Bayesian updating is just updating probability of future events as new information comes in. This way we will never stay too rigid to one stance and adapt to low-contrast events i.e small changes that trickle in that can slip us which we otherwise may have reacted to, if all the information came together. The marvel that is the human brain does this naturally as long as its not tied to an adverse stance or experience. Fear, greed, wanting to be right (ego) by choosing a side and sticking to it all come in the way though.
So allocations tied to probability (in this case asset allocations) should more or less reflect the new inputs that come in. Similar approach can also be used to allocations within an asset class like equity which has been discussed beautifully in this thread. Kelly’s formula as well is a wonderful tool for allocation within an asset class, similar to what’s discussed in this thread. In this context, applying Bayesian updating will require you to change stance i.e conviction as the thesis behind the conviction diverges from reality, requiring you to reduce allocation according to the probability/percentage of divergence.
All this involves costs though - both in terms of time spent and also cost of churning so it is very important to know the importance in terms of impact of new information that trickles in. This is something I have been practising recently and struggling with it to be honest, as it takes away both the highs and the lows. You never feel the highs of being right nor the lows of being wrong. It does get rid of the hubris of being right and the humbling of being wrong though.