mastershake

We can't ignore our OL issues or they'll come back to bite us. Not even just for 2012, but beyond as well. As of the past two years - Beatty has played only a total of 18 games, starting only 12. I believe he only has one year left on his contract as well. Diehl, Baas, Snee, Locklear are all over age 30. Snee and Baas dropped off huge last year. I don't know what Locklear has in him. Diehl is ok, but the oldest in the group, and not a dominant blocker. Booth is a career backup/spot starter. Petrus has looked solid in limited time. Brewer was inactive for most games last season. Mitchell Schwartz, Brandon Brooks, Jeff Allen, Tony Bergstrom. Kelechi Osemele, Zebrie Sanders are all solid 2nd and 3rd round value at picks #63 and #94 respectively.

Well at the end of the day, how can I not trust in Reese. He's done us well for several years. Also, on the plus side, Wilson is not a Big Ten back, so he's not associated with that "curse of the big ten RBs" on the giants (Bunch,Wheatley, Dayne, Woolfork, Adams etc)

I'll be rooting for him for sure. I just hope our O-line can block for him and doesn't get him killed.

Well then I think we're in somewhat of an agreement. If I didn't say it well enough before, I'll say it again - draft grades certainly aren't the only factor to use - I think they're one of many. They can be used to provide some key information on draft order projections, but certainly not all the info, they're not perfect and there are variances (the R2 is 0.59, which means there other important factors).

I did the correlation on espn and nfl's draft grades. I personally use cnnsi, but can't easily get the data into excel. We're spinning our wheels here denying the correlation exists, lol. And nobody is saying this should be the end all, be all of ALL draft decision making lol. Let's just agree that we hope things turn out the best for David Wilson here in NY... And that the giants draft the best graded player available at #63, haha (Ok, now here I'm being sarcastic)

I'm not disagreeing that larger sample sizes, to a point, aren't better. I'll do a full correlation on the entire draft once complete, fair? (it really takes me about 2 minutes, it's a simple copy and 4 clicks away in excel.

Storm, Here's a T-State Table... http://www.google.com/imgres?hl=en&client=firefox-a&hs=lif&rls=org.mozilla:en-US:official&tbm=isch&tbnid=-88BJ4F8NxRNgM:&imgrefurl=http://www.articlesfind.info/pdf/t-stat-table.html&docid=ukrtIdz8VBPehM&imgurl=http://web.math.umt.edu/elias/tTableDistribution.jpg&w=576&h=854&ei=-zqaT-bJNYr30gHa4s3qDg&zoom=1&iact=rc&dur=883&sig=116076501615528704684&page=1&tbnh=180&tbnw=122&start=0&ndsp=11&ved=1t:429,r:0,s:0,i:70&tx=67&ty=61&biw=1092&bih=538 On a sample size of 30, for even a 99.9995% confidence level (or P-Value of 0.0005), the T-State needs to be + or - approx 3.6. The T-Stat in my test above was 5.6, well beyond 3.6. And that's the highest level of confidence the chart has.

The test statistic at 31 degrees of freedom is statistically significant. That's not up to me or you. That's up to how statistical significance is calculated, and probabilities are assigned. Anyone can go look up a t- stat chart and see the same thing. Besides, the P-Value gives you all the information you need. Think of it this way. Imagine you have a hat full of 1000 M&Ms, half red half blue. You know what the probability is to draw 30 M&Ms in a row, all of the same color? It's like 1 in a trillion.

No-one is saying he isn't going to be a good player. Just that there was a high possibility that he'd be available if we traded down.

No it's not. I just calculated the statistical significance based on a sample size of 30, and the probability that the true mean falls outside the bell curve is 0.0004%.

According to you, I'm saying a lot of things I never said nor even implied. Please be sure to use the comments section to demonstrate your own ignorance and unfamiliarity with empirical data, facts, and math. Thank you.

Here's the T-Stat on a sample size of 32 -5.6152 Here's the P-Value 4.1E-06 That means there's a 0.004% chance that this correlation was random. Haha. In other words, that's the probability that what your denial has any validity.

Actually they do. 30 is considered the minimum sample size. It's done all the time. The T-Stat (bell curve associated with this sample size) on the coefficient is statistically significant. You're not going to win going toe to toe with me on the math, I do this on a daily basis.

And they don't try to assign draft grades to order?

Good, ignore it... to your own peril Like I said, it's not the only variable, but it provides a wealth of predictive power and information about draft order projections.

How much do you want to bet Reese and the giants staff are well versed in math and statistics? And are likely using similar types of analysis in their grades, and projections of draft order?

You guys just all said the draft grades have no meaning. I just showed how large of a predictor (among several) they can be in predicting draft order.

Correct, R square takes the correlation coefficient and squares it.

I just calculated the correlation on NFL.com's draft grades... the correlation is approx 0.73. Also very high.

You can do the correlation of player rankings rather than grade, and you'll get the same high correlation to draft order

Actually, I did prove it. It was a tonight's actual draft, and a sample size of 32 observations. You just didn't like the results, or don't understand the relationship. What, you think if I do this with any other site, I'm not also going to get a strong correlation? That's your claim?

Did you listen to a thing I said before that?

I just gave you the data and the equation from the correlation calculation. The R and R2 prove it. I'm a statistician by trade.

You made claims that the draft grades have no meaning, and I just demonstrated they have substantial predictive power in terms of predicting draft order. A correlation of 0.77 is significantly high.