THE BOOK--Playing The Percentages In Baseball

It doesn’t sound like the underlying data will be public for the foreseeable future.  Will ratings based on the data be made public?

Doubtful, since the teams will control the data and will have no incentive to release rankings based on their proprietary metrics.

Would like to know what you guys think about the idea of keeping the data proprietary vs. making it public.  I mean, I know you guys would like the data to be public in some workable form because people who contribute on this board would love to see it.  But even beyond that, is this even good for the Boston?

It’s obvious (I think) why Boston would want to keep it private as they would “only” be competing against the analysts for the other 29 clubs and have a leg up on most. Short run, it works.

But long run...don’t they want the state of the art advanced for their own benefit? Or develop analysts who know how to work with the data?  Or are these aspects minor in effect compared to trying to keep a competitive advantage?

And, if the data is kept within the teams, what’s to stop a group of teams from sharing their analysis to try to combat the head starts of clubs like Boston?

All the small market teams that don’t want to spend 250,000$ in a quant lab will want the data released.

MLBAM is owned by the 30 teams.  MLBAM is hugely successful, and they will have more pull than the Boston Redsox. 

If the Redsox and Yanks are going to push for this to remain private, I can’t imagine a better reason for the other 28 teams to push for this to be public.  Those teams would want it private because they can buy their way easily for that to happen.

My bet is that the public will have huge appetite for this, ESPN and Fox will beg for hang time and running times, and those two networks realize that the general public will do wonders with the data.  And there’s NO WAY MLBAM will so no, just to appease the Redsox and Yanks.

I think there is no doubt that the utility of PITCHf/x data to MLB teams has benefited greatly from the data having been made public.  Were the HITf/x and FIELDf/x data made public, I believe the same would be true, although perhaps to a lesser extent now that many MLB teams have hired analysts who cut their teeth as unpaid amateurs (as opposed to paid professionals).  On the other hand, lots of R&D has gone into developing these systems by the good folks at Sportvision.  They need to show a profit to keep doing the good work that they do.  It is hard to show a profit if they simply give away the fruits of their endeavors.  For that reason (and for no other reason) I would not expect either HITf/x or FIELDf/x data to be made public.

Tango - It’s not just the Red Sox and Yankees.  It’s every team that has invested in its own data analysis department.  The Pitch F/x summit is attended by almost half the teams.  Most of the teams that I have talked to feel that they have an analysis system that is at least the equal of any of the other teams.  None of them wants to see the information available real time to independent analysts like Pitch F/x has been, because they feel that that might compromise the investment that they have already made in their analysis departments by allowing the other teams that have not made similar investments to make use of “free” analysis by independent analysts. 

Networks and local broadcast stations will probably have access to summarized data like they use now.  The complete data dumps like the teams will get would be of no practical use to them.  Several years down the road we (the general public) may be given access to the data after it is considered less timely.

#5: that is a good point; I totally overlooked the investment.  I suppose the raw data is not likely to be public.  But Tango has a good point about media outfits wanting the data/results they can use. 

I hope at very least the raw data will become available for a reasonable fee (like BIS data, which seems attainable since sites like fangraphs and THT feature it).

Peter: right, every team thinks they are the best.  And they have a ton of overlap I’m sure.  It’s extremely inefficient.

This is the exact same thing when hockey tracked plus/minus back in the 1960s privately, and each team thought they had gold.  It’s inefficient, and there are going to be better uses with your time once some functions are centralized.

I mean, how much time does Mike spend trying to adjust the miscalibration?  Now, why have EVERYONE do that?  It’s a complete waste of 30 teams time, if that can be centralized publicly or at MLBAM.

They should be begging for this data to be public, so they can spend their brainpower analyzing the adjusted data, rather than spend their time adjusting the data to something useful.

I believe the attitude among teams that want to keep the data private is that

1. their analysts are good, better than the other teams
2. if the data is in the public domain, they could likely get better analysts/analysis, but so could the other teams
3. by keeping the data out of the public domain they are denying other teams access to the better analysts/analysis, thus maintaining their perceived advantage

But, if all the teams keep the data private, none of them really know how good the other teams are.

They would be deliberately spending more for a lesser quality product in order to keep the better stuff out of the hands of their competition.

Tango, couldn’t the exact same argument be made about scouting? “Why should every team hire an army of scouts when they mostly watch the exact same players as everyone else?”

That’s far more wasteful than stat analysis because not only do you have to hire more people to do scouting but there there are tons of travel costs as well. Teams clearly would rather try to keep as much in house as possible than be efficient.

Yes, there is inefficiency there.  From what I’ve read, there is huge turnover in scouting.  If one team pays a scout 30K, and another offers him 32K, he’ll jump.  There’s no such thing as “loyalty” in baseball any more than there is in regular workforce.  Indeed, from the sounds of it, there’s much less loyalty in baseball.

So, with all that jumping, how “unique” is each team’s scouting?

Now, what if there was a lot of centralized scouting?  Then each team can have their unique scouts do their unique thing, while they all share in the centralized scouting.

Things like the NFL combine, the NHL releases their scouting bureau rankings to the public on a quarterly basis, and so on.

When you reduce inefficiency, that’s how you grow.

***

In any case, how much do you think the average team is worth?  300MM$?  400MM$?  Something like that.  If MLBAM were to go public, they’d get a valuation 10 times that much.  A MLB team is a very small potatoes entity (Yanks notwithstanding).  MLBAM is a monster.  And they are not a monster by keeping things private for their 31 owners (commish owns 10%).  They are a monster because they feed the starving public.

Hang time on batted balls, run time on steals and close plays, throw times by catchers and outfielders… are you kidding?  That’s f-cking gold right there.  They’ll show it on Gameday, ESPN/Fox will demand it.  Do you really think some small potatoes team worth 400MM$ that doesn’t put any more money into MLBAM is going to influence anything?

Think big picture.

It will be interesting to see how things play out.  While Tango’s arguments are powerful, the same argument could have been applied to STATS and BIS, and MLBAM never bought them out and/or found a way to get that data to the public.

But Sportvision is a partner of MLBAM, unlike STATS and BIS.  The things that STATS and BIS did (hit location) MLBAM *did* make available via Gameday.

Gameday is different from Sportsvision, right? Is Gameday a partner with MLBAM? If so, I was under the impression that MLBAM only provided Gameday direction, not depth data, which is why Sean could incorporate the Gameday ‘slice’ data into infielder TZ but couldn’t do anything about outfielders.

I find it hard to believe that Gameday would have direction but not depth and trajectory data.  You need the last two items to get a UZR-type system.  MLBAM did not choose to release the data so that we could all do UZR.

By the way, I like the fact that somewhere in the article, reference was made to collecting data in one stadium from a different provider “to keep everyone honest.” Sounds like a good idea to me.

MLBAM owns the Gameday product. The location data in Gameday is unrelated to Field F/X; it’s sourced from human stringers. The reason the depth data from the stringers is unusable (other than the fact that human stringers don’t do a very good job of marking location data to begin with) is because they’re marking the ball where it was caught, not where it landed.

(And of course, Field F/X - last any data was released to outside analysts, although under an NDA - wasn’t tracking the batted ball at all.)

Colin, I’m having trouble understanding your last statement (in parentheses).

If Gameday only marked where balls were caught, not where they landed, how could they have provided “hit location data” as Tango said above?

The larger point I’m making is that MLBAM could have probably purchased BIS outright and the fielding data of STATS outright for less than the cost of developing or partnering with Sportsvision. The fact that MLBAM didn’t think it worthwhile to buy out STATS/BIS in order to release interesting data to fans leads me to think they won’t be inclined to release Sportsvision data when it comes on line.

Of course, the world keeps changing.  I could see MLBAM/Sportsvision releasing historical location and hangtime data, while keeping the other stuff in-house.  The ‘other stuff’ would be less relevant to ‘bottom line’ ratings of fielders than in diagnosing their particular strengths and weaknesses.

As Colin pointed out, FIELDf/x was not yet tracking the batted ball as of the time of last summer’s summit.  It is not so easy to do this with cameras, at least with automatic tracking algorithms.  Those of us attending the summit concluded that the absence of ball-tracking data probably means that it is still under development.  On the other hand, the TrackMan Doppler radar tracking system was invented for this job.  TM was installed in two MLB venues last year.  A bunch more have signed on for this season.  Those data are owned by the teams that have have paid to have the device installed, not by MLBAM.  So it is unlikely any of the data will ever see the light of day.

MLBAM’s original interest in Pitch F/X was to provide real-time pitch tracking data in the Gameday product - that is, for entertainment purposes. Sportvision, meanwhile, is largely in the business of providing data for broadcast use - again, entertainment. (Their most famous product is the first-down line on football games.) The analytic applications of the data only became apparent to everyone after the initial release of Pitch F/X.

Public release of the data seems to be most likely to occur if:

* The data can be processed in very close to real time, and

* It can be used to enhance the viewing experience of the fan (in other words, it can augment a product like Gameday or be used in a live broadcast like K-Zone, another use of Pitch F/X data).

That’s how you can make revenue off of data while still making enough money to recoup the (substantial) development and installation and maintenance costs. If you can’t subsidize the data through entertainment applications, then your primary revenue stream is going to be analytics, and Tippett is very correct in noting that the value proposition of the data is greater when fewer people have access to it.

Colin, I think you nailed it.

Nice thread. Mentioned nowhere in the linked columns that I could see was anything about the MLBPA accepting the use of field fx. There had been some talk some time ago about how use of field fx to prove that a player was slowing down wouldn’t be accepted by the MLBPA. The precedent was a similar system for European soccer players (UEFA, maybe), who rejected use of that system. I’d like to see the owners sell that feature of field fx to the heirs of Miller and Fehr.

What?

You don’t need Field F/x to see if a player is slowing down.  We already have technology for that.  It’s called a stopwatch.

Pitch f/x can tell you if a pitcher’s arm is slowing down, from season to season, or even inning to inning.  Did the player’s assoc have any objections to that?

Alexi,

I think the radar gun precedent might make that argument hard to win - for many years it’s been possible to track a pitcher’s velocity, and its trend over time.  I’m not sure how one would go about logically excluding one but not the other…

The soccer one must be different.  It’s not like a soccer player has many particular constant occasions to run full out.  A runner attempting to steal a base is such an occasion.  Perhaps running times going 1B to 3B, or running times on stand up doubles, etc, are not it, and I agree.

Any event where a player is not required to run anything close to all-out is going to be noise.  Outfield putouts are going to have tons of noise, and so, the soccer-objection would come into play here.

It’d be like tracking the fastball speed of a pitcher throwing an intentional ball.  That’s noise.