I believe the point discussed was that any type of leveraged trading system employing monthly bars is untradable because the drawdowns necessarily become catastrophic. I asked Surf if he did any research verifying his assumption on historic data (say over the last 20 years on 8-10 of the most liquid future markets, margin/equity R 20%) he said he didn't because it's "common sense". This is not about the entry/exit accuracy of EOD/weekly/monthly bars vs. tick bars in an adverse walk forward test. This all started out looking at historic data (Harris, Yilmaz 1993-2008).
yes, thanks for articulating my thoughts here. the 'social science' people don't seem to grasp the entire picture. regards, surf
research isn't needed to state the obvious. the longer the time frame between points, the greater the risk of ruin---monthly--you gotta be kidding! surf
More nonsense that is besides the point. As expected, you don't reply to the argument that EOD bar DDs and EOM bar DDs historically don't differ strongly enough to have catastrophic margin results assuming reasonable margin/equity ratios. But what am I doing here? Arguing with a guy that doesn't trade.
how many moves, or series of moves, in one direction increases the odds that the next move or series will be in the same direction? untill this can be answered, using "trend" is nonsense. viewing past trends can be extremely decieving, leading many traders down the wrong road. surf
Why were Harris and Yilmaz able to extract positive out of sample returns out of monthly currency data. That is that question you have to answer. You are making wild claims although there's evidence to the contrary both in research as well as in three decades of managed futures CTA returns. The burden of proof is on you, not on anybody else.
positive runs occur in random data. can these runs be succesfully traded? 3 decades of CTA returns? only the winners last in this game, means nothing to the viability of a strategy---- AND not all CTA's are trend followers. surf
Out of sample. And yet you have problems coming up with 5 trendfollowing CTAs with AUM >$1bln that closed up shop because of catastrophic drawdowns over the last 20 years. By your standards, no evidence whatsoever is able to give any credibility to a given strategy. You're getting a bit repetitive here.
A semantic quibble, but OK -- to be very precise I should have written "drawdowns as [mis]understood by Mr Makloda."
Ok fair enough, I slightly missed the point then. I would still say that 20 years backtesting is quite possibly inadequate to judge catastrophic risk. Backtesting over any time period, no matter how long, may be inadequate because certain scenarios can occur which have not occurred in the past in quite the same way. So IMO 20 years data doesn't prove much other than that "mundane" drawdown risk was not a big issue. Having no stop for a whole month does have a clear theoretical risk of multiple correlated positions all falling a huge amount before the EoM, I don't see any getting around that. It does seem to me like using time-delayed stops is basically trying to collect a small put premium for free. If we go back 25 years, the stock market and other major markets never fell 50% overnight, does that mean that there's no risk of that happening? IMO there clearly is, and selling those puts is risky, even though the chances of getting creamed is very low, it is still there, and your "profits" are probably totally illusory. I feel that a system using decent leverage and month-delayed stops is to some extent disguising its risks and trying to get something for nothing. You can reduce whipsaws by holding on longer, but the flip side is that you increase the chance/size of loss in a genuine blow-up situation. Now it's quite possible that markets are mean-reverting enough to make this a profitable approach, even with the odd big hit here and there. But I'm not convinced that data back to the mid 80s is sufficient testing.