Is someone a good coach?

[Sleeping in EQ]

The first two sets of numbers I'd look at to decide would be:

1. Turnovers

and

2. Penalties

I'd look at the total numbers and yards, the ratios, the types of penalties and turnovers, comparisons with opponents numbers etc.

If those numbers look consistently good, you probably have a good coach. You may still have players without adequate talent.

So what are Kyle's and Bronco's numbers on turnovers and penalties for the last two years? Can one of you stat masters look them up?

[pelagius]

Originally Posted by Sleeping in EQ

The first two sets of numbers I'd look at to decide would be:

1. Turnovers

I would guess that turnovers probably has more mean reversion that just about any other statistic in football. My guess is that over any period, sorting on turnovers is mostly a sort on noise and extracting that signal portion is very difficult because the signal to noise ratio is very low. Indy or Jay do guys have any evidence of persistence in turnover margin?

[Indy Coug]

Originally Posted by Sleeping in EQ

The first two sets of numbers I'd look at to decide would be:

1. Turnovers

and

2. Penalties

I'd look at the total numbers and yards, the ratios, the types of penalties and turnovers, comparisons with opponents numbers etc.

If those numbers look consistently good, you probably have a good coach. You may still have players without adequate talent.

So what are Kyle's and Bronco's numbers on turnovers and penalties for the last two years? Can one of you stat masters look them up?

Florida 2006 - 66th in turnovers, 109th in penalty yardage

Do you honestly think LaVell's teams excelled in these areas? Turnovers and penalties are not a valid indicator of good coaching.

[jay santos]

Originally Posted by pelagius

I would guess that turnovers probably has more mean reversion that just about any other statistic in football. My guess is that over any period, sorting on turnovers is mostly a sort on noise and extracting that signal portion is very difficult because the signal to noise ratio is very low. Indy or Jay do guys have any evidence of persistence in turnover margin?

BYU always had terrible turnover margin under Lavell. I think it correlates to the style of offense and wouldn't be a good measure of coaching ability.

All that matters is W's and that's really all you can measure a coach on.

The only way to measure a coach IMHO is to take W/L (or even better take computer ratings to remove SOS effect) vs expected W/L. The expected W/L could be for a season or it could be long term. The season analysis would be how good of a coaching job a coach did with a given talent level, and the long term would also include an element for recruiting and developing talent. But the expected W/L will always be subjective.

Thus far, I'd rate Bronco and Kyle as doing a fairly equal job so far.

I'd say Bronco slightly underachieved his first year and slightly overachieved his second year. I don't know as much about the U's program, but I'd say Kyle has overall probably met expectation right on.

[pelagius]

Originally Posted by Indy Coug

Florida 2006 - 66th in turnovers, 109th in penalty yardage

Do you honestly think LaVell's teams excelled in these areas? Turnovers and penalties are not a valid indicator of good coaching.

The conventional wisdom is that penalties are a proxy for discipline. There may be some truth to this, but this doesn't mean it is a good proxy for coach quality even if discipline is something that helps teams win. For example, one could a imagine the penalties also vary according to offensive and defensive style. More complicated offenses may natural incur more penalties but the trade-off may be worth it. Also, penalties is probably strongly correlated with number of snaps a given teams has and over a season more successful teams will have more snaps [the same is true for turnovers]. The last part is easy to get around by going to a penalties per snap statistics but very people actually go to the trouble of even normalizing measures when the discuss football statistics.

[Sleeping in EQ]

Originally Posted by Indy Coug

Florida 2006 - 66th in turnovers, 109th in penalty yardage

Do you honestly think LaVell's teams excelled in these areas? Turnovers and penalties are not a valid indicator of good coaching.

Do you honestly think I was referring only to the numbers you mentioned?

[Tex]

Originally Posted by pelagius

The conventional wisdom is that penalties are a proxy for discipline. There may be some truth to this, but this doesn't mean it is a good proxy for coach quality even if discipline is something that helps teams win.

I agree with this. I especially look at false starts as a measure of team discipline. False starts are easily the most unnecessary penalty in a game, and it's especially vexing when you get multiple.

Close on it's heels is unsportsmanlike conduct, and third is probably holding.

[Indy Coug]

Originally Posted by pelagius

The conventional wisdom is that penalties are a proxy for discipline. There may be some truth to this, but this doesn't mean it is a good proxy for coach quality even if discipline is something that helps teams win. For example, one could a imagine the penalties also vary according to offensive and defensive style. More complicated offenses may natural incur more penalties but the trade-off may be worth it. Also, penalties is probably strongly correlated with number of snaps a given teams has and over a season more successful teams will have more snaps. The last part is easy to get around by going to a penalties per snap statistics but very people actually go to the trouble of even normalizing measures when the discuss football statistics.

For starters, it would be interesting to see the % of running plays negated by holding versus the % of passing plays negated by holding.

Unfortunately, that kind of specialized data is not readily available to the public, if it exists at all.

[Sleeping in EQ]

Originally Posted by pelagius

The conventional wisdom is that penalties are a proxy for discipline. There may be some truth to this, but this doesn't mean it is a good proxy for coach quality even if discipline is something that helps teams win. For example, one could a imagine the penalties also vary according to offensive and defensive style. More complicated offenses may natural incur more penalties but the trade-off may be worth it. Also, penalties is probably strongly correlated with number of snaps a given teams has and over a season more successful teams will have more snaps [the same is true for turnovers]. The last part is easy to get around by going to a penalties per snap statistics but very people actually go to the trouble of even normalizing measures when the discuss football statistics.

And of course I would look at those things.

[pelagius]

Originally Posted by jay santos

The only way to measure a coach IMHO is to take W/L (or even better take computer ratings to remove SOS effect) vs expected W/L. The expected W/L could be for a season or it could be long term. The season analysis would be how good of a coaching job a coach did with a given talent level, and the long term would also include an element for recruiting and developing talent. But the expected W/L will always be subjective.

I like this conceptually at least, but it is tricky too. Coming up with a good ex ante measure is difficult. I think for this season I would probably do Edwards era season end computer rank average for seasons where BYU started a quarterback with zero or no experience (say three or less). If I had the data I would condition on the number of returning non-quarterback starters too.