Im new at technical analysis and even though i have read alot of stuff about it i still dont understand some things , all i see people using are indicators like moving average and oscillators etc , i have yet to see anyone use a basic trendline analysis or price patterns as the core strategy , i also notice that its really hard to code those in order to backtest them , no backtesting software includes code for these strategies , whats the purpose of these things? cant you use them as stand alone strategies?
How important are basic trendline and price pattern analysis?
- Thread starter Wach80
- Start date
Big Fat Yes
Your kind of all over the place. I promise a few months from now and you'll look back and go " what was I asking?" .
The reason you won't see head and shoulders/trend/ support/resistance line as a quick and drag function in a backtest software is their really discretionary from trader to trader.
A backtesting software works of mathematical/computer code logic where you the coder, will define if "a" happens then look for "b" and buy at "c", sell if "d" happens. This is just random talk but it's all it is, it's like basic algebra, buy = a+b.
Now he beautiful part about a backtesting software/coding is that you can go across very large databases with very complex rules ( that would take years to manually calculate), get a trade list, then review that trade list and analyze it with whatever formulas you want. What was the w/l ratio? What was the sharpe ratio? What was the max consecutive losses?
What your looking to do, by reading your previous post about Amibroker with drawing a trend line and having it backtest...... Is have a bunch of discretionary rules( you drawing a trend line) and then somehow have a computer backtest your system( which has no rules, just you drawing stuff with forward bias most likely).
I'm still lost to what your trying to achieve and seem like your confused of what backtesting actually is. You could continue manually backtesting, with a large enough sample base, you can justify having a system to go forward with.
In my mind it's no different then a coded backtest, were always assuming that past results will reflect future results and their are ways using a system with hard coded rules to test out of sample and with other statistic tools to build some assumptions on the system.
Issue with backtesting manually/ discretionary is their are no hard rules like code, if future results don't look like past results, either your intuition, or method isn't in sync with the market and you'll find yourself wondering if your method works.
Then you'll find people like Linda Raeshky or Adam Grimes that will do lots of backtesting/ modelling and apply discretion to their models.
Slow down with the posting, I'll bet you'll find more success by doing your own homework and reading.
The reason you won't see head and shoulders/trend/ support/resistance line as a quick and drag function in a backtest software is their really discretionary from trader to trader.
A backtesting software works of mathematical/computer code logic where you the coder, will define if "a" happens then look for "b" and buy at "c", sell if "d" happens. This is just random talk but it's all it is, it's like basic algebra, buy = a+b.
Now he beautiful part about a backtesting software/coding is that you can go across very large databases with very complex rules ( that would take years to manually calculate), get a trade list, then review that trade list and analyze it with whatever formulas you want. What was the w/l ratio? What was the sharpe ratio? What was the max consecutive losses?
What your looking to do, by reading your previous post about Amibroker with drawing a trend line and having it backtest...... Is have a bunch of discretionary rules( you drawing a trend line) and then somehow have a computer backtest your system( which has no rules, just you drawing stuff with forward bias most likely).
I'm still lost to what your trying to achieve and seem like your confused of what backtesting actually is. You could continue manually backtesting, with a large enough sample base, you can justify having a system to go forward with.
In my mind it's no different then a coded backtest, were always assuming that past results will reflect future results and their are ways using a system with hard coded rules to test out of sample and with other statistic tools to build some assumptions on the system.
Issue with backtesting manually/ discretionary is their are no hard rules like code, if future results don't look like past results, either your intuition, or method isn't in sync with the market and you'll find yourself wondering if your method works.
Then you'll find people like Linda Raeshky or Adam Grimes that will do lots of backtesting/ modelling and apply discretion to their models.
Slow down with the posting, I'll bet you'll find more success by doing your own homework and reading.
Good to see this thread started off on the right track.
To just reinforce what's been said,
Answer: Yes, they all will work. Patterns as a stand alone take some time.
Moving averages, trendlines, and patterns are what I use. Fibonacci angles and retracements/failures will work too but it took some effort. I wouldn't personally mess with much else.
To just reinforce what's been said,
Answer: Yes, they all will work. Patterns as a stand alone take some time.
Moving averages, trendlines, and patterns are what I use. Fibonacci angles and retracements/failures will work too but it took some effort. I wouldn't personally mess with much else.
Therefore if i understand correctly what you are saying is that software cannot backtest tl or price pattern analysis since those are discretionary unlike indicators who are always specific , thats ok but if we dont backtest a strategy then dont we lose alot of prestige about it?
You also make it sound like that on rare ocassions people backtest discretion which makes me think its not the most standard thing in the world , in that case ,if we plan to use software do we backtest the specific stuff only (thats the core of our strategy) and we add discretion as an extra bonus to our strategy but sadly not tested?
Btw i search about manual backtesting , it sounds very time consuming but not neccesarily hard , you gotta scroll the chart and mark same setups as yours , is that right?
As for software we cant of course write draw trendline as a rule but it should be possible to give specific metrics to a code as to where the tl is or not? cant we for example create a h&s into a code by giving metrics as to where the support and resistance is and then write a crossover code if price surpass the resistance point , well it sounds extremely complicated , i dont know if its even possible to write a whole chart or a specific part of it as a code.
Last edited:
When you are saying dynamics you mean moving average and oscillators right? from what i read those monitor the momentum of a price , if thats right then you say that TA without indicators is really bad.
No. When I say Dynamics it means "Price Action". The very thing that matters. The underlying. MA, Oscillators, Indicators, Patterns & Co are Framework. It's another channel of information. It's all about last prices. This is the source of everything. What I said is that whatever tools you use. DO NOT FORGET TO TRADE THE UNDERLYING ! Since it can breaks every tool you use. Study the behavior of the price. And after use whatever tool you want for filtering or adding redundancy. But PA is the base.