FanPost

Scouting Director Randy Flores, video, and the non-outcome-based statistical era

The market for conventional statistics is saturated. All 30 teams have statisticians on staff, and collect stats for all leagues. Remember, the Cardinals famously used Junior College statistical data to find Matt Adams, but that was back in 2009. So one can assume that other teams have now emulated the Cardinals' success, and the race for the next statistical frontier has advanced quite a bit in the last 6 years. Conventional stats presumably cover almost *every* amateur game at this point. So it's probably almost impossible to find an edge over the other analytically-minded teams via statistics alone. The hiring of Flores is evidence that the Cardinals plan to pioneer a new frontier.

This new frontier, presumably, is information-dense scouting video. Imagine you had comprehensive footage of every player-of-interest (say, by hiring 500 new, part-time roving video scouts, who don't need much more than an iPhone, basic training on how to position the camera, and an internet connection to upload their video). That would enable your best scouts/analysts to do 10x the work with 1/10th the travel. That would give your statisticians access to reams of proprietary information. Your video analysis would provide an enormous edge over every team that is reliant on open, outcome-based statistics.

The low-level "smurfs" (to borrow a term from Breaking Bad) collecting the footage could be easily-replacable, $10 an hour high school/college kids with minimal training and no baseball background. Further, you could hire interns to analyze the video and catalogue the number of hard-hit balls, rather than relying on "hits", and o/z swing rates, rather than walks.

Theoretically, you could collect video of every at-bat/every pitch from every player under consideration. By accumulating that data, make decisions that *transcend* statistical outcomes. Instead, you could observe player *processes*, and do so at a price that is dwarfed by the value of even a few draft successes.

Let's use a little back-of-the envelope math to ballpark it. Assume that each player is engaged in about 1000 hours of game play per season, and that the Cardinals are considering 500 players per draft. That means we'd need to collect and organize 500,000 hours of footage per season. That's 312.5 Terrabytes of data (at 5 Mb per second). Very achievable, using cloud storage systems. And it would cost roughly $5 million per season to collect the entire season's worth of footage on these 500 players (at $10 per hour).

Let's assume the intern who edits/analyzes the video is slightly better paid - $15 an hour - but that he only needs to watch an average of 30 minutes of each game to get the info he needs, and distill the video into a usable form. That's another $1.25 million.

At this point, you have detailed stats on all 500 players, which use advanced, non-outcome-based metrics, and cost you $6.25 million. If this system yields one additional WAR per season of draft success, it's worth it. If that's what Flores aims to do, and if he has the brains and leadership to make it happen, I think this could be an excellent hire, and provide the Cardinals with a durable competitive advantage over the next decade.