StockFetcher Forums · General Discussion · tau Zweig’s Genetic Programming Engine<< >>Post Follow-up
ham1198
169 posts
msg #42555
Ignore ham1198
3/27/2006 8:10:41 AM

http://www.tau-zweig.com/stockfetcher_GP_code.html

Has anyone ever heard of this b/4? they say nice things about stockfetcher. they mention stochastic and macd are the 32 best ta indicators.

in regard to their first filter with the naz100, i plugged it in and came up with 96 hits. thanks, that's almost everyone.

any thoughts on their filters?


heyen
124 posts
msg #42556
Ignore heyen
3/27/2006 8:57:59 AM

Yes, but look how restrictive this filter ended up.
Its a more random try/error method. Doesnt follow stock evolution.
I didnt find the filter that well performing in my backtests either.

The idea isnt new - let a dumb machine search for the perfect
technical filter. Is there any? If so, let me know, or post it here...


markcrisp
187 posts
msg #42576
Ignore markcrisp
3/28/2006 3:44:04 AM

Amazing.....it's funny how complicated some people make trading.

Look at TIE, RACK and HANS..simple Momentum stocks make me lots of $$'s..NO indicators EVER. I think 99.99% of T.A. is a complete waste of time...

Some people spend more time/money and effort trying to "guess" the future. I simply follow and make $$'s. It's your time I guess.


shaikhhali
42 posts
msg #42578
Ignore shaikhhali
3/28/2006 6:05:22 AM

Trading is a science and Art, and because of that there must be human touch in selecting the wining stock.

Therefor i do not think that this software is capable of replacing human mind





maxreturn
745 posts
msg #42582
Ignore maxreturn
3/28/2006 7:52:15 AM

Markcrisp, couldn't agree with you more. I have been trading for 15+ years now and tried everything under the sun, including fancy, complicated filters on stockfetcher but my best results come from buying stocks that are strongly trending and making orderly pullbacks towards strongly rising weekly 10 period ema which is in turn above an upsloping weekly 40 period ema. I also want to see steadily rising RS and check IBD to cover the fundamentals. KISS!


yepher
353 posts
msg #42584
Ignore yepher
3/28/2006 8:26:23 AM

ham1198,

Incase you have not already done it you can search this forum for "TAU" and see the discussions he started here and what the results were.

I personally disagreed with his results. He and I sent some emails back and forth for a while and I think he too did not agree that his test supported his claims.

markcrisp,

I understand your point about following the trends and agree with it. That is a fine way to make money. But I must say I completly disagree with a statment like "...complete waste of time.." in relation to research new techniques for market prediction (which is the same thing you are doing with trend following by the way). I feel there is much research and discover still to do in the field of computational finance. We have only scratched the surface of what is possible. I also have high hopes for genetic algorythmns, Hidden Markov Models, and the like.

I am now stepping down off soap box ;)
-- Yepher



TheRumpledOne
6,358 posts
msg #42583
Ignore TheRumpledOne
modified
3/28/2006 8:27:34 AM

Fetcher[

/* STRONG TREND WITH EMA DISPLAY */

/* ENHANCED TREND DISPLAY */

set{T10, count(10 day slope of the close above 0,1)}
set{T40, count(40 day slope of the close above 0,1)}
set{T200, count(200 day slope of the close above 0,1)}

Set{a1, T200 * 1}
Set{a2, T40 * 10}
Set{a3, T10 * 100}

Set{aa, a1 + a2}
Set{TREND, aa + a3}
set{T10a,days(10 day slope of the close below 0, 100)}
set{T40a,days(40 day slope of the close below 0, 100)}
set{T200a,days(200 day slope of the close below 0, 100)}


/* EMA DISPLAY */

set{ MinEMA, min( weekly ema(10), weekly ema(40) ) }
set{ MaxEMA, max( weekly ema(10), weekly ema(40) ) }
SET{ FilterEMA, weekly ema(5) }

set{ GoLong, count( weekly ema(5) crossed above MaXEMA , 1 ) }

set{E5b,days(close is above weekly ema(10),100)}
set{E5a,days(close is below weekly ema(10),100)}
set{CxE10, E5a - E5b}

set{E5bb,days( weekly ema(10) is above weekly ema(10) 1 day ago,100)}
set{E5ab,days( weekly ema(10) is below weekly ema(10) 1 day ago,100)}
set{E10xE10, E5ab - E5bb}

set{C_E10 , close - weekly ema(10) }

draw weekly ema(5)
draw weekly ema(10)
draw weekly ema(40)


ADD COLUMN C_E10
ADD COLUMN TREND

ADD COLUMN T10a
ADD COLUMN T40a
ADD COLUMN T200a

and add column GoLong
and add column CxE10
and add column E10xE10

draw T10a
draw T40a on plot t10a
draw T200a on plot t10a

/* enter your Upper Limit criteria */
set{UpperLim, High 26 week High}

/* enter your Lower Limit criteria */
set{LowerLim, Low 26 week Low}

set{LimDiff, UpperLim minus LowerLim}
set{PPDiff, CLOSE minus LowerLim}
set{PPDiv, PPDiff / LimDiff}
set{BallOn, PPDiv * 100}

set{PCT, LimDiff / LowerLim}

set{HiOp,high - open}
set{OpLo, open - low}

set{CCb,days(close is above close 1 day ago,100)}
set{CCa,days(close is below close 1 day ago,100)}
set{ClxCl, CCa - CCb}
/*
set{HHb,days(High is above High 1 day ago,100)}
set{HHa,days(High is below High 1 day ago,100)}
set{HixHi, HHa - HHb}

set{LLb,days(Low is above Low 1 day ago,100)}
set{LLa,days(Low is below Low 1 day ago,100)}
set{LoxLo, LLa - LLb}
*/
set{VolUp, days(volume is below volume 1 day ago,100)}
set{VolDn, days(volume is above volume 1 day ago,100)}
set{VlXvl, VolUp - VolDn}

set{vck1, volume 1 day ago }
set{vck, volume / vck1 }
set{vdbl, days(vck < 2, 100)}

Set{Bullpower, high minus weekly ema(13)}
Set{Bearpower, low minus weekly ema(13)}

set{BLb,days(BullPower is above BullPower 1 day ago,100)}
set{BLa,days(BullPower is below BullPower 1 day ago,100)}
set{BLxBL, BLa - BLb}

set{BRb,days(BearPower is above BearPower 1 day ago,100)}
set{BRa,days(BearPower is below BearPower 1 day ago,100)}
set{BRxBR, BRa - BRb}

set{RChg, count( UpperLim equal UpperLim 1 day ago, 1 ) }
set{SChg, count( LowerLim equal LowerLim 1 day ago, 1 ) }

/* Volatility Direction Filter Display */

set{BBDiff, Upper Bollinger Band(20) - Lower Bollinger Band(20) }

set{BBDb,days(BBDiff is above BBDiff 1 day ago,100)}
set{BBDa,days(BBDiff is below BBDiff 1 day ago,100)}
set{BBTrend, BBDa - BBDb}

set{ATRb,days(ATR(14) is above ATR(14) 1 day ago,100)}
set{ATRa,days(ATR(14) is below ATR(14) 1 day ago,100)}
set{ATRTrend, ATRa - ATRb}

/* profit had you bought the open 5 days ago and sold at the high */

set{High5, High 5 day High}
set{Profit, High5 - open 5 days ago}

/* Narrow Channel Breakout Display */

set{ High20, high 20 day high}
set{ Low20, low 20 day low}

set{ New20High, count( high above High20 1 day ago , 1 ) }
set{ New20Low, count( low below Low20 1 day ago , 1 ) }

/* column display */

add column HiOp
add column OpLo
add column Profit

add column BallOn
add column UpperLim {Resist}
add column LowerLim {Support}

add column RChg
add column SChg

add column New20High
add column New20Low

AND ADD COLUMN BBTrend
AND ADD COLUMN ATRTrend

/* PLOTS */

draw UpperLim on plot price
draw LowerLim on plot price

/* NORMALIZE accumulation distribution */

set{adval, INDPOSITION(accumulation distribution, 40) * 100 }

/* NORMALIZE momentum(12) */

set{moval, INDPOSITION(momentum(12), 40) * 100}

/* NORMALIZE rsi(2) */

set{rsval, RSI(2) }

/* NORMALIZE williams %r(10) */

set{wrval, INDPOSITION(williams %r(10), 40) * 100 }

/* DRAW THE MP RSIWLR CHART */

and draw wrval
and draw adval on plot wrval
and draw moval on plot wrval
and draw rsval on plot wrval

add column wrval
add column adval
add column rsval
add column moval

and add column clxcl
and add column VlXvl
and add column Vdbl
/*
add column HixHi
add column LoxLo

add column Bullpower
add column Bearpower
add column BLxBL
add column BRxBR
*/

/* Selection Criteria */

trend equal 111

WEEKLY RSI(10) increasing for 2 weeks

weekly ema(10) above weekly ema(40)
weekly ema(10) increasing for 2 weeks
weekly ema(40) increasing for 2 weeks

avg vol(5) above 1000000
avg vol(30) above 1000000

add column industry

Sort column 5 ascending
]



C_E10 is close minus the weekly ema(10). Sorting on this column so you can see stocks near their weekly ema(10).

MAY ALL YOUR FILLS BE COMPLETE.






TheRumpledOne
6,358 posts
msg #42585
Ignore TheRumpledOne
3/28/2006 8:33:33 AM

I think Mark's point is you don't have to do anything fancy to make money in the market.

Learning, practicing and mastering the basic fundamentals of trading and money management is all you really need.

On the other hand, using a computer to sift through thousands a stocks decreases the amount of time you have to spend looking for stocks to trade.

So long as you don't confuse stock SELECTION with stock TRADING, you'll be fine.

It only takes ONE STOCK to make you rich if you trade well.




yepher
353 posts
msg #42587
Ignore yepher
3/28/2006 9:09:42 AM

TRO,

I was not disputing that point. I agree with the philosphy that you don't have to do anything fancy to make money in the stock market.

But from markcrip's post he stated just follow trends like TIE... and make lots of money. TIE has special interest to me in that I too was trading this the first part of the year and exited the stock just before the big gap down (I would call it luck that I got out and nothing else).

On another note "fancy" is higly relative term and I am sure there are many who would believe an Exponetial Moving Average (or maybe RSI is better example :) is fancy and complex. Imagine the arguements when John Murphy first stared doing technical analysis. As you encapsulate and hide this complexity it become considered simple, tride and true. The same is possible with the more complex learning systems.

A more concrete example: Many consider face recognition and background recognition to be far to complex to have everyday usability. Check this out: http://www.riya.com/ I have tested it and it works well. It even recognized a picture of Big Ben in the background of an image and put it on the map without being told anything about the image. I personally believe similar things are possible with stocks price/volume pattern. With proper trianing a system should be able to learn that TIE is a failing trend. Maybe there is an indicator (albieit fancy in our current frame of thinking) that can tell you TIE is about to get chopped off at the knees.

My $.02
-- Yepher





TheRumpledOne
6,358 posts
msg #42590
Ignore TheRumpledOne
3/28/2006 10:53:47 AM

And I wasn't disputing what you said.


StockFetcher Forums · General Discussion · tau Zweig’s Genetic Programming Engine<< >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus