July 24, 2008

Cheating Eight Year Old: Player banned from an Under 10 tournament for 'receiving instructions' from her father through a headset.

posted by owlhouse to tennis at 06:50 PM - 16 comments

This time Belichick has REALLY crossed the line!

posted by Venicemenace at 07:15 PM on July 24, 2008

Okay, I'm a Patriots/Boston fan, and even I laughed Venicemenace. :) That seems questionable as "cheating", although we don't know what was said- christ, these kids are under-10, at that age isn't it more about fun? Personally, I think all sports should make use of radio technology. I think it'd be awesome if every player in football- not just the quarterback- had a radio in his helmet, and the guys in the booth could play 'em like Tecmo Bowl, Madden-ing a play diagram and having computerized voices dictate the customized play and key points in 5 second blurbs to each player ("Attempt to push through left of center" or "Stay on the WR but only for 15 yards" or "Forward 10 yards, break left 5 yards, turn and await pass"). I think it'd be awesome in baseball, too: guy at the plate has a little voice in his ear saying "computing... 84% chance of slider on the next pitch. Given current count, outs, runners, and game context, as well as batter proclivity with sliders, it is recommend you take this pitch and sit on outside fastball on next pitch" Or the pitcher has a little voice say "throw to first!" when some laser-guided multi-camera unit has determined that the runner on first has, with his known, median, and averaged speed and lead, is just about to shift his weight in the wrong direction to take that one step too far and the computer predicts it and warns the pitcher just in time. Christ, I don't even know if I'm being sarcastic anymore.

posted by hincandenza at 08:45 PM on July 24, 2008

If the players are supposed to keep sore themselves, then I believe they should do it. It seems a great way to learn the game, which, I think, is pretty important:) I do feel bad for the young girl. Getting kicked out of your first tournament must be the pits. I guess she can always blame her dad.

posted by BoKnows at 01:12 AM on July 25, 2008

Oh for gods sake... I fail to see how this is cheating, unless I seriously underestimate the power whispering "hit the ball harder" into an eight year olds ear... *runs off to find an eight year ol... Actually on seconds thoughts, maybe not...*

posted by Drood at 02:05 AM on July 25, 2008

Anything that gives one player an unfair advantage over another is cheating. Period. Why are people so willing to bend the rules when they have been so obviously broken? Although the father said he was just helping her keep score, that was a bunch of baloney. A 5 year old could keep score in tennis. More than likely he was telling her where to hit the ball which is cheating. Its too bad the child was penalized, but it was a good and necessary lesson learned: don't cheat and if you do and you get caught, you will get penalized. In Pro Tennis, the coaches in the stands are not even supposed to make any kinds of signals to their player or the player could be penalized.

posted by daveny5 at 09:50 AM on July 25, 2008

Does anybody know what a "heightened degree of questioned calls" means? Apparently, that is what tipped the officials off, but I don't even know what it means.

posted by graymatters at 10:01 AM on July 25, 2008

I guess rules is rules, but the father should have been penalized instead of the child. It seems as if he was the grandmaster of the plan anyways.

posted by lsutigers96 at 10:28 AM on July 25, 2008

It's called cheating until it's called a rule change. Like the forward pass in football. The more I read Hal's ideas, sarcastic or not, the more I liked them. Perhaps one man's cheating is another man's progress. Certainly many sports are beholden to their traditions; that's what provides much of their charm, such as baseball. But take for example a sport like auto racing. New technology and progress is taking it in new and more exciting directions, while providing a greater degree of safety for the participants. We probably wouldn't be as interested in it if the races were still at blistering 25 mph speeds. If the idea of a linebacker being played like a video game is unappealing, what if it the player and coach could interact and share info? My first thought would be that it would elevate the game, limit mistakes, and actually shine more light on good coaching instead of luck or freakish athletic ability, which would always be there. As for the gal and her dad, it was dumb to punish the girl. She's eight years old, fer chrissakes. She may not even have known it was against the rules. But you can bet dad did.

posted by THX-1138 at 01:12 PM on July 25, 2008

BoKnows: If the players are supposed to keep sore themselves
Drood: runs off to find an eight year old
Well, it sounds like Drood is going to be in charge of keeping them sore. Hey thanks folks, you've been a great crowd, and I'll be here all week! Be sure to try your waitress, and remember to tip the veal!

posted by hincandenza at 05:14 PM on July 25, 2008

The more I read Hal's ideas, sarcastic or not, the more I liked them. Perhaps one man's cheating is another man's progress.
Well, while the radio in the helmet might be a bit extreme, is there a rule against computers in the dugout? For example, I've long wondered why a canny MLB team doesn't kick a few hundred grand towards a "research project"; buy a couple hundred or so cheap used PCs and put them in a linux-based cluster using free open-source search-engine technologies such as Hadoop. 300 cheaper PCs would run about $150,000 in capital costs, about the same for 3 year's of datacenter costs (space, power, cooling, etc), plus money for the handful of people to maintain the system and manage the software and algorithms. All told, it'd probably be about league minimum ($390,000) to run this system each year. Next, load all the statistical data that's available, including and especially the PitchFX data from the last couple of seasons. Stat-headed fans, just for the love of the game, are already doing amazing work in graphing and analyzing pitches and pitch counts by location, type, break, sequence. A well-funded group working for, say, the Seattle Mariners, could be using this large, relatively inexpensive farm of machines to constantly run (and re-run with every pitch) lots of parallel analysis about pitcher's behavior. These would include but by no means be limited to questions such as:
  • How does temperature, humidity, pitch count, and game time affect the location and break of pitches?
  • For every given count, what is the break-down of next pitches for a given pitcher?
  • The previous question, but broken down by situation, particular batters, handedness of batters, etc?
  • Analyzing the habits and patterns of the hitter at the plate and his best and worst pitches
The beauty is a large cluster of cheap PCs, with all those parallel CPUs, could do massive parallel analysis of hundreds of different views and permutations of the data at the same time. Thus, math savants could come up with countless additional Bayes-ian views of the data, or even idle questions, and begin to make a computerized baseball mind that was insanely good at predicting the next pitch or pitches in a sequence. This system would be refined over time, using past success to modify the weightings of different analysis. Any given realization about tiny advantages in working pitch counts to get certain pitches might be miniscule, but in sum those realizations are the reason Tony Gwynn won so many batting titles. The truth is, the human mind thinks it's clever, but often has learned patterns it falls into. With sufficient reams of data, we could divine these patterns, and determine that there is an "84% chance of a slider on the next pitch". If such a machine was running, a small PDA in the dugout would allow a coach to access this data in real-time and signal to the hitter some shorthand version of "Look for the slider on the next pitch, but don't swing- you can afford the strike." If the next pitch was a slider as predicted, the system would respond, through the hitting coach, "Sit on the high fastball, which you eat for breakfast..." Hell, even if it wasn't during the game, it would be great prep-work before hand: the nice color-printed analysis of the starting pitcher, how he trends, and how each hitter individually should approach that pitcher to work the count and get the most favorable pitches for them. Hitters perform differently and have different habits and trends themselves, and you want to account for that. It basically would allow a team to give every hitter, no matter how Fasano-esque they might be, the mind of Tony Gwynn or Ted Williams while they are at the plate. Musings... Let's say such a technology turn 3% of outs into hits/walks. I'm totally making that figure up, but it's basically saying that 1 time in 33, what would have been an out in an at-bat becomes a walk or a hit by being massively "smarter" as a hitter. The thumbnail "average OBP" in baseball is about .338, give or take; the current team leading the pack is Boston with .356. Anyone watching the recent Boston-Seattle series noticed how much that Boston patience hurt Seattle, when even great pitchers like Felix Hernandez found themselves walking middling hitters like Coco Crisp to force home a run. Seattle has as a team 3,840 plate appearances with a majors-worst OBP of .312, so they've made 2,642 outs as a team. That 3% outs- into- hits would be only 80 extra hits so far this year, less than one extra hit or walk per game for the entire team. That seems so little, yet it would be enough to raise that majors-worst on-base percentage up from .312 to .332, which is just below major league average and tied with Philadelphia at 14th overall. And if those outs became hits- even just singles- average and slugging would by that nature also go up. Scoring runs would naturally go up as well. This is purely a speculation as to how much effect being Tony Gwynn/Ted Williams brained would have an entire club, but using my "1 time in 33 it leads to a hit instead of an out" doesn't seem wholly impausible. Using that, here is Seattle's line before and after the implementation of such technology for a cost of maybe one league minimum player: AVG___OBP__SLG__OPS .255__.312__.371__.683 .278__.332__.394__.726 As a team, Seattle's .683 OPS is the second-worst in the majors, just barely ahead of Washington's .681. Our fictitious .726 OPS would put Seattle at 20th, right along side Houston and Cleveland. How does that translate into runs? Well, there's not a perfect OPS- to- runs calculator, but OPS does have a decent correlation. Using Dan Fox's work here at Hardball Times we can estimate that an OPS of .726 correlate to about 699 runs, +/- 25 runs. Houston, which actually has an OPS of .726, is on pace to score 693 runs. Seattle, on pace to score 634 runs, would be expected to put up 699 runs instead, or 436 runs through their first 101 games instead of 395. Conclusion: Using the premise that if every hitter had the insight into pitcher patterns that a Tony Gwynn or Ted Williams have, then at least 1 time in 33 they'd get a single where previously they made an out, we can estimate the following improvement. A team that scores 395 runs and allows 468 should have a .416 winning percentage for a 42-59 record; the Mariners are at 38-63, so they're underplaying their own stats. A team that scores 436 runs and allows that same 468 runs should have a winning percentage of .465, or 47-54. Using heavy computerized analysis and real-time feedback for your hitters might potentially lead to scoring more runs, enough that the 2008 Seattle Mariners, on pace for 67 wins, would finish with 75 wins instead. That's an 8-game difference, and using the notion of Bill James' "win shares" a player who leads to a 10-win difference is having an MVP-caliber season. Don't even get me started on how this same technology would allow profiles of opposing hitters to improve your pitcher's chances... Am I that outlandish to suggest that being able to have the greatest virtual hitting coach in the world for the league minimum turn in an MVP-caliber difference for your team is maybe... worth investigating?

posted by hincandenza at 06:56 PM on July 25, 2008

Hal, I seem to remember Davey Johnson in the Baltimore dugout with a laptop. He was noted for digging through statistics, looking for an edge in matchups. Unfortunately, his statistical prowess didn't translate into wins. He was also working in an era before many of the more esoteric number crunching tricks had been thought up.

posted by Howard_T at 07:57 PM on July 25, 2008

Right- but I think with Johnson, it was more about things like the charts of run expectancy for different situations to guide when it's a good time to bunt or not, or how different hitters match up for doing PH moves. As has been noted, even a good manager can squeak maybe 2-3 wins out of his team making amazing moves throughout the entire season. What Johnson was doing is now standard: every manager, except maybe Seattle's John McLaren before he was fired, has basic charts on these kinds of expectancies on stealing, hit-and-run, etc. And even in my example, it only took a near-historically bad team into one that's still under .500; no one would notice "success" if it was only 67 wins to 75 wins. But "success" that goes from 81 wins to 89 wins can be the difference maker. I'm talking about something orders of magnitude more informative than pre-generated expectancy charts: looking at information on the physical paths of every pitch, and looking for edges and trends that people would never even think about in the past, such as how 10 degrees of temperature affects the break of a curveball, or the flight of the baseball for a high FB/GB ratio pitcher. Between math savants, statisticians, Bayes analysis, "conventional wisdom" from actual players, and idle thinking, a team dedicated to this with the resources available could find edges that, if played out in thousands of at-bats, only amounted to maybe one hit a season- but if you add them all up, you might add a few percentage points to your team's OPS, which as I've noted above can translate into real wins. Analysis into whether certain pitchers- whether they realize it or not- have with their catchers a tendency to go to certain pitch sequences or "safe" pitches. About how your hitters handle certain pitches- given a player's swing, if the game time is 85 degrees or more, he should go the other way on sliders but under 85 degrees he should pull a slider. There's a wealth of information we are barely starting to look at these days, but if you could run simulations and analysis on what's a ginormous database of pitches and results and their context, you might find some fascinating- and incredibly useful- information. After all, if you have three pitches in your arsenal, it's like playing Rochambeau (rock, paper, scissors). And if you played hundreds of games of Rochambeau, I with my super computer would possibly find trends and patterns in what you were doing, such that I could get a 3% edge on guessing your next move. If I just played random hands to your moves, I'd win 1/3 of the time, lose 1/3 of the time, and tie 1/3 of the time. If however I always responded to certain sequences that even you weren't aware you had with a specific move and it won 50% of the time and lost only 17% of the time... Sure, it doesn't sound like much, but the difference between losing at blackjack and winning is about 1%; that's what card counting buys you, the difference between 49.5% and 50.5% as a player. And the difference between struggling for a roster spot and a plaque in the Hall of Fame is 5 percentage points.

posted by hincandenza at 09:11 PM on July 25, 2008

A short puzzle I know I'm generating a lot of words here, but let me offer this story. When I was working at Microsoft a few years ago, when we still did the famous brainteaser questions during interviews, I suggested one that I'd read online. The puzzle is this: you're a terrible darts player, no real aim but good enough to hit the board on each throw. You're given two boxes of darts, one box of a few hundred red darts and another of a few hundred green darts. The dartboard itself is broken up into 20 wedges, some colored green and some colored red- in fact, there are 14 green wedges and only 6 red wedges. You are told that you can throw 100 darts, and every time the dart you throw matches up with the color of the wedge, you earn a dollar. How do you choose to throw your darts? The correct answer is "You throw all green darts". You might be tempted to say "for every 14 green darts I throw 6 red darts", or a 70/30 split. That way, surely you'll make the full $100! But in fact, since your aim is so poor, if you choose that tactic you'll only get 70*.7 green darts and 30*.3 red darts, or 49+9=58 darts. If you throw 100 green darts, you'll hit green wedges 70 times. When I told this puzzle to a colleague, he thought it was ridiculous- the puzzle was so easy, he said, that no one could possibly get it wrong. I had to tell him the bad solution for him to even realize that someone might think to go 70/30 on the darts. Yet he later used that in an interview, and was surprised when people did get it wrong- or spend a long time puzzling it out. It's less a puzzle than a test of whether people will overthink or overcomplicate a problem. I tell this story as an example of how sometimes the right choice is to pick the best strategy, and ride it out. No one gets on base 100% of the time; but if you have the best possible strategy and play it out, you'll be a little more successful. In the case of the darts, the strategy is simple. But in baseball, with so many pitches and situations, the "correct" strategy might be so detailed that we humans couldn't conceive of it. It'd be like the shape of the darts and the ratio of red to green on the dartboard changed with every throw. But if we could suss out the optimal strategy... maybe we'd be a little more successful in our at-bats.

Let me append this by noting a tiny smattering of work being done already, by unfunded amateurs. Here is an archive on Hardball Times of articles using Pitchf/x data from around the web. For example, the first one listed is this article from Cubs f/x about Tim Lincecum, written a couple of weeks ago. It has, using Pitchf/x, detailed information on the movement and velocity of Lincecum's pitches, the break downs of when he uses which pitches, in what counts, against which type of hitter. This is the kind of information, both historical, rolling average, and real-time in that particular game, that hitters have for ages done "in their heads"... with far less accuracy and reliability. John Walsh at Hardball Times has done yeoman like work on this, in particular this article about the effectiveness of pitches and pitch sequences. It's this kind of work in particular that I'm thinking of as beneficial to hitters. And John Walsh is just one guy, who has to come up with the idea, then massage the raw data to produce charts and perform analysis. What I'm describing isn't one fan in his office working these numbers, it's a small dedicated team of guys like John Walsh coming up with 100 of these theories and ideas and having them all performed constantly on a farm of cheap PCs running in parallel, where they can come up with a notion and type it in as easily as conducting a Google search, with results in the same sub-second time. The construction time of these ideas in such an "idea factory" would be measured in seconds or minutes, and once added could be run perpetually as new data came in. Eventually you'd have hundreds of algorithms crunching away all the time- with every pitch- and weighting the innumerable percentages of different views to predict the likely next pitch, etc. The people running this system would have the ability to generate more and more accurate predictions for pitches as well as possible explanations why a pitcher can seem to struggle one start and dominate the next. An obvious example is how knuckleballer Tim Wakefield has trouble in domed stadiums, likely due to the lack of open-air turbulence, humidity and temperature. All pitches, however, are affected by the air; knowing this might help us understand why a pitcher might want to adjust the ratios of his pitch selection depending on the weather. One element of this work is in confirming or disproving "conventional wisdom", such as protection, platooning, etc (some of this is already well proven/disproven). Another element is in basic pre-generated charts such as the Lincecum effort, which I suspect most major league teams have in some form at least by pitch types- it's still a common thing for younger pitchers, and pitchers on their days off, to sit in the bullpen and chart pitches manually. However, Pitchf/x makes that manual work unnecessary. That much is as far as I know already done at the major league level. The real effort and explosion would then be to turning out predictive information on pitch sequences as a group, and thus prepare and update each batter with a clear game plan of what to do in each at-bat beyond "Loves the fastball with a 1-1 count". It could include an understanding of how going the other way on a slider for certain matchups yields a higher batting average, etc. It might include tips on trying to intentionally make an out- I recall a few years ago that one person had studied human eye movement and tracking, and found that hitters will tend to overestimate the height or depth of high and low pitches; by teaching hitters to try to pop-up the high pitches and try to hit the low pitches into the ground, they instead found themselves lasering line drives left and right. And the hitters could glance at the dugout for an up-to-the-second suggestion on what the next pitch will be and what to do with it.

posted by hincandenza at 10:15 PM on July 25, 2008

Holy, Hal. I'll have to read it all tomorrow. It's late. But, if you say something in there about just needing your own facility and then you can conquer the entire world of baseball...I'll, I'll...just turn into a red headed Portuguese that climbs mountains. And if my joke is so inside that only I get it. Sorry. I just need funding for an inside joke facility. The musings part reminds me of a strength of schedule analysis for fantasy football, I just read. I look forward to reading it all.

posted by tselson at 11:41 PM on July 25, 2008

Wouldn't this concept also be used the other way? Somehow signaling the pitcher which pitch the hitter is now expecting, and which different pitch would be most effective in combating that specific hitter's expectation. It seems that historically there have been gains by pitchers / hitters that tilt the advantage on way or the other. Then those gains are negated in a few years by the other group making appropriate adjustments.

posted by opel70 at 10:01 AM on July 28, 2008

Exactly, opel70. But just like "Moneyball" and any other number of strategies- across the sports- the first ones to that form of innovation experience a benefit for a few years. Most teams- excepting the Mariners, apparently- have long since recognized the value of OBS and OPS even though "professional" analysts and baseball people would say "oh-pee-ess? What's that?!" just a few years ago. Billy Beane made a competitor out of a lower to mid market franchise by exploiting that evil sabermetrics to find the best cheap talent, and train them well in the farm system, that produced the most runs while other teams scratched their heads and tried to figure out how he kept getting these productive hitters and pitchers. Now that other teams have caught up, Oakland isn't the perennial contender like they once were. My assumption is that both the pitchers and hitters on one team would benefit- the pitchers becoming smarter, and the hitters as well- and eventually as word spread and success was its greatest evangelist, the other teams would follow suit using their own supercomputer analysis. However, unlike some other strategies such as hit-and-run or situational expected runs, etc, the particular algorithms and predictive techniques would be NDA'ed or otherwise proprietary. It'd eventually be like what we see in the world of high finance, where competing brokerages own competing million-dollar supercomputers attempt to find the best margins in currency exchanges or hedge funds, in millisecond real time. :) With average payrolls approaching $100M a year, these are the kinds of funds where that kind of edge matters. It's the smartest teams that will realize it first.

posted by hincandenza at 09:01 PM on July 29, 2008

You're not logged in. Please log in or register.