No. neither am I !
But I would put to you that whatever it is that you see is a product of your own interpretation of what there is to see. After all, we are all looking at the same thing.
Part of my trading methodology, involves taking completely random trading decisions. These crazy trades result in a distribution of gains and losses. In addition, I also take trades based on technical and statistical analysis, and these trades also result in a distribution of gains and losses.
If you compare the distributions from both approaches, they are different, and the difference is statistically significant. The difference are not imaginary, and the statistics suggest that they are not due to random chance. The difference between the means are statistically significant, and the shape of the distributions in returns are completely different.
Experience shows me that I can profit by trading with my TA signal, but more importantly I can equally well make a loss by trading against that signal. A key part of the ongoing development of my strategy is to continually demo trade fading the TA signal.
You might argue that gains may be due to good luck (its an argument that I've made), but the probability of simultaneously incurring losses from fading TA signals on other instruments and timeframes, on the basis of chance, surely has to be quite astranomical (although I concede its possible).
I would therefore argue that there is no imagination involved. I have a bunch of equity curves with positive gradients based on TA, and a bunch of equity cures with negative gradients based on trading against the TA, but the evidence for TA is actually much stronger.
The TA that I use, in common with most of the stuff out there has adjustable parameters. Some weeks parameter X may provide the optimum results, the week after parameter Y, and the week after parameter T etc. The returns between the best and worst can typically differ by a factor of 2. For the last 3 years, I've been banging my head against the problem of determining methods for self optimisation of these parameters.
Until this particular problem is solved, I'm forced to trade a larger number of small diversified positions and to take the rough with the smooth.
You are not going to like this one little bit, but if parameter X results in the best result over a given time period, then fading parameter X in the majority of cases results in the worst results over the same period. I'd argue thats quite compelling evidence.
Before anyone says, well of course, the opposite to a winning system will be a losing system, I need to make things clear. If I get a signal to buy EURUSD, the systems that are fading the TA signals do not necessarily take a sell trade, it happens on occassions, but its actually quite rare. They are totally independant systems that are running out of phase, they may already be in another trade, or they may be in a dormant state as they are programmed to randomly activate and deactivate throughout the day.
The other issue is that all of the elements of my systems contribute to my edge, so even in the case where you fade a TA signal, the other system components tend to compensate, and the system performs better than might be expected.
However, If I strip systems down to their most basic elements, the equity curves from the TA signal, v fading the TA signal are practically mirror images, one goes up, the other goes down.
I've been trading this method since 2005, I have tens of thousands of trades, and more statistics than you can shake a stick at. I use some pretty clever proprietry indicators, but only because I spent a great deal of time and money devloping them, to be brutally honest I can get similar (if not better) results from practically any indicator that I've tried. If you think about it, that really shouldnt come as a surpise.
I've had arguments with the "all is known in advance crowd", Mr Socco, Starspacer, Mr Marcus etc. I've argued that its possible that there is no cause and effect, and that a profitable traders results could be due to luck, and intellectually, I understand that allthough very improbable, it is a possibility. It was entirely due to these issues that I took a far more statistically rigorous approach to my trading.
There's loads of evidence I could present, I'm only scratching the surface as most of what I do is proprietry. Its one of those things you cant prove or disprove without revealling full details of the methodology, but anyone whose done a bit of basic research into system design, probably knows enough to realise that TA can provide a mechanical edge.
Everyone goes wrong by assumming TA is predictive and it isnt.