9:14 am Pacific: Well, I'm here in San Francisco for another Sportvision Pitch F/X Summit. The name continues to become less and less descriptive over time – last year they rolled out Hit F/X and this year they're introducing Field F/X. Throughout the day, I'll be making observations here at BP and on Twitter – follow me at @cwyers or look for the #fx2010 hash-tag. Sportvision is also graciously providing a live, streaming webcast of the event. –CW

9:21 am Pacific: John Walsh kicks off the discussion looking at how to use Field F/X to measure infield defense. The rest of the Field F/X presentations are supposed to go this afternoon, but John is joining us remotely from Italy so he's being snuck in early to accommodate him. Complete nerd moment – I think I recognize the software he's using to make these graphs. (GNU R, for those wondering.)

9:28 am Pacific: And we get to the part where we talk about distinguishing grounders and line drives for infielders. We don't have full trajectory yet for these batted balls (I believe the next presentation will cover that) so there's some margin of error for this. For those who haven't been following this – the Hit F/X data collection system gives us the launch angle (both vertically and horizontal) and the velocity of the ball; it doesn't tell us how the ball is spinning. That affects two things – the "carry" of the ball (a ball with backspin will go farther than one without) and landing spot (typically batted balls will curve toward the foul line closer to where they're being hit). This is a problem for evaluating defense, obviously (although perhaps not as big a problem as where we're at currently, with stringer data). What's being introduced this year is fielder (and baserunner) positioning data, which opens up the availability to do sorts of fielder analysis we haven't been able to do before. But the overall accuracy of the metric is determined by how good our batted ball trajectory data is.

9:28 am Pacific: Lot of really cool stuff here about tracking the first step of fielders. This is the sort of data that is going to be able to give us the ability to break down fielding into components (range, the ability to choose routes, hands, arm, etc.) instead of just looking at fielding as one component.

9:50 am Pacific: Now we're moving to Pitch F/X data – Matt Lentzner talking about how to evaluate the movement of the pitch from the proper perspective – that of the batter. Some incredibly cool illustrations on these slides (if you go to the link at the top of the post you should be able to download these slides yourself).

10:16 am Pacific: Really good followup to Matt's presentation – his copresenter, Mike Fast, showed us some very nice graphs looking at whiff rates on swings, based upon pitch movement. That's not the whole picture, though – I can't tell who asked, but someone asked about looking at what induces a batter to swing, and if some movement "freezes up" the batter. Now we're hearing from Glenn Schoenhals, who runs Scientific Baseball, which uses a private Pitch F/X installation for coaching. Cool stuff.

11:01 am Pacific: After a short break (this will surprise nobody – I spent most of it talking about bias in collecting batted ball data from stringers) we're back, and Matt Lentzner is presenting about Hideki Okajima's rainbow curve. Finally got a chance to meet Jeremy Greenhouse, who confirms this live blog has at least one reader. For those following along at home, Matt's referencing last year's presentation about the "pitching peanut," and how to identify pitches based upon it. You can find it (and all of last year's presentations) at Sportvision's website.

11:12 am Pacific: This is the main attraction, folks – Professor Alan Nathan talking about the spin of the batted ball. Given the data currently being collected, this is essentially the holy grail of deriving the full trajectory of the batted ball. Given my obsession with measuring batted balls, you can guess I'm rather excited to hear what he has to say. Last year he presented on the spin, and there wasn't much good news. I'm hoping this years' presentation is more encouraging in being able to derive spin from what we're currently measuring.

11:22 am Pacific: I am not disappointed – getting to see 1000 frame per second videos of spin coming off the "bat" in laboratory conditions. Seeing how the offset of the bat and ball affects spin is really, really cool. "The spin of a batted ball does not depend in any crucial way on the spin of the pitched ball – and that's a new piece of information." Wow.

12:14 pm Pacific: Afternoon session starts with a presentation on how Field F/X data is collected. One thing I want to discuss – someone asked on Twitter if the findings of batted balls (that incoming pitch spin doesn't significantly impact outgoing spin) means we know less about the spin of balls we know before. And I think the answer is no, we know more. Knowing what we don't know is I think important, and it's vastly preferable to knowing things that aren't true.

12:30 pm Pacific: Dan Turkenkopf just asked on Twitter if the positioning of the cameras for Field F/X should give us bias problems, like I've been discussing with human stringers. The short answer is no. The reason is because the data is being generated in a systemic way, given known distances to various points on the field. Assuming the camera's position is properly known (this is an engineering challenge – it's hard to picture it, but the weight of 40,000 baseball fans actually distorts the shape of the stadium, thus displacing the cameras) you can do the reckoning with a pretty high amount of accuracy. (I'm not communicating the full extent of how hard it is to do this well – this is a lot of impressive engineering by Sportvision. My point is that its a solvable problem, not that it's an easy one.)

And it is possible to do very accurate ball tracking with the current cameras views we have, it's just very labor intensive. Hit Tracker Online is doing it for home runs – if we had Greg Rybarczyk doing that process for every batted ball we'd have an astonishingly good data set. But we regrettably only have one Greg, and instead of asking people to measure what we're doing is asking people to make judgement calls. This isn't to say there aren't any concerns about park effects on data collection, but it's a very different beast than park biases in stringer data.