**************************************************** OK, yes, point taken. You are evidently aware of all this and I don't mean to lecture, but many traders, especially new ones are unaware of this and lose money trying to trade systems based on lagging indicators because they look good on past charts. As for the psychology part all I can say is that I wrestle with it big time. I'm going to put up a sign on my monitor that says ..."No Fear, No Greed...... Patience." ***************************** "When I read about the evils of drinking, I gave up reading." - Henny Youngman
********************************** You know I probably should not have spouted off on this topic because frankly I'm a little fuzzy on it myself. It does seem, as the others here have pointed out, that once the bar is closed the point is fixed for the curve. The point is, after all, determined by past action so future action should not effect it. But it seems that I've watched charts develop and seen the curve and crossovers change with subsequent price action after the bar closes. But I could be wrong. I was once before. ********************************************* "It was irritating to have one's physical shortcomings pointed out quite so plainly twice in one evening, once by a beautiful girl and once by a dying badger." - Tom Holt, Expecting Someone Taller
Back testing is hard to do at best. The use of closed bars is just flim flam and denial level of backtesting. In making money there is a timing aspect. If a person has a strategy of using closed bars for trading he is a simpleton at best. If you look at signals from indicators almost every action taken is done from a signal in time while a bar is forming. Almost all signals and alarms are created during bar being formed. Using closed bars for back testing is merely a humorous and misleading head in the sand childish thing.
Really I just use what makes money. I've been making money so i'm happy with the method. No one has access to information from the future. A "closed bar" represents a user defined end to a sampling interval. So if you really believe what you've said, try this: (a)Do your backtest using open bars. You are a sophisticated and wise backtester. (b)Take the same data, but now increase the sampling rate (use smaller time interval bars). Oh my goodness! It appears now you've made decisions based on "closed bars". In your own words, "head in the sand childish," etc. It appears we have a paradox on our hands. Which is it??? A or B???? Also, I lied. I DO have access to information from the future. I know what you are going to say next. "You can't tell when you'll get the signal; if you increase the sampling rate so that one of my signals comes right after a closed bar, most of my other signals will come during open bars." Good point. Which brings us to the heart of the matter: Your sampling rate needs to be much higher than your decision rate for a good backtest if you are making decisions based on closed bars. Which for me it is, with 30 trades on month on 5 minute bars. A scalper would have problems using closed bars for signals. I'll bet you are a scalper, Jack?
This is really a question about slippage not Sto or MACD. Just use an estimate. On an end-of-day system you can get in on the next days open plus costs & slippage. Are you talking about Divergence? Sto does not give a crossover signal in isolation. Combined with chart patterns and cycles, Momentum oscillators do have predictive value.
Sorry. I pulled the post to which you refer and put it over on the sma thread. ....I like the sign It is true that estimates can be used to guess the price one would have gotten, but most systems don't have much profit margin and wrong guess here or there can really throw things off. Besides, I could never make ma's work for me so they must be stupid and ineffectual. Or is it the other way around? *********************** "Delusions of grandeur make me feel a lot better about myself." - Jane Wagner
Well, you can backtest but it won't be accurate. Or perhaps there are some methods that would allow it but what I'm saying is this: Say you're testing a moving average crossover system that gives a signal to buy after completion of a bar and a ma crossover. Looking back on it, one can't always be sure which bar completion initiated the crossover at the time, therefore you can't be sure of what price you would have gotten. I suppose with special software that factors in rate of change during the crossover period one might get a better handle on it..... But I'll admit that I threw out a blanket statement that may not be entirely correct in all instances.