Here's my view (at D3, still working on D1)...
First, for me, it's much more of a sliding scale and depending on which range you are looking at, I value the ath/def/speed analysis differently. I have a baseline minimum for each that they must hit (these are absolute minimums and a player that just hit each of the 3 baselines would not intentionally be on any of my teams; if it just hits the minimum in one stat, they better be superior in one of the other 2). After that, I get into what I think is more valuable. I also have a point of diminishing returns, above which I start to value the other stats a little more
Baselines:
At the 1/2: Ath = 40 ; Speed = 60; Def = 40
At the 3: Ath == 45; Speed = 50; Def = 40
At the 4/5: Ath = 50; Spd = 10; Def = 40
Once the baseline is established, I value improvement in those attributes over the baseline as follows:
At the 1/2: Ath then Speed/def on a fairly even basis
At the 3: Ath then Def then speed
At the 4/5: Ath then def than way down in a canyon somewhere, buried under 100 feet of rock and molten lava, speed (that said, I like to have 1 or 2 bigs who are also fast for matchup situations, but they would get spot starts)
I do find that speed helps with steals in M2M, BUT, I don't think it creates a shutdown defender. For instance, I think a 45 Ath/90 Speed/40 def guard would get a lot of steals but a good opponent would also score a fair bit. On the otherhand, a 75 ath/70 speed guy with 75 def would get fewer steals but will make the opponent a much less effecient scorer.
I tend towards making the other team innefficient versus going for steals