Sloan Analytics: Farhan Zaidi on A’s Analytics

Farhan Zaidi isn’t the most famous member of the Oakland A’s front office — that would be Billy Beane — but he might be the smartest. Currently in his fifth year as the club’s Director of Baseball Operations, Zaidi has a Ph.D in economics from the Cal Berkeley. He also has a job description befitting the A’s Moneyball reputation. According to his bio, his primary responsibilities include “providing statistical analysis for evaluating and targeting players,” and “analyzing data from advance scouting reports.”

Zaidi talked about his team’s saber-slanted approach to roster construction between presentations at last weekend’s MIT Sloan Sports Analytics Conference.

——

Zaidi on the value of taking risks:
“If you’re a small or mid-market team, you’re compelled to engage in a high-variance strategy. We don’t want to just run our operation the same way everyone else does, with the same blend of stats and scouting, In some sense, the optimal strategy is to take risks. We make trades that might be perceived as risky. Sometimes they pay off, like Josh Reddick. Sometimes we acquire guys it turns out we were wrong about.

“If there isn’t some residual between how you evaluate players and how other teams evaluate them, then you’re just using industry values to put together the second-lowest payroll team in the league, and likely end up being the second-worst team. You kind of have to take those risks to outperform your payroll. Sometimes it’s going to backfire, just because you have to try to do something different.

“If I was the Yankees, that wouldn’t be my strategy. All I’d have to do is be as good at scouting and analytics as everyone else, and my payroll gives me the advantage. If you don’t have that advantage, you have to do something else.

“We know the risks when we make deals. There’s never going to be a time where I read something about a guy we traded and go, ‘Oh my god, we didn’t know that.’ We can be wrong, but we had all the information. We just interpreted it, or judged it, wrong.”

On Josh Reddick and residuals: “We talk a lot, in our office, about the residual. We define residual as the difference between how we view a player and how we think that player is perceived within the industry. Basically, we go after players we feel have a positive residual, guys we like better than everyone else. Anything you do that plays to that, or amplifies the differences between your evaluation and other team’s evaluations, creates more opportunities to find value.

“Our assessments are based on both numbers and scouting. As an example, our scouts thought Josh Reddick was a starting right fielder in the big leagues. They thought he was a 50, basically. On top of that, we had his performance data from Boston, in 2011, where he was worth 10 runs in the field in half a season. The scouting reports matched that.

“A lot of times, particularly at the corner outfield spots, you’ll have a guy who doesn’t have a great defensive reputation, but his fielding metrics are above average, or even way above average. You don’t know what to make of that. But when the scouting reports and metrics line up, you really have something.

“The final part of it with Reddick was that we believed in the bat. We didn’t think he was going to hit .300, but we believed in his bat more than what the popular perception was. People wondered whether he would hit enough to be an everyday big league outfielder. We thought he would hit .250 with 20 homers. He wound up exceeding that, and even if he regresses to a .250/20 level, and is worth 10 runs in the field, that’s still a very good player. He was an all-star-caliber player last season.”

On building from the bottom: “We approach improvement from the bottom-up. We talk a lot about bang for the buck — marginal runs and marginal wins per dollar spent. We’ve found that the best ways to improve our team were to limit the downside. We manage the roster from the bottom.

“A back-up catcher that plays once a week isn’t a guy you’d necessarily see as a focal point of your team, but over the course of a season, or even a half season, having an average player there as opposed to a replacement-level player there can make a difference. That was one of the reasons we went out and got George Kottaras last year. Our back-up catching situation wasn’t good, and he had some value.

“This year we acquired John Jaso, although he forms more of a tandem, as opposed to being a back-up. George was a guy who was only going to play once or twice a week. Jaso gives us some insurance for Derek Norris, who we like a lot. We want to bring along Derek at whatever pace he’s ready to play. Jaso is a guy who could play two or three times a week, or he could play four or five times a week. We have some flexibility there and will figure it out as time goes along.

“Jaso walked more than he struck out and he had over a .400 OBP against right-handed pitching. For a catcher, that’s pretty terrific. He is kind of the traditional Moneyball OBP guy. What’s interesting is last year’s team wasn’t what you’d call a traditional Moneyball lineup. We set the league record for strikeouts, we didn’t walk a lot, and we didn’t have a high OBP, but we hit a lot of homers. Jaso adds versatility to our attack, as he’s got good plate discipline and is going to have some long at bats. He’s a nice complement to our lineup.”

On providing data to advanced scouting: “There is a very deep-rooted conventional wisdom within advanced scouting. Opposing hitter reports and opposing pitcher reports have looked basically the same for the past 10 years. We’re trying to advance that. We’re asking ourselves, What information do we want our hitting and pitching coaches conveying to our players beyond what they’re doing right now?

“If you go to a hitting or pitching coach and ask, ’What do you want?,’ they’re happy with what they have. They have their hands full with the job at hand and don’t spend a lot of time thinking about what other potential pieces of valuable information are out there. We have a broader perspective. We constantly have people coming to us and talking about new advanced report services, offering new products and data sets, and that clues us in to what more we could be doing.

“On an opposing hitter report, you might have a player’s batting average against 0-1 fastballs or 0-2 fastballs. Guys would use batting average in very specific, narrow, circumstances to determine whether to throw a pitch in a certain count. Batting average is certainly not the right metric to use in those situations. We also need to understand how to slice the data into meaningful segments where you’re getting statistically significant sample sizes. You don’t want your manager necessarily viewing a match-up favorably if a guy is 2 for 3 against somebody. It’s the same sort of sample-size challenge that comes with advanced reports.

“We try to analyze skill, not results. Ultimately, what you want in an advance report isn’t batting average on fastball counts. You want something more along the lines of skill level, like how often a guy makes hard contact on a fastball on an 0-1 count. That approach is very applicable in this area.”

On proprietary information: “Analytics, today, is kind of like 30 guys with 30 radar guns: That’s not meant as disrespect to scouts. I go out and scout, and a lot of times I’m one of the guys holding up a gun. It’s more of an analogy to recognizing what data is commoditized, and what data really gives you a competitive advantage. Knowing that — knowing when you’re using data that other teams have access to, versus data that is legitimately proprietary — is an important point to be able to recognize.

“Everybody is holding up the gun and everybody writes down the reading like everyone is collecting performance data and evaluating it. There’s a lot of stuff that goes on behind the scenes that every team is doing individually, but really, much of that exercise is just running in place.

“The question of what you’re doing that other teams aren’t is a tough question. You have to give your competitors respect. A lot of things you’re thought about, they probably have as well. You have to try to go a step beyond and do it better.”




Print This Post



David Laurila grew up in Michigan's Upper Peninsula and now writes about baseball from his home in Cambridge, Massachusetts. He authored the Prospectus Q&A series at Baseball Prospectus from February 2006-March 2011 and is a regular contributor to several publications. His first book, Interviews from Red Sox Nation, was published by Maple Street Press in 2006. He can be followed on Twitter @DavidLaurilaQA


52 Responses to “Sloan Analytics: Farhan Zaidi on A’s Analytics”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. semperty says:

    Love getting these quotes! Thank you so much!

    Vote -1 Vote +1

  2. Highball Wilson says:

    Great interview.

    Vote -1 Vote +1

    • Persona non grata says:

      Don’t think this was an interview. More of a dictation from the conference.

      Still awesome though.

      Vote -1 Vote +1

  3. Kris says:

    Really interesting article, I always love reading about the behind the scenes stuff that the A’s are doing.

    Vote -1 Vote +1

  4. Mike says:

    Joe Posnanski wrote at HardBallTalk while he was talking to Zaidi at Sloan, Zaidi said the A’s performance model actually found Cabrera to be slightly more valuable than Trout last year. Make anything of this?

    http://hardballtalk.nbcsports.com/2013/03/04/revisiting-trout-vs-cabrera-mvp-debate-with-a-twist/

    Vote -1 Vote +1

    • Kris says:

      “Well, that’s not exactly right. He was quick to say that the difference between the two was so slight as to be almost invisible — they were, for an intents and purposes, in a virtual tie. But their system did have Cabrera ahead by the tiniest of margins.”

      It’s hard to say why this is because we don’t know much about the system they use obviously. Maybe they don’t value base-running as highly as others or something along those lines.

      And anyone who’s honest with themselves probably agrees that it was a very close race last year.

      Vote -1 Vote +1

      • Ty says:

        Billy Beane has made it very apparent base running means very little to him, so that’s a very real possibility.

        Vote -1 Vote +1

        • Ben Hall says:

          Has he said that? The team he built last year had the second best BsR numbers in the league, according to the FanGraphs leaderboard.

          Vote -1 Vote +1

        • Kris says:

          Beane has said that in the past but it does seem like they shifted their philosophy a bit last year simply because they didn’t have a great team OBP.

          I was just offering that out as a possibility based on what he said in the past but my broader point is that it’s hard to know why Zaidi makes that claim simply because we don’t know much about his system and he’s obviously not going to reveal much of it. Zaidi did talk about defense in that piece above so I imagine that’s important to him but I never once saw him mention baserunning which is a large part as to why people thought Trout had the better season.

          Vote -1 Vote +1

        • payroll says:

          I assume the A’s discounted Trout’s baserunning value due to the fact that the Angels and Tigers were both top 5 offenses, where baserunning has a smaller marginal value. BSR and fWAR values don’t account for that player’s team-specific run-scoring environment, I believe.

          Vote -1 Vote +1

        • That’s not entirely true. The A’s only care about baserunning that works; they let Coco Crisp do what he wants because he’s an excellent baserunner and base-stealer. Everyone else has a much shorter leash. It’s simply an extension of Moneyball: don’t make outs (and certainly don’t make them on the bases).

          I suspect they discount baserunning in Trout’s case due to his manager being rather run-happy.

          Vote -1 Vote +1

        • Bab says:

          Where has Beane said this? I’m not immediately doubting you but I’d like to see the context of the statement if you can provide it.

          Vote -1 Vote +1

        • Jay29 says:

          @Kris, Beane is always going to be looking for the market inefficiency, the part of the game where a low-budget organization can build a better team than their budget would dictate, as Zaidi says. That means shifting your focus as the rest of the league shifts theirs — right now OBP is clearly in vogue, so the A’s are looking at other market inefficiencies through analytics. What those are isn’t entirely clear because, well, it’s probably a lot more complicated than it was in 2002.

          Vote -1 Vote +1

        • Forrest Gumption says:

          Beane changes his outlook and ideals every six months or so, I wouldn’t quote him from years ago. If anyone changes with the times, its him.

          Vote -1 Vote +1

      • vivaelpujols says:

        I don’t think it was a close race at all. BPro, FanGraphs and B-R each have Trout 2-4 wins better than Cabrera last year. I have no idea what the A’s have that would lead them to think the two were even. I’m guessing that they are answering a difference question than WAR. They are probably looking at projection rather than single year value.

        Vote -1 Vote +1

        • Kris says:

          Ahh, yeah you probably have a point there. Since we don’t have a direct quote from Zaidi it’s hard to say for sure.

          Vote -1 Vote +1

    • Travis L says:

      I would guess that it was context-adjusted.

      Vote -1 Vote +1

    • bjoak says:

      WAR does not account for the luck inherent in hitting the way it does for luck in pitching. Trout had a lot of luck on balls in play, whereas Cabrera’s numbers were in li ne with his skill set. It is basically the same reason why most analysts expect regression from Trout this year but not so much from Cabrera. As Zaidi said, it is about assessing skills rather than results.

      Vote -1 Vote +1

      • Doug Gray says:

        I don’t know that we can say that Trout had a lot of luck on balls in play. While it appears that way on the surface, we don’t have enough “other” data to suggest where his true BABIP talent lies at this point in his career. I wouldn’t bank on him repeating his BABIP from 2012 by any means, but maybe he is one of the few outliers due to how hard he can hit the baseball and just how fast he actually is. It is a rare combo to be able to do both as good as he can, so there is a chance that he just is that rare outlier type of guy who can beat BABIP somewhat.

        Vote -1 Vote +1

        • Steve Staude. says:

          Yeah, you beat me to it. Trout overachieved in his BABIP, but perhaps not by as much as people think. When you consider Jeter’s .354 career BABIP, something in the .360s for Trout seems pretty reasonable to me.

          Vote -1 Vote +1

        • Blofkin says:

          @Steve Saude
          5 hitters in history have BABIPS over .360 (min. 5,000 PAs). Outside of Jeter, no hitter in the last 50 years has a BABIP over .350. While he MAY be able to sustain that over his career, there’s no way anyone can predict it to happen, hence the regression.
          Jeter may well have a generational ability when it comes to BABIP. Not to argue Trout does or does not, but sample size and precedent suggests it is incredibly unlikely he maintains anything close to that BABIP over his career.

          Vote -1 Vote +1

        • bjoak says:

          No offense guys but as soon as you start trying to guess his BABIP, you’re putting yourself at a disadvantage. You have to look at the factors that feed into BABIP. His batting average should have been .298 last year according to his peripherals. With a bit of regression to his power and speed he’s likely to end 2013 a bit below that. I have him at .287. While it is expected for some guys to overachieve in BABIP, it is very rare that players consistently beat out (or change) their expected betting average.

          Vote -1 Vote +1

        • Steve Staude. says:

          blofkin: yeah, I mean, I wouldn’t bet on it, but it doesn’t seem all that improbable to me. And I’m only talking about this year, not over his career — he’s going to lose his speed eventually. Also note that BABIPs are historically on the very high side over the past couple decades, and that there are 5 present-day players with over 2000 PAs and a BABIP over .350, a bunch of players in the .340s, plus Austin Jackson, who has a .370 over 1960 PAs.

          Vote -1 Vote +1

  5. rageon says:

    We constantly have people coming to us and talking about new advanced report services, offering new products and data sets, and that clues us in to what more we could be doing.

    Translation: “If you bring us your ideas, we won’t buy them, we will take them.”

    +15 Vote -1 Vote +1

    • Dirck says:

      Getting good ideas from all sources as cheaply as possible is a very intelligent way to run a business ,rageon . It sucks to be the guy giving your ideas away for nothing ,but one should be smart enough to avoid that .

      Vote -1 Vote +1

    • Travis L says:

      Ideas are worth little. Execution is what matters.

      It’s also important to note that in analysis, you really need to own the data to fully utilize it.

      EG, I “understand” what ZIPS projections are, where they come from, etc. But Szymborski knows the results inside and out, almost to an intuitive feel. Unless you build it yourself, it’s awfully difficult to internalize the analysis.

      Vote -1 Vote +1

      • Bab says:

        Dude, this is a cool little collection of ideas except for the slogan “Ideas are worth little. Execution is what matters.”

        I’d hate to waste time executing analysis on projects with no ideas backing them.

        Vote -1 Vote +1

        • Dustin says:

          I think what he’s getting at is that good ideas aren’t in short supply, but the ability to turn those ideas into something useful is much rarer.

          Vote -1 Vote +1

    • Aaron says:

      Well there just isn’t enough time in the day to implement other people’s ideas. When someone pitches the idea to you, you think “What problem does it solve” “How long will it take to implement by myself”. If its over 100 hours you probably just buy it.

      Vote -1 Vote +1

  6. jwise224 says:

    This is so fantastic. Huge thank you for putting this up!

    Vote -1 Vote +1

  7. JeffMathisCera says:

    Interesting comment about Jaso’s OBP adding versatility to a SLG heavy lineup. It’s in line with the FG article from a little while ago that said Callaspo was more valuable than Trumbo for the very same reason.

    Vote -1 Vote +1

    • Forrest Gumption says:

      He’s not well-liked amongst A’s fans, but i’d love to see the other heavy-OBP guy, Daric Barton, in this lineup alongside John Jaso. Now 1B isn’t a position that needs power on the club, with Reddick , Cespedes, Donaldson and Moss all likely providing 20+ HR power, Barton’s unpopular “defense+OBP+nothing else” actually works well.

      Brandon Moss is bad at 1B and Barton is a gold glove 1B, so Moss should be DH and Barton 1B. Sadly Melvin is in love with platooning, there is a pretty outstanding team there if he wants to use it more traditionally.

      Vote -1 Vote +1

  8. Shalesh says:

    A team which outperforms its projection by definition has players with surprisingly good years and low payroll teams typically have low projections so they are more likely to have these surprises. Zaidi himself says the A’s were lucky with Reddick having an All-Star year. I doubt the media was lining up to hear Zaidi’s thoughts when the A’s were winning 74/81/75/75/76 games in the 5 years prior to this one. Zaidi does make a point that GM’s should be compared to their payroll peer group since the context is different across payroll groups. Why did the A’s and O’s outperform their projections in 2012? Will they be able to sustain their success in 2013 & 2014? How are the Rays able to do it year-after-year? How often do multiple teams in the same year with low projections/payrolls surprise and make the playoffs?

    Vote -1 Vote +1

    • West Coast Hard Baller says:

      The Rays don’t do it year after year. They have had about 5 good years which were preceded by a TON of terrible years, much worse than the A’s last 5 years…which were preceded by 8 or 9 very good years. You can’t just look at the last 6 years and say the Rays have been better…you have to look at the entire history.

      When you lose 100 every year and get a top 3 pick, you should have a good period coming and that’s what the Rays realized.

      Vote -1 Vote +1

    • mlbfan23 says:

      rays drafts from 1999-2008
      chose in top 5 8 times
      had 4 #1 overall picks

      A’s over that same stretch had

      8 above .500 teams, 5 playoff appearances]

      So everything goes in cycles, A’s were mediocre/bad from 07-11. 76/75/75/74 wins. 2010 they were .500 team

      If they totally sucked in those yrs you can draft higher potential superstars instead they settled for the simmons, weeks, grant greent, michael choice, sonny gray. Hopefully Addison Russell changes that

      Beane isnt perfect he’s made bad deals and players they thought were good turned didnt plqay as expected (barton, buck, suzuki, gallagher, de los santos, taylor, etc)

      Vote -1 Vote +1

  9. OaktownSteve says:

    Great article. A’s are such a fun team, on the field and off.

    Vote -1 Vote +1

    • baycommuter says:

      How true. Whoever could have guessed the formula the A’s have developed over the years?
      1) Select players who offer good value based on advanced proprietary metrics.
      2) Encourage them to wear creative facial hair, do odd dances and throw pies on one another.

      Vote -1 Vote +1

  10. DD says:

    “Batting average is certainly not the right metric to use in those situations.”

    Charlie Manuel – “Ridiculous!”

    “You don’t want your manager necessarily viewing a match-up favorably if a guy is 2 for 3 against somebody.”

    Charlie Manuel – “Preposterous!”

    Vote -1 Vote +1

  11. channelclemente says:

    Excellent article.

    Vote -1 Vote +1

  12. The bit on residuals should be posted on Rotographs. I’ll paraphrase:

    If there isn’t some residual between how you evaluate players and how other teams evaluate them, then you’re just using ADPs to put together an average team in the league, and likely end up being the average team. You kind of have to take those risks to outperform average ADP. Sometimes it’s going to backfire, just because you have to try to do something different.

    Vote -1 Vote +1

  13. Steve Staude. says:

    It makes a lot of sense that bad teams can afford to take bigger gambles — “when you ain’t got nothing, you got nothing to lose.”

    Vote -1 Vote +1

    • Antonio bananas says:

      That and stability is a tangible asset that teams will pay for. If a guy has high variance, his price tag goes down. If he’s incredibly consistent, it goes up.

      Vote -1 Vote +1

  14. Frediot says:

    I was there-the whole baseball panel was phenomenal. Zaidi was hugely insightful and is clearly a very very smart guy.

    Vote -1 Vote +1

  15. “Sometimes we were wrong…” like the Holliday trade

    Vote -1 Vote +1

    • Blofkin says:

      heh

      Vote -1 Vote +1

    • baycommuter says:

      I don’t know how metrics tell you that a player is going to hate your team and your league.

      Vote -1 Vote +1

    • mlbfan23 says:

      yes the wallace, taylor part of that deal just didnt progress. I still have hope for shane peterson who might turn into a solid player. But prospects that were in that previous rebuild have been passed over for the most part. Its all about the next wave of prospects and their 2012 draft who has nice upside.

      Vote -1 Vote +1

    • Forrest Gumption says:

      Its funny how the Holliday trade ruined the team for the next 4 years. If they don’t make that trade, and instead add Carlos Gonzalez and Huston Street to that 2010 team that went .500, they’d have been buyers at the deadline and a possible playoff squad.

      Its funny because for years, before people realized that overpaying for players worst seasons wasn’t entirely smart, people would constantly badmouth Beane for trading his star players. The one time he does go out and got a superstar, along with other “big names” (Nomar & Giambi) it failed dismally.

      Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>