I’m going to try to explain to him that many of these stats are simply ways of sorting out what’s luck and what’s skill, and ways of separating the average performers from the exceptional.

Ack!!!

Here's a word of advice -- DO NOT USE THE WORD "LUCK". Do not let it pass your lips until the end of the interview when you say "Joe, I wish you good luck and happiness." The debate is lost the second you use the word luck in the way you just used it. Talk about repeatable skill or some such if you want, use the word "random" if you're completely at a loss. Maybe don't even worry about arguing against notions like clutch but instead argue that clutch performance ends up being only a very small part of a player's value. So, no, I don't have any magic words for you but you'll get a snotful on here if you use the word "luck", on sports radio you're toast.

And ways of separating the average from the exceptional? HR and RBI and even BA already do a pretty good job of that thanks. They get some wrong but not many. Joe's obvious retort here is "I don't need fancy statistics to tell me that Albert Pujols is better than whatever schlub they've gotten playing first base in Kansas City."

I think you're better off trying to go for two arguments. On the one hand, one set of stats are trying to bring together offensive, defensive, baserunning and durability value to assess a player's overall contribution. On the other hand, another set of stats are trying to break performance down into its smallest components (in ways that can be reliably measured). Buried in both of those is the idea of positional adjustment -- fans can understand that SS (generally) hit worse and a big part of the reason for that is they need to be defensive wizards. These stats are trying to even out those differences such that a good-fielding SS might be worth more than a slugging LF. To do that, we need to measure defensive value as best we can. For example, one SS might make 20 errors and another 10 but if the first SS gets to 30 more balls that the other doesnt, that's a better performance.

But if you want some really winning points:

1. Stats show Manny's defense is horrible, doesn't Joe agree?
2. Stats show that most FA are horribly overpaid, doesn't Joe agree?
3. Stats show that X, Y, Z are the most overpaid players in the game this year, doesn't Joe agree (assuming none of X, Y or Z are Ryan Howard or other obvious superstars ... how does Joe feel about Jason Bay?)

And for god's sake, you've got to be prepared for the Jeter question. I'd suggest going for a joke -- "you kidding, we all want to be him, he's got like 10,000 BAR -- that's babes above replacement Joe" -- and then saying that the stats show Jeter's one of the (where is he at now?) 6 best SS in MLB history, we love Jeter.

Heh, Walt beat me to the punch with the LUCK advise.

Heh, Walt beat me to the punch with the LUCK advise.

Doesn't everyone who watches baseball think there is luck involved. We have all seen a player hit 4 screaming liners right at the defense and another player hit a "ground ball with eyes" or a "dying quail". I think that is understandable.

NERDS!!!

GO HOME AND PLAY WITH YOUR COMPUTAH.

I do wish you luck, sir.

I think people also need to understand at some level that sabermetrics really helps more at the front office margins, by helping to put minor league stats (where things like park size and oppositional strength have major effects) in proper context, helping determine if a potential 4th outfielder is on a major decline, looking for small inefficiencies or hidden strengths and weaknesses in a player that your coaching staff has shown the ability to exploit.

That's where I think a sabermetrically-inclined front office and analyst can really show off their stuff.

Doesn't everyone who watches baseball think there is luck involved.

Sure. The problem with luck (unlike chance, or random variation or other better options) is it's a potentially loaded word, particularly outside these walls. Used correctly, it can define that which happens by chance - the dying quail, the seeing-eye grounder. But it's also historically been misused by sports fans to describe a result that we don't like, with or without supporting evidence. Tell someone their victory was luck, or they got lucky, and the immediate response is that person will go on the defensive. If you're having a conversation with someone disinclined to believe you, that's a situation to avoid.

Frankly, I would hope you'd explain that sabermetrics is not just about rate and uber stats designed to measure value, which I find to be a somewhat boring applications of the process. Perhaps I'm in the minority, I'm much more interested in the developments that have enhanced our understanding of how the game is played, rather than whether this guy is a marginally better than this other guy.

The mere fact that they are having this segment on the show is a victory (albeit a very small one) for advanced stats.

The mere fact that they are having this segment on the show is a victory (albeit a very small one) for advanced stats.

Yep. It's a sure sign of progress. Think of how inconceivable this would have been 10 years ago, or five years ago.

I think the most basic issue out there right now is that of sample sizes. Save for the truly narrowminded I think most baseball fans (and certainly people inside the game) understand that there is something beneficial from having access to more information. I can see a discussion of stats being quickly derailed by a host saying "oh, so a guy hits four line drives but goes 0 for 4 so you think he stinks?" This is where the SS becomes so useful. I think sample sizes need to be much larger than most people realize to be useful.

Separately, during the discussion of PECOTA on MLBNetwork last year the host (Vasgersian I think) made the point that the advantage of a stat based approach is that while scouts only see a few games, stats incorporate everything a player does. I thought that was a really good point.

Separately, during the discussion of PECOTA on MLBNetwork last year the host (Vasgersian I think) made the point that the advantage of a stat based approach is that while scouts only see a few games, stats incorporate everything a player does. I thought that was a really good point.

It is, although it's actually just an amplification of the sample size issue, which cannot be overstressed.

These guys on WFAN have been talking about OBP for over a decade now, ever since Gene Michaels stocked the Yankees lineup with guys "who takes a walk" or "works the count." It's almost boilerplate for every host who talks about the Yankees at this point. OBP might not be an advance metric, but I think it'd help to show Joe that sabermetrics isn't as esoteric as he thinks it is since he's been using it to gauge the Yankees' performance for a while now...

Joe Benigno is very, very old school.

I haven't listened to WFAN regularly in over ten years, but my recollection is that while Benigno is old school, he is not 100% convinced that he knows everything like Francessa, so he is willing to have an honest conversation.

Topics will be the basic stuff like WHIP, WAR, BABIP, FIP and rate stats in general. I’m going to try to explain to him that many of these stats are simply ways of sorting out what’s luck and what’s skill, and ways of separating the average performers from the exceptional.

Echoing Walt and taking his advice a bit further -- don't start out trying to tell old-school Joe Benigno and his listeners that you have a better way of separating the average player from the exceptional. Just define the terms and go from there. "So 'WHIP' is just walks and hits per inning pitched, Joe. Wouldn't you want to know how many baserunners a guy allows before you decide to pay him $19M a year for seven or eight years?" Good luck getting anywhere with WAR, though.

Doesn't everyone who watches baseball think there is luck involved.

People believe in bad luck, sure. But good performance is rarely ascribed to good luck -- the player works hard or his talent is finally coming through or, when all else fails, he's on a hot streak or seeing the ball really well right now. C'mon, tons of Mets fans thought Jeff Francoeur's 2009 second half was repeatable ... which, of course, there was some chance it would be!

And consistent good performance -- that's never ascribed to luck and quite possibly that's a good call. But, again, this is a NY sports radio show. Derek Jeter is a clutch god and "luck" is not an acceptable explanation for that. Hell, even if you had stats to show that Jeter hasn't been a clutch god, you won't convince these listeners otherwise.

As to other posts, I agree that sample size is an understandable issue for fans although I don't think most fans consider a season to be a small sample size -- and from a "politcal" standpoint, the fact that it takes the stats 3-4 seasons to be worth anything isn't a winning point at all. Stats "incorporating everything" seems likely to be a pretty good approach to that though.

Good luck getting anywhere with WAR, though.

What would be the point of introducing WAR? Why would anyone care about it? What practical use does it have?

<i>- Good luck getting anywhere with WAR, though.

What would be the point of introducing WAR? Why would anyone care about it? What practical use does it have?

Nothing, absolutely.

What would be the point of introducing WAR? Why would anyone care about it? What practical use does it have?

Nothing, absolutely.

This bears repeating.

What would be the point of introducing WAR? Why would anyone care about it? What practical use does it have?

Nothing, absolutely.


This bears repeating.

I believe that you need to gather people's attention and have them listen to you.

I don't have any experience talking about things like this on the radio, but I do have experience both in trying to explain advanced stats to people who are hostile to them, and experience trying to teach students things that seem counterintuitive to them.

Based on this, my advice would be to avoid getting into details and instead focus on generalities. For example, in explaining DIPS or FIP, don't trot out the formula. Instead, say something like "While good pitchers may be able to get hitters to make more weak contact, this skill is more variable from one year to the next than other skills, so that a players FIP may be a better prediction of what he will do next year than his ERA or W/L record." For clutch hitting, avoid arguing about whether or not it is luck, simply point out that it is a fairly unimportant part of performance. In general, point out that the motivation of most advanced metrics is to help make better predictions about what will happen than using traditional metrics would.

I think if you start focusing on smaller points or debates about luck you will lose the audience and the hosts and leave yourself open to derision. Instead, focus on the big picture that advanced stats help us better understand what wins ballgames.

What would be the point of introducing WAR? Why would anyone care about it?

I couldn't disagree more.

For the casual fan, WAR is perfect. Win Shares is actually one of the things that caught on pretty well with mainstream media. Dewan's Fielding Bible caught on quite well in part because it finally put a single, simple number to something sportswriters had been writing about forever. (Sure, anytime their opinion was in conflict with Dewan, their opinion won ... not that I ever do that!).

The "beauty" of WAR is that it brings together offense, defense and baserunning into a single number. A pretty understandable number. Every fan understands that a lot of Ichiro's value comes from baserunning and defense ... stats like WAR are the only stats that get at that.

Brett Gardner ... 3.4 WAR ... 2nd best player on the Yankees this season ... largely because he's been great defensively. What Yankee fan doesn't want to hear that?

And what else is he supposed to go out there with? EQA? VORP? FIP? PECOTA? pitch/FX? WPA? UZR? OPS+? ERA+?

#21 Walt I'd take a different approach though. There's a fair chance that a "what about baserunning" (or defense, or productive outs) question will come up. Then if you wish you can talk about how we now know how often a typical baserunner goes first to third on a single to right. (and every other baserunning situation) And how the particular mix of fielder, runner, out situation and score all influence the situation. To the point where we can make a reasonably accurate estimate of the value of baserunning.

Also useful for mythbusting. Somebody called Rolen a historically great baserunner during the game last night. That can easily be checked.

UPDATE...Called off.

I received a call from Joe & Evan show producer Pete Billotti that the appearance Evan Roberts set up with me (on air) was going to have to be canceled. With the passing of Yankee owner George Steinbrenner, WFAN is going wall to wall with coverage on his passing. The call came as no big surprise to me, I was expecting it when I heard the news about George. Mike Francesa mentioned on air yesterday that this story looks even bigger than the passing of Mickey Mantle, which lasted for close to a week of non-stop coverage for WFAN.

Hey Repoz, thanks for the link. I received a call from the producer Pete Bilotti this afternoon, and it’s been rescheduled for 2:00 on Thursday (7/15). Hope your readers can tune in, but again I have my doubts about how much depth we’ll get into. I think I’ll just make the case that if fans want to criticize their GM, they should put themselves in his chair. Once there, SABR tools are a great way to asses value to a player, sort out luck from skill and find undervalued assets.

The "beauty" of WAR is that it brings together offense, defense and baserunning into a single number.

That's why it's useless. People are skeptical of being able to assign a single number to either defense or baserunning to begin with (and even offense, although to a lesser extent). If you try to tell people who are suspicious of sabermetrics that you can skip all that messiness and go ahead and assign a single number to the entirety of a player's value, based on stuff you haven't even tried to explain, all they're going to do is pick holes in your argument.

Take Brett Gardner. Tell people he's the second best player on the Yankees, and they'll think you're an idiot. Not even his mother would believe that. What you would need to do is explain what the advanced defensive metrics measure, and use Brett Gardner as an illustration, eventually getting to the point where you can say that Gardner has saved X number of runs on defense. Then you might even be able to say that, combined with his hitting and baserunning, he's X runs better than Jeter.

But if you start with WAR, the discussion won't be about Gardner but about whether WAR is valid or not. And that's an argument you're not likely to win.

I wonder if it would be worth it to note that things like Shandler's xERA (among many others) are a better predictor of future results than ERA itself.

So take 20 guys' ERAs, and ask if they will go higher or lower in the second half.

If you picked 20 overachievers by xERA, seems reasonable to think that you'd come out ahead.

Of course, based on sample size, if you lose 12-8, it will be 25 years before SABRmetrics ever reappears at the station, so there's that.

Take Brett Gardner. Tell people he's the second best player on the Yankees, and they'll think you're an idiot. Not even his mother would believe that.

Fine, but it;s still possible he's having the second best season on the Yankees, which is all 2010 WAR is saying.

No one thinks Robinson Cano is the best player on the Yankees either, but he was sure pretending to be for a couple months.

But if you start with WAR, the discussion won't be about Gardner but about whether WAR is valid or not.

The argument doesn't even have to be about whether WAR (TM) is valid. Even sabermetricians can disagree about how WAR calculates certain things. The point of it is in trying to quantify that which has not historically been quantified. OK, X hit .320 with 30 HR -- how many runs is that worth, compared to an average/replacement player? Let's look at all his baserunning exploits and determine, as best we can, how many runs that was worth. Let's study his defense, and instead of just assessing him as good, average, or bad, let's estimate how many runs that's worth. Put it all together and we have an estimate of how many runs X is worth in all facets of the game, rather than just blindly trying to compare the .320, 30 HR hitting 1Bman to the .270, 30 SB shortstop.

Best:

With Derek and Arod clearly aging I don't think it's outrageous to suggest that Cano is the best. Mark? I like the productive up the middle types.

I thought of qualifying that comment but decided not to. You're right, it is possible Cano actually is their best player right now (essentially an unanswerable question). I just don't think most fans would identify him as such.