Welcome Guest, please sign in to participate in a discussion. | Search | Active Topics | |
Registered User Joined: 6/30/2017 Posts: 1,227
|
Hi Bruce,
I found a "Simple S&P 500 Swing Trading Strategy" video that's only about 11 minutes long. It's not hyperbole, the presenter really isn't selling anything, he includes all the simple rules and parameters, and tests using 25 years of S&P 500 data. There were only about 3 trades per year on a daily timeframe but if we can TC2000-ify this I can also trade the QQQ's, stocks, other timeframes, etc. It's a pretty generic system.
The only thing is, I don't recall if TC2000 can handle Standard Deviation Channels. If not, we're screwed. :) ... if we can, then the only other things we need are Bollinger Bands and Candlesticks. Nothing a PCF Ninja can't handle.
If TC2000 can do Standard Deviation Channels, then the (it really is) simple system rules are in the video (I think) and if they're not all there they are in the text article.
The equity curve looks good. I figure if I can set this up in TC2000 I can paper trade, and then start drafting my plan for total world domination. :)
|
|
Registered User Joined: 9/17/2010 Posts: 484
|
I beleive the Regression Channel feature gives you the Standard Deviation channel you want, but don't think it can be referenced in a PCF
|
|
Registered User Joined: 1/28/2005 Posts: 6,049
|
That indicator is in Version 7 but it doesnt use stddev.
That said you can probably replace the middle line with a 200 period moving
linear regression line as the new language does have a stddev pcf.
Close enough for government work.
Thanks
|
|
Registered User Joined: 6/30/2017 Posts: 1,227
|
b, diceman, thanks for the info!
I noodled around at Starbucks and simulated a faux Linear Regression Channel by plotting the Lin Reg line, and a pair of 1-period moving averages offset by postive and negative amounts (see snap below).
Bruce,
Is there a way - given your mad Ninja math skills - to calculate the amount of moving average offset needed below given a lookback period of x and a Standard Deviation of y?
Also, now that I can access the forum, here are the parameters for the "system" referenced above...
A direction filter
For this strategy I use Standard Deviation Channels to identify the dominant trend. I think these channels are an excellent way to identify the market direction. The strategy is long only and if the channels are pointing upwards I will enter a trade.
-
The standard deviation channels were calculated based on 200 days and were offset by 0.5 standard deviations.
An exit system
The exit strategy that I am using is Bollinger Bands. Bollinger Bands expand during market volatility and contract during quiet times.
-
I set The Bollinger Band multiplier to 1.5 standard deviations.
An entry trigger
The entry trigger is a Japanese Candlestick pattern. In fact I use a very simple pattern. I wait for the price to close below the lower standard deviation channel and then look for a bearish candle. The bearish candle must have a body a minimum percentage of the height of the candle.
-
The body of the candlestick entry trigger must be at least 65% of the total height of the candle (between the high and the low).
Stop Loss
-
I had a stop-loss calculated at 3 times the ATR.
The test results on the S&P for 25 years look pretty decent. I won't clutter up the forum but folks that are interested can check the links. Basically, the win% was 72% and the average winner was just a smidge lower than the average loser. Of course, all the usual disclaimers about past performance apply. :)
|
|
Registered User Joined: 6/30/2017 Posts: 1,227
|
QUOTE (diceman)
... you can probably replace the middle line with a 200 period moving linear regression line ...
Close enough for government work.
diceman,
Noodling around now with the moving linear regression line (I use it a lot) and a pair of offset moving averages.
Until Bruce joins the party with his genius Jedi suggestions (he's waaay smarter than me) I'll continue noodling, ... and caffeinating. :)
Oh, and I've never worked for the government but I've been on the receiving end of their work. No thanks. :)
|
|
Worden Trainer
Joined: 10/7/2004 Posts: 65,138
|
As far as I can tell, the entry would be the following.
H > L AND O - C >= .65 * (H - L) AND FAVGC200 > AVGC200 AND C < 3 * FAVGC200 - 2 * AVGC200 - .5 * STDDEV200
And the exit would be the following:
C > BBTOP(1.5, 20)
-Bruce Personal Criteria Formulas TC2000 Support Articles
|
|
Registered User Joined: 6/30/2017 Posts: 1,227
|
Thanks, Bruce. I'll kick the tires on your PCFs. Appreciate all your help!
Even if I don't wind up using the system exactly as is, I always learn something ... I never considered using BB's as an exit tool before.
Is there an easier/better way to plot Standard Deviation Channels in v17 than my hack above?
|
|
Worden Trainer
Joined: 10/7/2004 Posts: 65,138
|
Your method is quite clever. It doesn't plot in any sort of multiple of standard deviation or standard error, but it does get wider apart when the slope is steeper.
I really can't think of a better / easier way off the top of my head.
-Bruce Personal Criteria Formulas TC2000 Support Articles
|
|
Registered User Joined: 6/30/2017 Posts: 1,227
|
QUOTE (Bruce_L)
Your method is quite clever. It doesn't plot in any sort of multiple of standard deviation or standard error, but it does get wider apart when the slope is steeper.
I really can't think of a better / easier way off the top of my head.
High praise from the Jedi Master! Thanks!!
I have my moments. :) ... to tell you the truth I didn't notice my hack gets wider as the slope increases ... though I did notice the phenomenon after I refilled my coffee and looked at other tickers and lookback periods.
I'm even smarter than I thought. ;)
Thanks for your help.
On to the next brainstorm ...
|
|
Guest-1 |