StockFetcher Forums · View by Author: (35 messages)  [ Display By: Date / Subject ]<< 1 2 3 4 >> 
Filter Exchange · TRO'S CROCK POT 2.0
frsrblch
msg #115983
10/16/2013 7:48:17 PM

It was with the filter I posted on page 2, which was a simple 52-week version built for easier backtesting.

Filter Exchange · TRO'S CROCK POT 2.0
frsrblch
msg #115973
modified
10/16/2013 2:49:32 PM

As you said, novacane, holding requires you to either keep a large reserve of cash to keep buying new stocks each week, or for you to limit the number of new buys you can make each week. Holding indefinitely is not something I'm comfortable with; the system was designed around finding stocks with a good probability of giving a small return in the span of a week, not finding stocks where you want to park your money waiting for a turnaround. With that in mind, I set two weeks as a arbitrary length of time to hold on to a stock after it missed its 2% target (for a maximum hold time of 3 weeks) and did some testing based on this.

The portfolio size was 5, and the testing was done back through the week of Dec 31, 2012. If a trade was not profitable, it was held for up to an additional two weeks, keeping the original stop of 2%. If a stock was held longer than one week, one fewer trades were entered on the next week, so that there were five positions open at the start of each week (you could do it differently, but I decided to test it that way). I'll spare you the full details, but the annualized return of this system was 79%, compared with only 52% when losing trades were sold on the first Friday. On average, each week was more consistent (lower standard deviation), and more profitable (1.13% per week vs 0.82%).

Filter Exchange · TRO'S CROCK POT 2.0
frsrblch
msg #115933
10/15/2013 12:17:47 PM

"I knew that when I made a comment about what I'd been able to do with the original filter, it wouldn't take long for someone to say "it needs to be modified/changed and you're full of BS".....and it didn't. It doesn't need to be modified."

The first thing I do once I have a system like this modeled in Excel is to start throwing ideas at it, analyzing the hell out of it in an attempt to make it better. Different combination, profit stops, and trading methods. How effective is the system if you take 8% profit instead if 2%? What if you hold each stock until Friday, regardless of what it did? What if you hold onto stocks that didn't perform well in the first week? I find this kind of analysis very stimulating - perhaps something missing from my actual job. The post you made about your results trading the original filter, other people's opinions of its accuracy aside, brought my interest back to a system I had originally overlooked for being too difficult to backtest within StockFetcher. Having finally looked into it, I can see the potential... but I'm still going to poke and prod to get my fun. Best of luck to you and yours, Frank.

Filter Exchange · TRO'S CROCK POT 2.0
frsrblch
msg #115922
10/15/2013 9:40:09 AM

olathegolf, what was the average length of time these trades were held for before they hit the 2% profit stop? It complicates the system a little, but it's fairly remarkable if you can change the win rate from 87% to 98%.

Filter Exchange · TRO'S CROCK POT 2.0
frsrblch
msg #115889
10/14/2013 1:21:39 PM

You can indeed backtest these strategies, so long as your trading pattern is simple enough (e.g., buy at open, and hold to close or for an X% gain). You need to do a date offset on your selection criteria, but then you can export last weeks results with this weeks prices in a CSV. Throw that into Excel and have at it. Be advised that holidays tend to affect your wopen, whigh, wlow, and wclose values, so you'll have to enter them manually for each stock on these weeks.

I've actually been looking at these systems for the past couple days, and they've been working well for the last year or so. Beyond that, they start to slide the other way, with too many misses happening to make it work. This 13wk Crock Pot 2.0 works splendidly, but only for the last three months before you stop getting enough results to trade on. Haven't looked at the 52wk version yet.

Fetcher[
/* TRO STAT SCAN for SWING TRADERS - Manual Backtesting, look at the Friday of each week */

set{whiop, weekly high - weekly open}
set{Long_Profit, whiop / weekly open }
set{wkProfitPct, 100 * Long_Profit }

set{twopct, count(Long_Profit > .02 , 52)}
set{threepct, count(Long_Profit > .03 , 52)}
set{threepctrecent, count(Long_Profit 1 week ago > .03 , 52)}
set{twopctrecent, count(Long_Profit 1 week ago > .02 , 52)}

set{twopcttriggered, count(Long_Profit > .02 , 1)}
set{threepcttriggered, count(Long_Profit > .03,1)}

set{filter1,threepctrecent/100}
set{filter2,filter1+twopctrecent}

add column wkProfitPct
add column separator
add column weekly open {wopen}
add column weekly high {whigh}
add column weekly low {wlow}
add column weekly close {wclose}
add column separator
add column twopctrecent {hit 2% in last year}
add column threepctrecent {hit 3% in last year}
add column separator
add column filter2

draw twopcttriggered {hit 2% target}
draw threepcttriggered {hit 3% target}

close 1 week ago is above 5
average volume(90) 1 week ago above 500000
TWOPCT 1 week ago ABOVE 42
THREEPCT 1 week ago ABOVE 39
sort column 15 descending

chart-display is weekly
]



Filter Exchange · TRADING DIVERGENCES ON THE S&P 500
frsrblch
msg #112738
modified
4/12/2013 8:22:00 PM

The G column in the Exit sheet isn't actually used. Originally, I had built the sheet around buying or selling when the signal goes from 0 to 1, regardless of what the other signal is doing, instead of the (entry=1,exit<>1) and (exit=1,entry<>1) it uses now. I changed how the exit was calculated at some point, and forgot to delete the unnecessary column.

Filter Exchange · TRADING DIVERGENCES ON THE S&P 500
frsrblch
msg #112736
modified
4/12/2013 3:54:34 PM

You can fiddle with the entry and exit criteria in my spreadsheet (the column with blanks and ones immediately to the left of the entry and exit price columns) to test different approaches, but my testing seemed to show that holding when you get both a buy and a sell signal on the same day appears to be the best approach.

Filter Exchange · TRADING DIVERGENCES ON THE S&P 500
frsrblch
msg #112725
4/11/2013 4:41:49 PM

Thanks, jackmack. I'm glad that people are finding it useful.

Looks like a great exit today. Taking money from less intelligent investors it may be, but it's the nature of the system that that money has to come from somewhere.

Filter Exchange · TRADING DIVERGENCES ON THE S&P 500
frsrblch
msg #112654
4/5/2013 1:43:47 AM

Fantastic idea, pulling data from Yahoo Finance like that.

Filter Exchange · TRADING DIVERGENCES ON THE S&P 500
frsrblch
msg #112578
4/1/2013 10:49:24 PM

Zen and the Art of Stock Fetcher

In other news, the spreadsheet is done. Data is pasted into the 'Data' tab, with the left column being the asset to be traded, and the right column being the asset used to determine entries and exits. As a default, these are SPY and the S&P500, respectively, but you can change these as you see fit, provided you make sure the dates line up across both columns. The sheet can handle up to 2500 days of data, beyond which you'll have to add more lines of formulas to each of the filters.

The 'Control' tab shows some limited statistics for each system, including annual return for the period backtested. I have included an option to limit the number of days a stock is held for in each system, but most of these systems react poorly to these exits.

The spreadsheet is designed to buy when the entry signal is 1 and the exit signal is 0, and sell when the entry signal is 0 and the exit signal is 1. Please feel free to check my calculations, and let me know if you find any errors.

If nothing else, this spreadsheet has shown me the importance of using ^SPX data rather than SPY data when it comes to choosing entries and exits. Most systems grossly under-perform if you use SPY data for choosing entries and exits. The poor results I have had backtesting most of Kevin's Top 10 in SF are likely the result of this. The data in the spreadsheet runs from Jan 1, 2006 through to today, and it's worth noting that these systems plowed through 2008 with barely a hiccup.

https://docs.google.com/file/d/0B0rUWi0Alg22ZnlWNEU1TkNGbDg/edit?usp=sharing

StockFetcher Forums ·  · << 1 2 3 4 >>

*** 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 2022 - 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


This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.