| StockFetcher Forums · Filter Exchange · HOW TO DESIGN A SYSTEM (NOT JUST A FILTER) | << 1 ... 1 2 3 4 5 ... 43 >>Post Follow-up |
| jkinghome 7 posts msg #101146 - Ignore jkinghome |
6/9/2011 8:55:28 PM Hi all, I am new to SF and loving it. I am also paper trading this system. I must say that as a newbie, and going through most of the historical forum posts, I have learned more from Kevin's posts than anyone. Kevin's posts ard filters are well thought out and very well communicated to the readers. Some of the posts here are just so cryptic! Thank you for helping the SF community, Kevin! I am looking forward to this strategy. Kevin, I have just joined Stratasearch as well. I am having problems trying to get the custom forumla to work. But I will try some other ways. Jacqueline |
| Kevin_in_GA 4,599 posts msg #101148 - Ignore Kevin_in_GA |
6/9/2011 10:55:19 PM PKI opened at 26.37, so it records as a loss of ca. 0.1%. I might have held this through the day given how the futures were looking, but the mechanical system says close the trade at the open price so that's what we'll do. Two more closed today - SWY and FMC. Will close them out using open prices tomorrow. Replace them with KIM and ALTR at the open price tomorrow. |
| calhawk01 135 posts msg #101149 - Ignore calhawk01 modified |
6/9/2011 10:56:17 PM I just did a small research on this system. I thought i would share it: 1. Exit variable= rsi(2)> 70 rather than the std gives better results. Backtested this fact in the increments of two years - ranging from 2002- present day And as a general rule: 2. Always set your entry price as "yesterday close" and make sure you enter a limit order at that price. I did a research of SPY and various other stocks. Backtested this for the past 5 years. There are 70% chances that your limit order will get filled the first day. 80% chances that it will be filled by the 3rd day. What does that mean for us? It reduces slippage. I tested this system and compared two entry prices. 1st was next day open and second was yesterday's close. Backtested it from 2002-present day in the increments of 2 years. Overall a limit order at yesterday close price improved ROI and decreased the numbers of trades that occur (obviously). |
| Kevin_in_GA 4,599 posts msg #101150 - Ignore Kevin_in_GA |
6/9/2011 10:56:20 PM Jacqueline - send me a PM at stratasearch.com and I'll try to help. |
| Kevin_in_GA 4,599 posts msg #101151 - Ignore Kevin_in_GA |
6/9/2011 11:09:46 PM 1. rsi(2)> 70 rather than the std gives better results. Backtested this fact in the increments of two years - ranging from 2002- present day +++++++++++ Define better ... what I am looking for is higher equity summary values and lower volatility, rather than annualized ROI or win%. Also, look at the Sharpe ratio of the stats I posted for this filter - that is what professional quants use as their metric for a "good" system. Here are the results for the same system and timeframe, but using the RSI(2) > 70 as your exit trigger: 
Compare this to the system stats in the first post in this thread. As you can see, total profit is lower, as is per trade return and a lower Sharpe ratio. |
| calhawk01 135 posts msg #101152 - Ignore calhawk01 |
6/9/2011 11:29:30 PM Well i can't argue with that. Your statistics make it more clear. I was using stockfetcher and it seemed to me that rsi(2)>70 increased your win% and roi%. By the way, sharpe ratio calculation = Return on portfolio - risk free / STD portfolio So.. Average annual return = 90.40 30year riskfreerate= 4.37 standard deviation= 41.88 90.40-4.37 / 41.88 = 2.05 = sharpe ratio How does your software derive .4914? Thanks |
| jkinghome 7 posts msg #101153 - Ignore jkinghome |
6/9/2011 11:57:15 PM I believe that Kevin's software uses monthly Sharpe ratio. So you need to multiply it by 3.46 (square root of 12) which equals 1.7 annual Sharpe ratio. |
| starfox182 24 posts msg #101157 - Ignore starfox182 |
6/10/2011 6:38:34 AM How does this work if you don't have 100k? Does holding a little longer yield greater % gains? |
| Kevin_in_GA 4,599 posts msg #101158 - Ignore Kevin_in_GA |
6/10/2011 7:30:43 AM Not really. I ran several variants that looked at exiting when the Z score was >-0.5 and >0 and neither did as well. I also looked at extending the maximum days held, but these trades are typically completed in 5 days or less, so it added nothing. |
| seanban 22 posts msg #101183 - Ignore seanban |
6/10/2011 3:09:18 PM novacane320000 "I took a look at setting a stop at the open of the next day after a sell trigger but that will not improve results-With the small sample size I looked at most times I got a worse fill with the stop than I would have by just selling at the open. As far as which is best all depends on which way the market and your stock is headed when you sell. Either way ,very excited at the potential here. Big thanks to Kevin for his work. " Thanks for your reply. One variant I'm building into the paper trade system is to buy on green i.e. only when it trades a few cents above the open price and round numbers. As today was predictably a red day, I'm wondering if there is a way to improve the approach. Indeed a big Thanks to Kevin for his inspired feedback. Sean |
| StockFetcher Forums · Filter Exchange · HOW TO DESIGN A SYSTEM (NOT JUST A FILTER) | << 1 ... 1 2 3 4 5 ... 43 >>Post Follow-up |