well.. i guess that analysing the whole picture is still needed to prevent going long at the top, even if there is a good OI bar setup.
I never meant to present a setup that'll work every time even on a particularly positive time frame... to me, its always needed to evaluate the market structure prior to trading.
So far, almost every breakout caught enough points to compensate stopouts, even if price comes back after that trigger bar is closed.
But then again, when you backtested this, it seemed like very positive results... even if automated strategies enter blindly in the market (disregarding its position inside channels, tops or bottoms...)
Is there a way that i can evaluate if this strategy is profitable for discrectionary trading (ie checking the market structure before considering entering long or short), based on historical data performance?
Or any way to add this market position analysis to the automated strategy? Something like, when a certain MA is blue, only consider longs...
What do you think?
(im thinking this works better on UNUSUAL time frames that people arent looking at... thus the weird number) But you still have to figure what works best for you: more setups per week = more fakes.)
I backtest ideas and then if useful I incorporate them into my trading. So this is possible. You just have to do a ton of testing by hand if it's not automated. I do this as well. Sometimes 6-8 hours/day just doing manual backtests on a chart, bar-by-bar (covering up the future) and keeping results in excel. I change a parameter I start over. I pick the most recent 1-2 months and do it. and when I'm confident I trade it. and today for example, trading CL in the afternoon nothing worked. but I know over the past 2 months it has worked well so i will shake it off and continue tomorrow.
automating can only do so much. not enough IMHO and that's why I don't spend a lot of time on it. I've spend hundreds of hours on a strategy only to watch it fall apart after a couple months trading live. So I keep my investment small.
I test things like avg move from a pattern, best target & stop, days of the week, times of day, etc. but nothing is profitable by itself so it's just to give me ideas and maybe a bias on the market.
really I hadn't done any strategies until lately. I went months without it.
I do program indicators to color bars and detect patterns, but nothing with the mathematical algorithsm.
and filtering with moving averages doesn't work cause they're lagging. in my own trading I'm often in a trade before the MA turns or crosses. price is faster than an indicator.
I know you didn't present the outside bar breakout as an automated system and it's totally possible you can trade it profitably by adding a bit of discretion. This is why I don't put a lot of faith in backtest results. if the backtest is positive though I know there is potential. if it's negative really bad then I try something else. if a strategy is near breakeven I can usually trade it profitably with discretion.
Jeffs dots will add a dot when its moving average turns. I don't use moving averages, they lag too much. once you can read price action, price will tell you all you need to know.
for the development, I really don't have time. you can ask BigMike or Roonius, they will do it for a small fee. or post your idea in the programmer's paradise forum and if the idea is worthwhile someone may program it for you.
forget dots & moving averages. learn to trade price. see the Ross Hook thread, it's a good approach, very similar to what I use..
YES! I have a volume indicator that you can read about in my volume thread. Other than that, no indicators. No need. Price is making a HH and I want to be long or a LL and I want to be short or neither and I want to be flat. Trading really is that simple. But simple is not easy.
Sorry for a bit late and not complete reply, I will write, hope tomorrow more.
Cunparis is right by all means.
However time which you've picked could have some sense (about 4 and 8 hours respectively) but ...
what you are doing is called "curve fitting", i.e. fitting results into rules or vice verse, always don't remember how correct
That style of research is also used, but for preliminary conclusions, then you have to test all on significant historical data, but on different market conditions seasonalities, not less than 1 year
For example, what you saw can work, only on sleepy market during holidays only, or that could work only during winter or summer or on new bull or on old bear or when market on last cat bounces when market made already 70 % profits or before that
All that is in line with Cunparis told, but that's true when creating any type strategy mechanical, bot, other you should be very careful and trade it only when it checked properly and seriously otherwise can happen following :
strategy working from the very beginning, but then no and a lot of money could be lost
or even worse strategy working randomly
so, the one having it could be caught and don't understand what is going on and what to do exactly to solve that or pass through. the only carefully historical research helps, but not in all cases as past is not indicative for future, otherwise historicans could be best traders
what I wanted to tell when sent links to simple volume indicators and what I also noted on Cunparis chart.
Look at meaningful levels on normal time frame, watch want market doing, usual colume and usual HH, LL, etc can tell a lot.
Those things which is really working is very simple, that's why I've paid attention to your 2 lines and suggested to think together about that further, then added to that simple volume.
When one able to "see" market that's the only thing needed to trade.