Well I don’t screen per se. Screens or sorts, really. So I wouldn’t buy a stock simply because it has a low P/E. I would distinguish between estimation/modelling and portfolio construction. Even if the modelling suggests that a low P/E stock has an attrractive return, then that wouldn’t necessarily mean that a stock should be included in a portfolio.
Further, I would note that I usually incorporate a momentum term. If the P/E is low because the stock has fallen, then that would get pickedup by the momentum factor. Also, as you note a P/E of 3 would be pretty extreme. I try to winsorize so that outliers do not make as significant an impact, in estimation or prediction.
Well, TA tends to be about supply and demand for securities, so you could almost certainly do it with treasurys, and those prices tend to be the baseline for other interest rate securities.
However individual company issues might be more problematic, because they won’t always have as much liquidity. Still, it can be worth trying, if only because base yields will affect all securties.
The bigger question in FI is whether you would use TA on prices or on yields. The techniques probably work better on price. Some of them may also work with yields, but since yields go down when prices go up, anything that is not a symmetrical indicator would act in strange ways.