tag:blogger.com,1999:blog-17185944.post8447121647175967970..comments2023-10-30T04:04:00.692-05:00Comments on Process of Illumination: The Diva DilemmaHhhttp://www.blogger.com/profile/01619039249146075142noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-17185944.post-36353009442046294482012-04-10T15:22:49.366-05:002012-04-10T15:22:49.366-05:00I watched several games of Ironside's on video...I watched several games of Ironside's on video this year, and it didn't seem to me that the o-line played conservatively. It seemed rather as if they hucked quite a bit. (Mike Lawler, I think is his name, did some statistical analyses of the ETP games in which he concluded that Ironside hucked too much, if I remember correctly.) If the d-line's strategy was to play conservatively after the turn, I suppose I could understand why, though maybe you're right about it not being the best use of personnel. I also don't think that completion percentages of short passes are anywhere near 99.8% at this level of competition.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17185944.post-60477836802303299462012-04-09T15:51:26.530-05:002012-04-09T15:51:26.530-05:00Muffin, if your super power is your bombs (which I...Muffin, if your super power is your bombs (which I agree that it is), do you believe that part of the issue is that you play on the D-line which means that in games against possession oriented teams (which include Revolver) you get fewer opportunities to use your super power and influence the game?<br /><br />Not that your assessment of the impact of your team's strategy on your influence is wrong, just that your role may also play a part as well.Kyle Weisbrodhttps://www.blogger.com/profile/12968840595720794251noreply@blogger.comtag:blogger.com,1999:blog-17185944.post-10091156383916002762012-04-09T13:59:41.273-05:002012-04-09T13:59:41.273-05:00This is a good assessment Charles, but the issue i...This is a good assessment Charles, but the issue is that people always treat each event independently and do these analyses, let alone do it with static dynamics. The problem we are dealing with is conditioned on events, specifically events with temporal dynamics. Therefore, you will need to make these probabilities conditional. <br /><br />Basically, you are issuing conditional dynamics to the problem by assuming the defense will change its pressure based upon the assumption that the offense is grounded in a no huck game and that they will always take the highest percentage non-huck option. You are issuing a temporal dynamic to the prior on the defense, as an offensive team throws more passes, the defense will adapt their prior knowledge (normal defense) and condition, or adapt, to a more stringent underneath defense. This is accurate, but you can't use the idea of independence in probability anymore and will need to use a Bayesian strategy to analyze this. Also, based upon the idea of temporal dynamics, you are now in the realm of sequential Bayesian. <br /><br />In reality, the offense would adapt to take advantage of some of this tight underneath defense to take high percentage deep looks. This could be taken into account, along with a spectrum of events, and you could make a transition kernal density to account for all these events in a markov chain like procedure. (ie a up line dump leads to a huck more often then not, while a backfield dump leads to a swing more often then not and these have probabilities based upon these conditions) In others words, you have the option to throw any option at any time (dump can throw a swing, huck, under, dump), but with specified probabilities conditioning on the fact you caught a dump.<br /><br />Lastly, looking at your 80% feed the cannon approach, using the same principles of your previous defensive analysis, the defense would condition its prior on a more liberal deep game and likely turn that 80% down to a 65%. I think if you watch enough elite level video and calculate a top hucker's completion percentage, it is roughly around 65% (2/3) at best.Brettnoreply@blogger.comtag:blogger.com,1999:blog-17185944.post-3068453685125795922012-04-09T12:05:34.907-05:002012-04-09T12:05:34.907-05:00The "hi-percentage, short-distance" stuf...The "hi-percentage, short-distance" stuff makes sense on paper, and if your team has good players, it looks like it works...<br /><br />But think about it for a second. What are the odds really like on a short reset throw? If I know the offense is trying the "death by a thousand cuts" strategy, I'm taking silly/crazy chances on every pass to the man I'm covering - if I knock it from a 99 1/2% pass to 90%, how often can I afford to get broken/end up way out of position and chasing?<br /><br />If your offense takes 10 90% passes but gets 10 nearly-free (99.9%) resets because of being out of position, it scores 35% of the time... this is worse than an 80% feed to a cannon for a 50/50 pass to the end zone.Charles Kanebhttps://www.blogger.com/profile/15624079849035204879noreply@blogger.com