They have only played 2 years and missed the entire off season programs the rookie years. By the end of the rookie deals its quite possible Prince will be best CB we have had since Webster if not better. Jacqauin Williams could be a starting WLB.Brewer could be the starting RT and Scott could be a contributing RB. To early to tell. Austin and Jernigan though do not look promising. not really surprised Greg jones wasnt any good and Sash i never thought was more then Special teams.
Originally Posted by TCHOF
Its possible Cooper taylor is similar to Jacquain WIlliams that he is rated lower because of lack of familiarity. When I went back and watched Williams tape I thought he looked fantastic.
Last edited by Redeyejedi; 04-28-2013 at 10:57 AM.
Slip, I wrote a whole lengthy reply describing why you analysis was meaningless, then realized that I was writing it for the wrong audience. Anyway, you can't assign a value to a subjective assessment or rating and add them subtract multiply or divide them. There are many other things wrong with this analysis, all of which means that you really need to find a better way to look at the data.
yea I know what you mean.
Originally Posted by Captain Chaos
Was just trying to compare pre draft data to whether the giants were reaching or having top guys fall to them.
Of course it means nothing for the actual players, and it's only one source.
Who cares, we got him and he's going to ball! That was an awesome pick.
Originally Posted by TheEnigma
It's the wrong audience because this thread, and method of analysis, is less about trying to grade a draft than it is to grade the fan's perception of our draft. Hence, the pick value addition.
Originally Posted by Captain Chaos
Without the pick value addition, drafts would be inflated and influenced by the online prospect rankers inability to predict the lower ranks. The draft in general is a crapshoot.
Good point. A little early to judge the 2011 draft
Originally Posted by Redeyejedi
The thing I would point out about this, is that as I'm sure you already know, the draft value trade chart changes every year. There is every reason to believe that the picks have a much lower standard of deviation from the mean (a fancy way of saying the lower and higher picks are much farther apart in value) this year. just look at the raiders trade with the dolphins (although I realize the raiders were rather desperate to trade out of the pick). The same is true of all of the early picks I believe if you compare the values from the trade chart that is available online to the actual trades were on draft day.
Originally Posted by slipknottin
So basically, at least for this year, the higher picks had less value than normal, and if you wanted to determine a more realistic number for the values in prior years' drafts, you would have to analyze all of the trades that happened, and then come up with all new numbers that roughly fit that data. Obviously that would be a lot of work and quite difficult to do though.
I also think that rounds 4-7 are all really had to figure out, since often, especially with the giants lately, a team takes a guy that's not on most teams radar, like Adrien Robinson or Taylor his year, who the giants FO think are hidden gems.
Lastly I think there is actually a bit of a disconnect between value of a player that is picked versus the value of the actual pick itself that is determined by the draft pick trade chart. Teams determine the value of the picks themselves based on the fact that you are more likely to draft a player who will be good/great for your team with a higher pick. However, if you draft a player who has a second round grade in the third round (for example Moore from this year based on the data you provided), shouldn't you multiply his value by the value of the pick he actually was graded at as opposed to picked at? To show what I mean, I'll give a good example. Imagine Eric Fisher had the number one grade in the draft. If we drafted him in the seventh round, the second last pick in the draft, why should you multiply that number by the value of the second to last pick? That is treating Fisher as a late seventh-rounder, which is clearly wacked.
Basically I guess I'm just pointing out that it's very difficult if not near impossible to evaluate a draft just based on round-grades given by scouts relative to where the players were actually drafted. You'd actually probably be better off doing it the way you did it the first time.
Maybe -- sticking with the draftscout rankings -- you take the player value (pick - ranking) and the multiplying factor could be his ranking in pick value - his selections pick value.
Originally Posted by ObsessedNYGFan
A little confusing, but I'll elaborate. You have CBS Sports' top prospect, Luke Joeckel. His player value would be -- in the seventh round -- 252 (253prospect-1prospect) and his multiplying factor would be 2999 (3000pts-1pt). 252*2999=big number
That's a really extreme scenario, but you get the point.
Yeah, that's basically what I was implying, and maybe that works ok (Fisher in 7th round obviously should generate an insane number). The only thing I don't like is that we don't have the real trade chart, as I mentioned (that oakland trade was an outlier, but funny enough I think it was still a good trade for both teams, which I think goes to the point of how off the chart an be in some cases, but I digress a bit).
Originally Posted by gmen0820
Here are some more examples to see where this gets us:
Fisher at 19 = (19 - 1) * (3000 -875) = 38,250
#10 ranked at 19 = (19 - 10) * (1300 - 875) = 3,825
#19 ranked at 19 = 0
#28 ranked at 19 = (19 - 28) * (660 - 875) = 1,935 (two negatives make a positive)
The problem we run into here is that the last example wants to express that the pick is bad because it is two early for the pick and also because you picked poorly on a very important pick, but with this expression is unable to because the negatives cancel. Also using absolute value does not fix the problem.
Instead perhaps we can try making the ranking a fraction when negative, ie. -10 = 1/10, and the point value as a fraction:
Fisher at 19 = (19 - 1) * (3000 / 875) = 61.7
#10 ranked at 19 = (19 - 10) * (1300 / 875) = 13.37
#19 ranked at 19 = 1
#28 ranked at 19 = (1 / 9) * (660 / 875) = .083
#224 (late 7th rounder) at 19 = (1 / 224) * (2 / 875) = .000102
So in the most extreme examples given (Fisher at 19 and 7th rounder at 19), the difference in value of the pick is roughly 650,000 times better.
#10 ranked at 19 being 13 times better than average value seems a bit extreme to me though, so maybe just view this as more of a point system. However it does at least seem to weigh hits and misses fairly equally.
Let's see if it punishes less for misses and rewards less for hits in late rounds: (obviously if you pick exact value you still get a 1 however)
#42 at 51 = (51 - 42) * (480 / 390) = 11.08
#60 at 51 = (1 / 9) * (300 / 390) = .0854
#74 at 83 = (83 - 74) * (220 / 175) = 11.3
#92 at 83 = (1 / 9) * (132 / 175) = .083
As you can see unfortunately it doesn't reflect that, it actually treats the rounds about the same as long as the picks are equal distance apart.
It might just be best to not divide by the value of the actual pick. Then you get:
#10 ranked at 19 = (19 - 10) * 1300 = 11,700
#28 ranked at 19 = (1 / 9) * 660 = 73
#42 at 51 = (51 - 42) * 480 = 4,320
#60 at 51 = (1 / 9) * 300 = 33.3
#74 at 83 = (83 - 74) * 220 = 1,980
#92 at 83 = (1 / 9) * 132 = 14.7
Doing this reflects the fact that if you want a high point total for your total draft (essentially your draft grade), the high picks are very important. I think doing it this way is halfway decent.