Smith has a huge sample of throwing ~17% deep. Standard deviation on Kaepernick's sample of 50 throws at 26.5% is .062, which is not gonna be a statistically significant difference from Smith given the standard frequentist analysis, but given a decent prior due to universal reports of "stronger arm" is almost certainly meaningful. (guessing on the last part but feel pretty confident about the guess.)

This conclusion taken at face value seems insane and wrong. I find it extremely, extremely hard to believe that you actually think those 49 attempts do not affect our estimate of p(kaepernick deep>smith deep) at all. If you do, I could not disagree more strongly.

If I were to say that I expect, for example, Smiff's career completion % on 3rd and short to be > Kap's for xyz reasons, that's reasonable.

If I were then to say 'and look, Kap has completed 0/1 passes so far in such situations, which backs me up', that would not be reasonable.

IRM is doing that.

you're still remembering it wrong, and even though Bledsoe took a few more weeks to be ready, he was still the cagey veteran that BB benched due to injury.

Brady or Bledsoe was a GIGANTIC debate at the time. Bledsoe was hugely popular and had just signed a 10 year, $100m deal that offseason. and he did some back and play in the AFCCG, and was bench again

Harbaugh wasn't forced to make his decision after 1 week, but there is still a valuable anlogy

assani said it much better than i can but this is 100% my take on this. there is nothing smith can do that kaepernick can't except have 8 years of experience under his belt. and everything smith CAN do, kaepernick can do much much better. the niners successfully built a team and offensive gameplan that can beat any team on any week with a huge constraint on the qb (last in the nfl in deep attempts this year and only freeman and mccoy among starters threw fewer deep balls last year). of all 8 division leaders the niners are the only one where the QB is their weakest position, and harbaugh has masked that so well by crafting the optimal offense for smith to do well in that people honestly believe smith should be the qb going forward. if this is what harbaugh and roman can do with alex ****ing smith, imagine what they can do with a qb who DOES have incredible physical upside

Alex Smith is bad, it should be a no brainer to start Kaepernick if it wasn't for the ******ed extension they gave Smith last year.

I don't accept the premise that the stats were complied in ~identical (or even similar) situations, which they would need to be in order to make them comparable. As I said, a great deal of Kap's snaps so far have been when his team has been down big with time running out.

i'm using the same metrics people are using at face value without considering all the reasons those metrics say what they do, only trying to apply it more accurately than they do. nothing more. i hardly expect kaepernick to throw 26% deep balls over a full season, but > 17 or whatever smith's career high is is pretty certain

Do you think the probability that Kaepernick's 'true' deep % is greater than Smith's 'true' deep%, or best guess as to the magnitude of the difference, is completely, 100%, not at all affected by those 49 throws?

If so, we might just fundamentally disagree there. Regardless of differing situations (fwiw, Kaep was never down more than 10 to the Rams and made 23 of those 49 attempts in a blowout victory), I feel that those attempts are absolutely not evidence to throw away (not that they should be overinterpreted, of course.) I strongly believe that the most accurate assessment possible of Kaepernick's tt deep% includes and properly weights those attempts, and that such an assessment will be, on average, more accurate than one completely ignoring them.

Bledsoe never saw the field again after his backup, Brady, replaced him due to injury.

No, but close enough.

Of course, but you're trivialising the process of accurately weighing those attempts. The small improvement in predictive ability which is obtained from accurate assessment of very small samples is massively outweighed, in my estimation, by the frequent gross misapplication of such data by many people when we're discussing such things in an open forum.

cliffs: small sample analysis does more harm than good yo

at the least, those stats should come with a public health warning attached

Great posts by Assani.

Please stick with Alex Smith all year, and preferably next year too.

Signed,
Seahawks fans

In other news, Donovan McNabb didn't know there were ties, Jim Mora didn't want to talk about playoffs, and Aaron Rodgers is a good QB

They don't have to keep him next season. The $7.5M base for 2013 is GTD if he is on the roster March 1.

The extension wasn't an estension anyways. Alex was a FA.

so who has the hot hand? MNF vs AZ, LOLtie vs STL, and MNF vs CHI...

Passing:
Alex Smith: 25 for 27, 304 yards, 4 TD, 0 INT
Colin Kaepernick: 27 of 40, 360 yards, 2 TD, 0 INT

Rushing:
Alex Smith: 3 rush, 11 yards
Colin Kaepernick: 8 rush, 88 yards, 1 TD

Start Smith, ldfo

I think you gotta explore the CK option for his kurtosis, but is he actually a fave to be better than Smith if we assume that Smith is ~ the 20th best QB in the NFL? Seems like he's still a dog based on limited info.

You guys are going nuts over 1 game sample size.
http://www.pro-football-reference.co...12/passing.htm
Alex Smith #1 in completion %, #6 in AY/A, #11 in Int%. He's actually been an excellent QB despite your eye test.

Kaepernick had one really good game against a Bears defense that for whatever reason couldn't do anything against the run or anything else. I trust the sample of 30ish games going back to 2011 more than the 45 minutes I saw last night.

I'm not very well educated on the scouting report on either QB, so I take your guys word on it....but how exactly did Smith go #1 overall(and ahead of Aaron Rodgers!) if he didn't have great natural tools?


which posts ITT have bashed Smith based upon the eye test?

Peyton chose Denver because he saw Kaepernick's potential and knew that at the first sign of trouble, Kaepernick would nab the starting job.

A bunch of scouting reports I found basically said the same thing as this: very intelligent (40/50 on Wonderlic), accurate, athletic, "leader", mediocre arm. Apparently he had a legit great pro day as well. fwiw seems like Rodgers and Smith were in tight contention for the #1 pick for the majority of the pre-draft process, I think Rodgers was considered to have done poorly on his interviews as well which probably didn't help. I think teams these days tend to be more knowledgeable about how college system can affect quarterback prospects and seem to place greater priority on "tools". In retrospect, Smith is a pretty classic example of a system QB.

kaep probably has one of the strongest arms in the league. he apparently threw in the low 90s as a high school pitcher and was clocked throwing the football the fastest out of everybody at the 2011 combine, which includes newton and mallett who both have ridic arms.

he also injured randy moss's finger with a pass last night.

Irrelevant to your point, but it bears mentioning that he has ridiculous touch on his ball when needed. See: the deep out (I think) he threw to Vernon Davis last night.