"The ONLY reason". That is FALSE, and Maestro may not be a maestro in this subject unless he wants to be in the name. Here is what people need to understand: A. Each strategy has a mean and a variance. B. If you allocate capital to 2 strategies with a given weight to each that sum up to 1 (which is called convex combination), then this implies is the mean of the overall strategy is the weight sum of the means of individual strategies, but the variance of the combined strategies depends not only on the variance of each, but also on the covariance of the two (on the covariance matrix in case of many). C. If the convariance in point 2 is zero or negative, then the variance of the combination is less than the variance of each strategy. The implication of point C are multifold: 1. You have a portfolio with a lower variance if the covariance is negative or zero. 2. If you are a broker, and you have a customer with a combined strategy that has a non-positive covariance then you will see his equity bounce less. The broker can then feel better and may impose less margin requirements (you can see this in case of SPAN) 3. A trader needs to understand that variance has an impact on the fraction of capital to allocate to a trade. If the variance is high, the fraction of cashthat should put in a trade decreases when compared to the cash sitting idle, as otherwise you decrease the mean return when the fraction goes beyond a give threshold value. 4. Unless your probability of success is way too high (almost 1), the cash sitting idle for the next trade in point 3 will be much higher than the cash working in trade or set of trades. Therefore, you do not need to acquire a higher margin. And acquiring it can actually be dangerous as it can reduce your return. 5. If you want higher margin, then understand that if you get half margin requirement, then you need to scale you trade size by half. Because if you do not then, according to point 4. you are doubling the size of your trades when your money management tells you that you should proceed with half as much. Conclusions: A. Margin increasing techniques are really the wrong issue. If you want to minimize it, there is much easier ways. You need to learn one topic only: It is called banking in the options market, where you can lend and borrow as much as you want at an interest rate equal to treasuries interest rates. The only requirement is that you will not be able to have a margin ratio less than 2% (check IB for instance). B. I hope that my above comments make you understand things differently and point you to the way to raise the correct questions and the right directions to answer them. C. If you want to learn math/probs/numbers/rigorousstuff, would you learn it from a poet/psychologist?
Poet / Psychologist (- type) here. (I'm not bashing on you. I know that you're helping Maestro explain himself.) Variance / Covariance / Standard Dev. / Correlation / Composites ... ... ... great. We all understand Maestro. All Quant. / Stat type services offered by institutions follow the above (plus more like delta and other stuff) to assess your margins, regardless of options and others, mainly monitored by having swap accounts. You borrow stocks, margin, and etc. by paying extra $$$ for those services. Now... I'll bite: The core / basic idea of portfolio management regardless of multiple systems or symbols, the basic idea is to diversify. In terms of what riskfreetrading has mentioned, covariance would relatively fit this (along with correlation). Simply, all this is equity curve analysis. I'm no expert but I understand some basic probability and statistics. Though, things don't work out that way. math / probs / numbers / rigorousstuff does not solve any problems with any backtesting or quantitative tests due to the fact that the market are always changing. It's only going around in circles. I don't believe that being rigorously mathematical, scientific, or computational will solve it. There are few ways to deal with this. One is to assess each system individually according to their nature of the system. Two is to understand the system and have a contingency point based on the market itself, along with the system performance. Three is to add other sources of information. It can be newsfeed, intermarket behavior and others. Four... well there are others but can't mention it here, except to TEST EVERYTHING. There is no easy way out, no magic formula, and no short cuts. I can have 4 systems trading based on one market edge. Under equity curve analysis I can have the perfect, neutral balance of portfolio. When the market edge deminishes, you're dealing with a bad system allocation. Like I mentioned... I'm a poet / philosopher / psychologist.
Riskfree, are you Jack Hershey's son? I thought Jack had the prize for long-winded posts utterly devoid of content, but you have surpassed him. Please tell me how I can borrow "as much as I want" in the options market. I want to borrow $100 mm. Or more. Are you insane? Are you, along with Maestro, Mark Brown, and Nitro, an inmate at Lakeshore Hospital Psychiatric Wing? It was a big mistake to allow Internet access in the day room! And I don't believe you have an account at IB. Minimum to open is $10k now. I know you have a $300 account at Oanda, but that does not qualify you as a real trader.
Hey T-Dog, have you heard of "Professor Burke Brown," leading light of Behavioral Finance? No? Funny, neither have I. No citations, no published research, nothing published in the field at all, no one but Maestro has ever heard of him... for such a famous theoretician, he sure keeps a low profile. You'd think he would want to disseminate his mind bugling [sic] discoveries. Maybe he was a professor at the same fictitious institution of higher learning that awarded Maestro his fictitious PhD.
it ALL boils down to 2 points: 1.) is your sharpe ratio higher after diversification. 2.) can you lever (have enough margn) to a return that fits to your needs. in very specific siutations the margin issue can become tricky. for example when people do a lot of notional funding, which in essence means that they try to run at highest possible capital utilisation. thus their cost of an additional strategy is to give up another one. now you might increase your sharpe by adding strategies, but you cannot lever them to the former level. so 1.) is in place, but 2.) is not. i think this is what maestro tries to say in a misleading way. so, he is not wrong, he is just talking about something else than the others here, who are at a point, where 2.) is not an issue ... IMHO.
In terms of this thread about margin and correlation... I have to agree. Having a lower variance (volatility... Std. Dev) is what you are aiming for. Having a low Sharpe does help. Having a low Margin does help. Though, it's not all of what's important. It's part of what is important. Lower variance and lower margin is not what you really want to be aiming for, it's not the goal of why you're developing a system. It's about assessing a system and portfolio for the balance between robust and curve-fitted. It's about developing and researching ways to keep your systems and portfolios managable. It's about... whole lotta other stuff... (which again, won't be mentioned here, but you get the point)
Another typical thread of this kind, this time after 156 replies, to reveal again another new discovery on ET! Thanks anyway! Regards from a newbie, OddTrader
well, we could discuss politics and brain science as well, but they are (as your comment on robustness vs fitting) not subject of the discussion at hand. which is about does noncorrelation make sense in a portfolio. necessarily assuming that they are nonFits, otherwise the discussion is useless to begin with ...