The well-respected trading technician and (imo) non-guru Adam H. Grimes has come to the same conclusion I have, which is that price behavior at moving averages is random and unreliable when used to predict future price direction. Here are some highlights from this discussion on his blog: "I have done extensive quantitative work on moving averages, and the answers I have found challenge many of our ideas and many of the ways technicians use moving averages. Based on my work: There are no “special” moving averages. (I.e., the 200 day is not special compared to the 193, 204 or any other average.) Pricing crossing or touching a moving average does not have significance for future market direction. The slope of a moving average is not a meaningful indicator of trend. Crossings of moving averages are not meaningful indications of trends. Indicators built from moving averages are not reliable indicators of trend. In short, most of the things that traditional technical analysis teaches about moving averages do not stand up to quantitative scrutiny. In most of my work, I’ve tried to stick to higher level questions and simply looked for some evidence that there is something interesting happening around moving averages. I cannot find that evidence, nor can I find it in any calculated level like Fibonaccis. When I’ve looked at VWAP on different timeframes, I find it’s no different than any other average, which is to say that price action following price engagement the moving average is random and unpredictable. Someone selling you something based on VWAP is going to give you many arguments why it should work or why it should be so, but ask them for statistical evidence. " http://adamhgrimes.com/blog/200-day-moving-average-work/ http://adamhgrimes.com/blog/reader-question-is-vwap-useful/
I agree there is no special moving average. It's pretty silly to think what 197 day is different from 200 day; but 200 day and 100 day don't produce especially different p&l. I disagree however with most of this. I think moving averages done properly still seem to be working at stock index level and on futures markets generally (although this is no guarantee for the future). [They don't work so well on individual stocks for which the intra asset class mean reversion effect dominates except for fairly long time horizons] By done properly I mean using a moving average crossover with volatility scaling (ideally with exponentially weighted MA, though this is not a deal breaker). However this effect is weak, and will not be statistically significant on a single market or crossover pair. Running a test of a single market and crossover pair might produce a negative, or a positive result, but this doesn't mean anything. You can cherry pick examples of where it has, or has not worked, to your hearts content [Indeed this is a favourite pastime of many bloggers]. You can also cherry pick an example which worked very well in the past and then stopped working, and then say all moving averages have broken, or that all moving averages are overfitted (when it's the author of the blog post that's overfitted of course). However running many tests of multiple crossovers on a large number of instruments produces a highly significant result. Hence why diversification, of a portfolio of different markets and multiple crossover lengths, is so darn important. This is a weak effect which works consistently. A small edge which on any given night on the table might not work. At an individual market level the efficiency of some pairs has faded*, however this is more than compensated for in the back tests of an entire portfolio by more markets appearing and increasing the diversification that's available. * I'm thinking here of faster pairings on equity indices GAT
In general, I get the feeling that most people here are more concerned with "proving that this or that doesn´t work", instead of dedicating the same ammount of effort to "finding something that DOES work". But hey, I´m just one guy in one internet forum, what do I know?
Adam Grimes was day trading in the 1990s, I have been for 5 years, so no this is not negative thinking leading to wrong conclusions from people with no skin in the game. The most surprising discovery in the moving average discussion for me is that VWAP is useless and that price behaves randomly around it, due to the dominance of algorithms in current markets I was expecting VWAP to have some non-random association with price.
OK... So how do you apply that "knowledge" to your trading? And I mean, in a "positive" way... Because, based on what you said, the only thing left to do is just ditch moving averages and anything based on them(since they show completely random results and thus, would be useless). That would be applying it in a "negative" way. If I were to write down here everything that (I think) doesn´t work (for me), I would spend my entire life writing things on this forum and die without completeng this work and worse yet, that would not result in my account growing larger.
I did a lot of trial and error and got a hint by some people that slow stochastics work nicely ... but it's a long way from there ... I apply them in a very specific way: multi-timeframe and with macd ... that's just one thing ... the other thing that I use is pure naked price action, again on multi timeframe in gold... that's my second ... I am working on a third thing: multi-timeframe in the dax ... there are similarities with what I am trying to do in gold but it's different nevertheless ... I'll leave the rest (details) up to you to find out
Well, my former mentor wrote that trading was much more appealing than poker because trading never felt like gambling and that day trading had no element of randomness/chance, unlike in poker. I believed that crap for years. By spreading awareness of the high degree of randomness and chance in day trading it could, ironically, lead to more success in trading, if my trading and others' is approached as a gambling business rather than a ratiocinative interpretation of price levels there could be more profit resulting from less holding onto losing trades.
there is much less randomness then most people think in daytrading if you apply the correct multi-timeframes for the appropiate instrument ... it makes no sense to use the same tf's for gold as for the dax for example ... however there are plenty of instruments I don't understand whatsoever ... only gold and dax are clear to me intraday ...