Stevens is about statistical analysis and everyone being on the same page. This means no rolling the ball midcourt unless time is running out, no gambling on defense, no pounding the rock.
In certain spots, gambling on defense could be the correct play from a mathematical perspective.
which sport is that??
In basketball gambling on defense usually doesn't turn out well. Rondo taking focus away from his man, trying to go for the steal somewhere else is risky. Player x predicting his player won't go to his left this time, when he can't go right is risky. Stevens wants high percentage plays from the games i've seen him coach. If that guy who can't dribble left very well, beats you going left once or twice, he is not going to care as long as you don't let him beat you going to his right.
Here are some numbers I made up in a simplified case. The numbers are not intended to represent accurate real-game scenarios.
Let's say the following happens with normal defense:
The opposing team scores two points 45% of the time and on the following possession will give up two points 40% of the time because they can set their half-court defense. The 55% of the time that the Celtics get a stop, they score on 50% of possessions because they sometimes have higher percentage opportunities in transition.
Let's say the following happens with a gambling defensive strategy.
Let's say that when gambling works, you give up zero points and have a 70% chance of scoring due to an even higher rate of having transition scoring opportunities. When gambling doesn't work, you give up points 60% of the time and have a 40% chance of scoring on the following offensive possession.
The value of the first strategy is:
0.45(-2+0.4*2)+0.55(0+.5*2)=0.01
The value of the second strategy is:
x(0+0.7*2)+(1-x)(-2*0.6+0.4*2), where x is the rate at which gambling succeeds.
If I did the arithmetic right, for the value of the second strategy to be higher than than the value of the first strategy, gambling for a steal has to be successful more than 27.8% of the time.
I don't know what the actual value of gambling for a steal is. I'd guess that always going for a steal is a bad strategy. I'd guess that never gambling for a steal is also sub-optimal. Rondo is probably better at stealing the ball than most other point guards, so he should probably be gambling for a steal more often than other players.