I've been working a bit on my HU game online (tournament style) and wanted some opinions on the following ideas concerning raises. My starting point here is really Harrington's HU strategy section.
First, OOP: Here, as Harrington points out, you really want a lot of folds because you're playing the hand at a disadvantage. So, it's a good idea to raise larger. I've been making it 4 BB as a general rule, but I'm actually thinking 5 BB as a starting point is probably a little better, and raising top 30% of starting hands.
As to the raising quantity, what you want here imo is at least 70% folds. So, if they seem to be calling more than that, I think you should make your raise a little larger (until you get the desired result), and if they're calling less, you can go down a little, decreasing your risk for the same result.
But what I'm really starting to think on this is that the more important issue is actually how often they're folding to your CB, which I consistently put at 1/2 pot HU. They should actually hit the flop only 1/3 of the time, and if they're calling much more often than that, you're probably losing money at least when you miss and they take down the hand: You've put in, for example, T100 PF, plus T100 on the flop, so you lose T200 when you lose, but win only T100 when they call PF but fold on the flop.
I think the appropriate response to this is to tighten up on your selection of raising hands but be willing to play these rather more aggressively. Rather than raising top 30%, tighten up to top 20% and also don't raise to more than 5 BB. These hands are good enough that I think you'll be compensated for your potential CB losses on misses by taking down very large pots on these strong hands.
With position, it's rather different, as you WANT a caller PF with your good hands. You have a good hand AND position. So, you also raise less. I usually make it 2.5 BB to start with. Here, I think you'd like to get a caller 1/2 the time.
If you don't succeed with this raising top 30%, then I think you should switch over to raising BOTTOM 30% as well as all hands that are top 30% but not top 10% (on which you just complete). That would mean you're raising 50% of the time and gaining a lot of value on your really bad hands while also keeping them guessing with a fair number of decent-strength real raising hands.
That's it for now. Opinions?