Identifying things that WERE bubbles is easy enough, depending on what you define a bubble to be. Inside a bubble, the intrinsic defining feature of a bubble is that it is NOT observable, or else everyone would make a fortune off those not observing it by predicting it and drive the price down to prevent it from ever actually being a "bubble". If you are capable of observing a bubble, then by definition you are NOT capable of stopping it.
You're right that timing is an often ignored part of predicting - making a bet costs money, and the juice is always running on speculative bets, especially contrarian ones (here's why). The market can always stay "irrational" longer than you can stay solvent.
Grantham's methodology is based on the historical likelihood of a price movement, which is the same methodology that gave subprime mortgage debt AAA credit rating. It's very reliable for a tight range where asset returns approximate well defined statistical distributions. However the problem with rare, extreme events is that they are easy to observe after the fact, but they are rare and by the time you can tell what is going on, they have an extreme effect on your balance sheet.
His "3 sigma" events are inherently things unlikely to occur, but most importantly its based off three things that change daily, exempli gratis: prices, equity, and the average P/E ratio. Companies with P/E ratios that have been more than 3 times the market P/E ratio include those failures like Microsoft, Google, Apple... Any company with rare growth potential should have an "excessive" P/E ratio. Tech stocks as a sector, for example, have high P/E ratios - but they have not "popped" in the long term (.com bubble being a blip in this longer term trend), they have actually increased the long term P/E ratio of the entire market because they have BECOME the market by crowding out slower growing industries. The price of creativity, innovation and technology is risk.