Technical indicators are they any good?

funky · Aug 20, 2003

Quote from vanilla2:

I think you could achieve this in backtesting by placing a simple condition in front of all of your strategy exits. For example if you were trading on 10 minute bars, and your target timeframe was 3 hour trades, each exit would only be triggered if bars since entry was > 18. Then if you optimize for the entry indicator length, your result will be optimal for the given timeframe constraint.

good points v....i think, however, that there isn't going to be any ideal setting above all others. its the feel of the market more than anything. i can drive my junky old car just as good as someone with a porsche -- why? because i know how my car feels the road, and how to react to it. its all the same road -- just a different reaction needed by me vs. the guy with the porsche.

to all of you new traders: just pick some reasonable settings and watch price along side your indicators. remember, its PRICE that you are trading, above all else.

i need another drink.....

OddTrader · Aug 20, 2003

Quote from funky:

good points v....i think, however, that there isn't going to be any ideal setting above all others. its the feel of the market more than anything. i can drive my junky old car just as good as someone with a porsche -- why? because i know how my car feels the road, and how to react to it. its all the same road -- just a different reaction needed by me vs. the guy with the porsche.

to all of you new traders: just pick some reasonable settings and watch price along side your indicators. remember, its PRICE that you are trading, above all else.

i need another drink.....

Very well said, Funky.

(Don't mean another drink, though.

)

Bolts · Aug 20, 2003

Quote from vanilla2:

I think you could achieve this in backtesting by placing a simple condition in front of all of your strategy exits. For example if you were trading on 10 minute bars, and your target timeframe was 3 hour trades, each exit would only be triggered if bars since entry was > 18. Then if you optimize for the entry indicator length, your result will be optimal for the given timeframe constraint.

Well yeah, like I said, you could backtest it, but I just wonder if there could be some mathematically expressed theory that could explain that optimal indicator period every time.

OddTrader · Aug 20, 2003

Quote from Bolts:

Well yeah, like I said, you could backtest it, but I just wonder if there could be some mathematically expressed theory that could explain that optimal indicator period every time.

Another phd thesis!?

vanilla2 · Aug 20, 2003

Quote from Bolts:

Well yeah, like I said, you could backtest it, but I just wonder if there could be some mathematically expressed theory that could explain that optimal indicator period every time.

help me funky brother, i think i'm missing the idea here. is it like what bruce lee said to his student in enter the dragon, "don't THINK, feeeeeel"? Or is it more of a belief that all statistically significant relationships based on historical indicator settings are a curve fit?

not to spark the discretionary vs mechanical debate in every corner of ET, but it sounds to me like the mathmatical relationship you're looking for is the specific one between a maximized profit outcome and an indicator period variable over a given sample of price history. i cant think of a way to generalize this relationship, since every indicator is different.

as a mechanical trader, i've never understood why it would be preferable to trade a perception or a guess at what your optimal variables settings are, as opposed to a statistically significant, probabilistically favorable backtested setting. Within reason, distributions of your variable setting's historical results can tell you a lot more about what's likely to happen than a gander based on what you ate for lunch, how much sleep you got, or what your p&l has been doing lately.

i haven't been following this thread, so my 2 cents could be completely missing the point. a drink would be nice.

OddTrader · Aug 20, 2003

Quote from vanilla2:

help me funky brother, i think i'm missing the idea here. is it like what bruce lee said to his student in enter the dragon, "don't THINK, feeeeeel"?

not to spark the discretionary vs mechanical debate in every corner of ET, but it sounds to me like the mathmatical relationship you're looking for is the specific one between a maximized profit outcome and an indicator period variable over a given sample of price history. i cant think of a way to generalize this relationship, since every indicator is different.

as a mechanical trader, i've never understood why it would be preferable to trade a perception or a guess at what your optimal variables settings are, as opposed to a statistically significant, probabilistically favorable backtested setting. Within reason, distributions of your variable setting's historical results can tell you a lot more about what's likely to happen than a gander based on what you ate for lunch, how much sleep you got, or what your p&l has been doing lately.

i haven't been following this thread, so my 2 cents could be completely missing the point. a drink would be nice.

Another part of holy grail is revealed

. Thanks v.

vanilla2 · Aug 20, 2003

Quote from OddTrader:

Another part of holy grail is revealed . Thanks v.

I wish

I assume you mean the Bruce Lee wisdom. He would have been a great trader. Assuming he didn't bust up the screen on every stop out.

I forgot to mention that my personal reliance on historical distributions of strategy variables (like indicator periods etc), is based on agnosticism about where price will go. If you have a strong belief that you see a setup forming, and discretion weighs into your trades, than trusting a backtested result is painful or impossible.

I'll can it now before I pollute this thread with off topic stuff. Hope this is helpful in any way.

peg_leg_joe · Aug 20, 2003

Traders can find success with discipline and a methodology that gives them an edge. For some this means technical indicators. For others, it means looking for price patterns and volume. I personally do not use any technical indicators outside of pivots/S/R, trendlines, and volume. Some people probably wouldn't call these things technical indicators at all. I also use patterns such as flags, triangles, double tops, double bottoms, and head-and-shoulders. The key is you've got to find a methodology that works for YOU. What works for me or Joe Blow won't necessarily work for you and vice versa. I will close my post with this though: for over a year I tried to use technical indicators to succeed at trading. It wasn't until I started using patterns and volume that I started to become consistently profitable.

OddTrader · Aug 20, 2003

Quote from BSAM:

iraj......

Read this afternoon's chat log with Rich Kenny. One of the most important things he points out is near the first part of the chat: "The beauty of technical analysis is that all of the indicators I have mentioned work some of the time----and none work all of the time."

To answer your question, YES technical indicators are good, but as electron points out, "They are only as good as the trader who uses them."

BSAM

The only thing never changes is change.

Bolts · Aug 20, 2003

Quote from vanilla2:

help me funky brother, i think i'm missing the idea here. is it like what bruce lee said to his student in enter the dragon, "don't THINK, feeeeeel"? Or is it more of a belief that all statistically significant relationships based on historical indicator settings are a curve fit?

not to spark the discretionary vs mechanical debate in every corner of ET, but it sounds to me like the mathmatical relationship you're looking for is the specific one between a maximized profit outcome and an indicator period variable over a given sample of price history. i cant think of a way to generalize this relationship, since every indicator is different.

as a mechanical trader, i've never understood why it would be preferable to trade a perception or a guess at what your optimal variables settings are, as opposed to a statistically significant, probabilistically favorable backtested setting. Within reason, distributions of your variable setting's historical results can tell you a lot more about what's likely to happen than a gander based on what you ate for lunch, how much sleep you got, or what your p&l has been doing lately.

i haven't been following this thread, so my 2 cents could be completely missing the point. a drink would be nice.

Yes I had a sort of curve fit in mind, position exit time as some function of indicator period time. To clarify, there would be a completely different function for each indicator of course. But I see your point about backtesting being better than any theory could be. So I guess it wouldn't be of much use to most traders, but maybe it would be useful for trading something that could not be backtested well, like IPO's for example.