@hcosah1. The way I look at pre-Covid versus post-Covid status...

@hcosah

1. The way I look at pre-Covid versus post-Covid status is whether my proportional ownership of the businesses I like has increased relatively compared to the prior period, rather than the specifics of the portfolio numbers- and in my case, I certainly think my current position is fundamentally better than my previous position, irrespective of the prices of the various securities I own. The jury for me is still out whether this pans out in the future. I can't speak for others but I most certainly did not breeze through this period- my gut was wrenching making some of the buys I made in March, which seems easy in retrospect. I also have had huge personal volatility in my equities- at one stage I was down about 70% from my pre-Covid peak, although I have done quite well in the subsequent period (based upon market pricing anyway).

2. Re:modelling errorThe truth is every single company needs a bespoke model. I honestly think DCF is BS- I believe it is only reliable for things like utilities, and there is too much sensitivity to key inputs which are fundamentally arbitrary. To be very clear, I do not believe in setting a different discount rate for different stocks in my methodology. I manage risk at the portfolio level and in % allocation and not by setting different rates at the valuation level. To me, this fundamentally makes more sense.If you want specifics of modelling error, that I have personally been involved with, I can lay out the following two cases:- I bought XRO at around $15 when tech was going through its downturn. Basically I realised that they were being managed to breakeven and that the subscription revenue was very sticky and predictable as was the growth trajectory- ie the valuation methodology at the time was incorrect. Now it is most definitely priced with this in mind- there is no longer any advantage in that insight.- I lucked into APT via TCH (I bought TCH when its equity value in AFY was more or less its market capitalisation). The MD died and there was a reverse takeover by the fledgling business. Basically noone understood how afterpay worked back then (and most still don't appear to). I basically figured out that there wasn't an existing fast receivables turnover model, and that the way to model it out was working out the growth trajectory (which is possible to do with D2C products), use the bad debts as a form of customer acquisition cost, work out the sensitivity to inputs (which was with a very complex sensitivity analysis) then TEST, TEST,TEST my bespoke model as new information came out. When you have a bespoke model, if new information confirms the model, if you use a Bayesian framework, the probabilistic assessemnt of the accuracy of the model increases over time. However, you need to pre-specify your hypothesis in a probabilistic format. Anyway the essence of it is- I have developed my own model over many years that has gone through so many tests that I am reasonably confident that I am right. I know the assumptions that my model relies upon and if any new information proves any of my assumptions wrong, I would sell quickly. Obviously having very smart people that are short the stock makes you triple check that your analysis is right (I have probably checked >100 times in different ways over the years).

The sum of it is- you basically have to do the work yourself as it is not that easy to do. But the really big rewards don't come easily.