So after I quickly put together this post this morning I thought "you know. I can't do anything about the second two, but I can track GAA when leading in the third period."

So guess what, I went ahead and looked at every game where the Avs were either ahead or tied in the third period and calculated their GFA (Goals For Average) and GAA (Goals Against Average) (EN goals didn't count in the GFA). I think the results are telling.

So here's how I calculated it. there's a tied or leading part, and just a leading part.
the Tied or leading part:

I counted GF and GA as goals scored when the Avs are tied or leading by 1 or 2 in the third period. Any goals after that didn't count, unless the Avs went back to up by 1 or 2 or tied. The time counted was time counted when the Avs were tied, up 1 or 2. Any goals scored outside of this didn't count in the calculations. If the Avs are down a goal their strategy changes, and the goal of this is to see if the "cling for dear life" strategy is working. If the Avs went up by 3 goals, i figured it's not really a close game and playing defensive isn't really clinging for dear life.

Here's an example: On December Colorado Lost to Columbus 5-4. The Avs entered the period up 3-1. Here's the scoring summary for the third period of that game:
Columbus 7:07, Rick Nash 18 (Sergei Fedorov, Gilbert Brule)
Columbus 9:26, Joakim Lindstrom 1 (power play) (Kris Beech, Curtis Glencross)
Columbus 14:44, Nikolai Zherdev 10 (Curtis Glencross, Jan Hejda)
Colorado 15:25, Scott Hannan 1 (Ryan Smyth, Paul Stastny)
Columbus 15:55, Dan Fritsche 4 (Jason Chimera, Kris Beech)

So the Nash, Zheredev and Lindstrom goals counted as GA, because the Avs were up 2, then 1, then tied. The Hannan goal didn't count as a GF because the Avs were down one, but the Fritsche goal counted because the Avs were tied again. That's 4 GA and 0 GF. Also all the time in that period counted into my calculations except the time between Zheredev's and Hannan's goal. during that Time the Avs were down, and presumably not trying to go in their shell (even though with Q you never know).

Here's another example that doesn't look like I'm making criteria to fit my theory:

On October 7th the Avs and Sharks were tied 2-2 entering the third period. Here's the scoring summary from the third period:
Colorado 0:51, Paul Stastny 4 (Milan Hejduk, Jordan Leopold)
Colorado 2:47, Milan Hejduk 2 (Ryan Smyth, Paul Stastny)
Colorado 7:48, Marek Svatos 1 (Jeff Finger, Tyler Arnason)
Colorado 8:04, Paul Stastny 5 (unassisted)
Colorado 8:06, Ryan Smyth 1 (Paul Stastny)
San Jose 17:53, Ryane Clowe 2 (power play) (Steve Bernier, Matt Carle)

So the first Stastny goal, the Hejduk Goal and the Svatos goal all counted as GF. (since the Avs were tied, up 1 and up 2). Neither the second Stastny goal, or the Smyth goal counted as GF and the Clowe goal didn't count as a GA. The time used in the GAA and GFA for this game was 7 min 48 seconds. (when the Avs then went up by 3 on the Svats goal).

After doing this, here was the results:

Month GF GA Min
Oct 6 6 120.92
2.02 1 2.48 2.98
Nov 3 0 79.37
1.32 0 2.27 0.00
Dec 10 9 179.85
3.00 1 3.00 3.00
Jan 4 3 107.98
1.80 1 1.67 1.67
Feb 5 6 113.32
1.89 2 1.59 3.18

Total 28 24 601.43
10.02 5 2.29 2.39

All in all, this isn't actually too bad. When tied, up by 1 or 2 in the third the Avs score slightly less often than they are scored upon, and have a GAA of 2.39. Not great, but certainly not awful (of course a solid November helped that tremendously)
Now let's look at whan happens, only when the Avs are playing with a small lead in the second or third period:
Month GF GA Min Min/60 ENF GFA GAA
Oct 4 2 39.67 0.66 1 4.54 3.03
Nov 3 0 76.02 1.27 0 2.37 0.00
Dec 7 6 127.20 2.12 1 2.83 2.83
Jan 1 1 10.87 0.18 0 5.52 5.52
Feb 4 4 56.33 0.94 2 2.13 4.26

Total 19 13 310.08 5.17 4 2.90 2.52

Look at that GAA balloon. Especially in Jan and Feb of this season. First notice the Avs only played with a small lead in January for under 11 minutes. WTF!?! Also look at those GAA in Jan and Feb when protecting a lead... 5.52 and 4.26. Compare that to October when they still had a GAA of 3.03, but were scoring at a 4.54 rate. In November they didn't give up a goal in 76 minutes of leading time. I think it's fair to say that going into a defensive shell (at least the defensive shell the Avs employ) actually INCREASES the likelihood you will give up a goal.

The reason behind this is partially because the other team is pressing, but I think it has more to do with the fact that both Theo and Budaj are great when taking shots of over 60ft.

So to recap STOP PLAYING LIKE YOU ARE ON THE PK WHEN YOU HAVE A LATE SMALL LEAD!!! You have a better chance of holding the opponent to no more goals if you forecheck and keep the puck in their zone. I know this is counterintuitive to the men behind the Avs bench, but really it's true.