Have you found anything that works through backtesting? Linda Raschke who has done an assload of backtesting said she has only met two traders in 20 some odd years of trading who made aliving trading an automated system which as I see it is the only kind of system you can come up with backtesting as most methods employed by traders use some descretion and there is no way to test descretionary elements is there? p.s. It has brought an assload of new traders into the market so in that respect it is certainly good.
I don't know what you mean by "discretionary elements". If that means that a trader buys and sells on a "gut feel", then you are right, you can't back test this. Pretty much everything else other than a gut feel can be tested. In regards to your Linda Raschke argument, I'd point out that automated trading is a fairly new paradigm which didn't exist in much of Linda Raschke's time. So suppose she knows 10 futures traders who use exclusively automated systems, and 2 of them make money. But it's already a better ratio than the ratio of 90 losers to 10 winners among the "discretional" futures traders. Without knowing the population size of automated traders that Linda has known, two is a meaningless number.
So have you found anything that works? p.s. That was a fairly recent comment by her and Id be willing to bet she is still on the cutting edge of whatever is going on.
Actually you backested the system without even knowing how it works, so we can't really say that your backtest counts for that system, can we? *** For a computerized system to work Bear, someone would have to know how to trade extremely well and code extremely well, and that's about a rare combination as you can get, that's probably why Linda could only find those examples. Good trading, JJ
Don't let the lack of bactesting software keep you from doing this, eventually you'll definitely come up with a few ideas that work (if you haven't already).[i/] JJ
I backtested it precisely as specified in the original post of that thread, and I also tried a few variations, as suggested to me. If you have an alternative interpretation of that system description, I'd be happy to test it as well.
See, you are back to your train of thought. Look, whether I found anything that "works" doesn't change anything about the merits of back testing. Think of it this way: it works 100% for eliminating the systems that do not work.
PEOPLE! Automated trading doesn't work unless it has some sort of artificial intelligente that would handle moneymanagement in it's most complex form... And when you have that, congratulations, you have just created an android ) Taking a more serious approach, there is a way of backtesting discretionary trading, and that is done in a "training" kinf of fashion .... You get a chart vendor that can give you a huge database of historical data and you click forward one bar at a time. And there you go, do that for a couple of years and nobody can beat you at your profession, provided you have recorded every trade in detail in a journal and then, after each 10 trade set, go back to the journal, review them and make sure you avoid the mistakes you made the next 10 trades and so on .... Real life fast-forward! My 2 cents, it's the only way you can call yourself a "trader" and it's the only way you can gain an edge! He who trades automated systems is not a trader! Behold
agree with you 100%. I would just add that there is no simple purely mechanical system which works under any market condition . Discretion is an inevitable part of a profitable trading .
I'm not saying backtesting doesnt work as Im not very knowledgeable in that area. I backtested, manually, this system by just going back over charts for a number of days entering and exiting on a close above/below the 3sma+2 and got quite excited because the results were very good but when I tried to trade it live it did not work. The reason it did not work is that with live data you often have to make decisions intrabar. Looking backward you do not have this problem because the bar is complete. This is why I have lost faith in any kind of looking back at historical charts. Trader28 says he has been making money with this system for a couple of years but you have eliminated it by your backtesting as a way to make money. I have a feeling that backtests keeps a lot of people out of systems that work with some practice.