What's with the attitude? You just ruined my image of you - before I pictured you raking in the bills intraday all day long when I saw your posts, now I can only imagine you yelling down the hall "Mom, someone is being impulsive and thoughtless on the internet again!". But I will try to be less impulsive from now on. Seriously, I am not going to criticise you for not reading my post carefully enough, but propseeker understood correctly. I am not referring to any theoretical convergence while ignoring transaction costs. I am just saying most of us would get more payoff on our scarce time by adding more strategies to a portfolio than by elaborately fine tuning their relative weights. If that does not apply to you then congratulations on your large bankroll and excellent quant skills. Diversification is easy, optimization is hard. Plus for many of us a robust optimization process would just tell us to switch between trading 1.4 and .83 contracts instead of 1, what good does that do? Here's a thought experiment. Or a real experiment if you have way too much time on your hands. Assume a universe of 10 strategies. Make up true expected returns, vols, and correlations for them. If you are hung up on transaction costs, assume they are all day trades and your rebalancing horizon is a day. Simulate a time series of returns for each over 10 years, and do it 1000 times. Now imagine it is year 5. Trader 1 only knows about 5 strategies, but can optimize like a champ. But only using the noisy historical returns to date, not knowing the true expected returns and var-covar matrix. Trader 2 knows about all 10 strategies but just equally weights them. They both trade the last 5 years. Out of 1000 simulations, who does better more often? I bet on Trader 2. And I think Trader 2 would do even better if you make it more realistic by throwin in things like regime shifts, time varying vol, etc. Higher n is better.
What is that? I though we had a serious discussion here. Have you found a broker with zero day trading commissions? Oh my, this thread is a total waste of time. Bye.
No - but the commssions are not affected by your rebalancing. I don't see why this is not clear. The point of the previous post was to make a simple example where transaction costs don't complicate the issue. I'll make it even more simple. At the beginning of the day you have 50% of your capital allocated to Strategy A and 50% in Strategy B. Strategy A holds apple during the day and sells at the close. Strategy B holds bannana and sells at the close. You paid your costs already and now you have cash. Your costs are not affected by tomorrow's allocation - you are free to go 50/50 again or 100/0, your costs are the same either way. You still say transaction costs are unambigously higher for 50/50 than for the optimal weights?