Why does the market have to be open for you to make a decision on your multiple MA chart? (And it is, in fact, open.) Earl Grey, Darjeeling or Bergamot?
Aw, shucks. Are you trying to boggle this truck driver’s mind? Variance ratio? Another poster I respect used the abbreviation “GBM” this very week. Lordy, if I were smart, I would get my a$$ off a trading website where the big boys roam and hang out on my CB, talking about bears, babes, and 4 wheelers. So here I stay. 10-4?
What makes MA's suck? 1. Lots of lag. 2. Too popular--everyone is using them. 3. Trying to use them in isolation. 4. Trying to use them as short term signals. Since they have a lot of lag, they work more effectively for exploiting the bigger moves on longer time frames. 5. Using MA's with poor position sizing. 6. Not constantly re-optimizing the MA's as market conditions change. You probably need to re-optimize frequently if you are using MA's in isolation rather than filtering the MA signals with something else. Relying on too few signals or too many is likely to give you inferior results. Keep in mind that your second strategy--buying below the MA and selling above it--fails in prolonged downtrends. In that case, you are cutting your wins short but letting your losses run! So you will have a lot of small wins occasionally wiped out by large losses unless you have other elements in your system to prevent this outcome.
"A trading system consisting of entries and exits solely on moving averages should not provide an edge." I haven't tested it, but my intuitive guess is that it would provide a small edge over random entry. It probably would not beat or even match buy and hold in a prolonged bull market, but it might in a choppier market, or over periods of time long enough to include major sell offs. The key to making that slight edge eventually pay off in a major way is position sizing. Once you have an edge, position sizing becomes the most important ingredient in the trading recipe.
"if you backtest a randomly chosen indicator, over a randomly chosen market, over a randomly chosen time period" -- I know may traders do this, but why bother? Why would anyone want to trade based on a single signal with very low potency (and anyone can see that these signals used in isolation are very weak even without formal statistical testing--I mean, it quickly becomes very obvious to any beginning trader). The trick is to find a group of signals that work together in a logical way like coordinating the pieces on a chess board, and I remain unpersuaded that testing individual ingredients one at a time helps you to put together a better recipe (pardon the mixed metaphor). The individual ingredients in a cake taste nothing like a cake. The complete recipe is everything. The trader should look for a convergence of signals that increases the reliability of his trades, and they should do so for logical reasons to reduce the odds that it is just a lucky coincidence. I still maintain that thinking is superior to purely mechanical approaches that computers have elevated to unprecedented popularity. Nor should anyone waste his time looking for the Holy Grail of Trading Systems. The trader just needs a big enough edge to make position sizing pay off. I also don't understand why some retail traders want to trade like a hedge fund. The typical retail trader has a set of advantages and disadvantages that is quite different from those of a hedge fund. Furthermore, trading like a hedge fund means you are going head to head with them, whereas doing something different, something that leverages the strengths rather than the weaknesses of having a small account versus the huge capital resources of a fund, allows the retail trader to trade with an edge by focusing on niches that big funds can not efficiently exploit for various reasons.
I'd agree that you should never trade just one signal (I use dozens), but it doesn't follow from that you shouldn't test individual signals. To bend your analogy, I'm not going to throw an ingredient into my cake without checking to see it's edible and hasn't passed it's expiry date. I'm adding together signals in a linear way (with weighted averages). The weight a given signal has will depend on how good it is, and whether it diversifies the other signals. If I don't test in isolation I can't find this information. If a signal adds value despite being crap individually, or without being an excellent diversifier, then it must be because of some weird non linear effect and frankly I wouldn't want to add a signal that added value like that because I wouldn't be able to explain what was going on - a big no, no in any trading system. Computers are indeed over rated, but a trader who thinks accompanied by sensible backtesting is a potent combination that will beat a trader who is acting purely on gut instinct I guess it depends on what you mean by 'trading like a hedge fund'. Obviously there are advantages and also disadvantages to having modest capital, but generally speaking hedge funds make fewer stupid mistakes than the average retail trader. So I'd argue that most retail punters could learn a lot from better understanding the way that quant and systematic funds in particular manage their risk and trading costs. GAT