The value of deferral
http://www.valuewalk.com/2016/08/how-warren-buffett-used-deferred-taxes-to-make-twice-as-much-money/
"Imagine that Berkshire had only $1, which we put in a security that doubled by yearend and was then sold. Imagine further that we used the after-tax proceeds to repeat this process in each of the next 19 years, scoring a double each time. At the end of 20 years, the 34% capital gains tax that we would have paid on the profits from each sale would have delivered about $13,000 to the government and we would be left with about $25,250. Not bad. If, however, we made a single fantastic investment that itself doubled 20 times during 20 years, our dollar would grow to $1,048,576. Were we then to cash out, we would pay a 34% tax of roughly $356,500 and be left with about $692,000.
The sole reason for this staggering difference in results would be the timing of tax payments. Interestingly, the government would gain from Scenario 2 in exactly the same 27:1 as we – taking in taxes of $356,500 vs. $13,000 - though admittedly, it would have to wait for its money."
Could perhaps be a reason for why non-market cap indexes are actually flawled?
Lets say you invest in a equal weight index (or one of those Arnott fundamental indexes). They will have to rebalance every year (generating taxes), in the mean time, if the market cap index catches a few huge runners (and keeps it holding for a really long-time) it will produce a significant compounding effect due lack of tax payments. It feels to me that the data (of market cap vs fundamental indexes) is extremely noisy and very much dependend on outliers, so therefore, historical tests for 1 or 2 decades are not enough to say that 'fundamental indices' are better than market cap
I got no strong view one way or the other (other than preferring market cap indices just because its more of a proven concept, so, as a precaution) but I do think that the level of sensitivity to outliers, especially when taxes are taken into consideration could create a significant difference in long-term performance