all these Central Division pitchers saw their BABS rating tick up slightly (and Flaherty’s dropped — why wouldn’t it?) — Bauer improved from (e,KK) to (ER,KK) and Woodruff (ER,k) to (ER,KK) but these are not huge gains. Only Musgrove took a leap, but that was in part due to his move to San Diego.
Should streakiness/extra variance deserve a liability of its own? Trevor Story and Jose Ramirez have roughly the same number of assets, but I am taking Story every time because he’s immune to slumps and has a much higher floor whereas Joram is as streaky as they come, with the floor being the first half of 2019. I feel as though some warning is needed for players like him.
If only it was that simple. The Mayberry Method at BaseballHQ.com — the forerunner of BABS — does have a Reliability rating for Consistency. In essence, that is what you are asking for. It is a metric that identifies players who’ve been inconsistent year-to-year in the past. While interesting, what it has revealed is that most players are just naturally inconsistent. In this year’s Baseball Forecaster, only 6% of batters and 13% of pitchers received a grade of “A” in Consistency. Those who rated an “F”? 28% of batters and 22% of pitchers. It’s easy to cherry-pick players like Story and JoRam at opposite ends of the spectrum (and Story has a rating of “A” and JoRam “F”, so the system does work), but most players are somewhere in between. When developing BABS, I did not think that variable was intrinsic enough to a player’s makeup to include, but more important, it is not really projectable.