First off, let me give a giant shout out to Mr. Taylor over at BleacherNation.com. There is a less than 0% chance I would have found this unless he did. If you frequent this site, I am sure you frequent there where I am a frequent poster during the day, and I must say, he was the inspiration for me starting this site several years ago.Anyway, thanks Ace!
According to the CEO of Cray (a supercomputer manufacturer), and as reported by Peter Ungaro over at The Economist, an unnamed team has purchased one of their computers. Furthermore, it is a team that 5 years ago, would have never dreamed of such a purchase. That rules out most teams. Either because they are strident sabermetricians or 5 years ago they weren''t hilariously behind the rest of MLB.
Who does it leave? The Phillies, Diamondbacks, White Sox, Brewers, Dodgers and Cubs (rough thinking here). The White Sox and Brewers have, roughly, the same management as 5 years ago. The Diamondbacks would qualify if he said last week. The Dodgers and Cubs have recently gone under some serious changes at the top in the last 5 years (Cubs more than the Dodgers) and the Phillies have been, what appears to be, trying to undertake some shifting in how they operate. Frankly, the Cubs and Phillies make the most sense but the Cubs have been spending a lot of money on this sort of thing in the past few years. So we will just say they got their own version of Hal.
Neat-o! Now the Ricketts can say they have done something without really doing something, right? Well, let's think about what has been happening the past few years in baseball. Information overload. In the linked article above, and I wasn't able to independently verify this, 95% of all data currently available for teams, has been generated within the past 5 years. That's 95% of the data in 3.5% of the time. Now, extrapolate that over the next 5 years or the 5 years after that. Teams that don't take advantage of every morsel of data that is available are going to be quickly left behind and it seems that computing power may be a market inefficiency that the Cubs might be looking to make a stranglehold on.
So, what could you do with all of this computing power? Well, just given what is available in today's world, you would be able to breakout every pitcher into subcategories and, during a game, utilize the computer to crunch a pinch hitting decision. If you know that Rizzo and Schierholtz are on the bench and that Justin Verlander is on the mound you can break out what type of hitter both Rizzo and Schierholtz are, what kind of pitcher Verlander is and then look up what players like Rizzo and Schierholtz have done against pitchers like Verlander, run a simulation a few thousand times given the current weather conditions and how Verlander is pitching that day and know that Schierholtz is going to provide you a run expectancy that is .00000001 higher than Rizzo.
In addition, with advanced scouting, you could determine that if you put in Rizzo or Schierholtz the opposing manager is likely to sub Verlander for Pitcher X based on that manager's history, current bullpen usage and game situation. In this case, it would spit out that Rizzo should produce .001 more runs than Schierholtz. Lineups could be optimized that way as well, a long with bullpen usage, defensive positioning and grass height. If you know that teams that profile similar to your next series would score fewer runs with shorter grass, by God, cut the grass a little shorter. Maybe the other team fares poorly with a dry infield or the alleys need to be left longer than straight-away.
It could spit that out instantaneously, while letting the FO review trade candidates or amateurs for the draft. Imagine that you could plug in a HS hitters stats and review the players that have come out of a similar league with a similar profile and project future stats/development. Now imagine that same scenario, just with a AA ballplayer and it's the Trade Deadline. Getting the information on a prospect minutes earlier allows more time for debate/discussion and further negotiating.
That's just a brief touch on what can be done with the data available today. What's going to happen when the Field F/X data gets loaded into the system. Now, not only can you load in stats for the hitters, you can enter the information about how the ball leaves their bat which can be used to create more profiles of hitters and how those profiles stack up against certain profiles of pitchers. The 3rd base coach now has more information on when to send the runner. How players perform in certain weather conditions.
Also, I would imagine the Cubs would want this kind of information in their minor league system. Place Pitch F/X and Field F/X in each of the minor league stops and suddenly you have some advanced scouting against rookies and more information that you can analyze when working trades or callups.
It's not just about game performance or trying to determine who the best player should be going forward. It's also about looking for trends within the data. A supercomputer could look at all of the data from 2013, and in the matter of minutes, see if there is a common denominator for why certain players performed better/worse that hasn't been identified. Bill James spent the better part of 30 years creating the various Sabermetric stats we use today. Now, take a box filled with millions of Bill James and more data than he ever could have dreamed of and put it to work. Just spend 10 minutes watching this video from last year's SABR conference and imagine what sort of work could be output in those same 10 minutes if the computing power equivelant to the SABR conference were brought to bear.
Ultimately, we are, probably, talking about the difference of a win over the course of a season that something like this could bring to bear assuming no major breakthroughs. If it leads to another revolution in sports, being the first team to the block could pay huge dividends. All for somewhere south of $5M plus operating costs.
Dear Korean Jesus,
Please let this be the Cubs that bought the supercomputer.
(warning, strong lanugage in this video)
No comments :
Post a Comment