| StockFetcher Forums · General Discussion · BEHOLD THE HOLY GRAIL | << 1 ... 6 7 8 9 10 ... 14 >>Post Follow-up |
| Eman93 4,750 posts msg #145842 - Ignore Eman93 |
1/2/2019 8:28:56 PM K, No problem. |
| mahkoh 1,065 posts msg #145844 - Ignore mahkoh modified |
1/2/2019 9:39:01 PM Stratasearch code: Entry string: close>0.1 and close<10 and volume>1000000 and crossabove(rsi(2),98) and close/high<0.9 Exit string: 1=1 Order types: Entry: ref(open, 1), Exit: close |
| KSK8 561 posts msg #145846 - Ignore KSK8 |
1/2/2019 11:04:56 PM Nibor, 1. Portfolio123 is what I used for the backtest. Fantastic software. 2. The first thread was a gag.. I even said it was if you take another look. 3. And what do you mean by you have doubt? Stockfetcher confirms the superiority of this filter! In January 2018 if you would have invested into 1 candidate each day whenever it gave one, playing by the system rules, (if it was all based upon price) a $1,000 portfolio would have been turned into nearly $25,000 by June. -Excluding slippage,commission, and all that crap. Check it out for yourself if you'd like. I'm surprised you haven't realized the profit potential it has. 4. Hmmmmm, now if only SF would implement reverse splits as a factor onto the chart as well as dividends & earnings. That'd be nice... I already know a code to identify fda approvals through price & volume action which I would add to the filter, but if that were the case of those things being added to SF then yeah it would become "mechanical" I suppose. And to answer your question; those are the only two catalysts that could cause higher highs and possible whipsaw as to what I have observed |
| KSK8 561 posts msg #145847 - Ignore KSK8 |
1/2/2019 11:06:09 PM mahkoh, Don't just stand there? RUN IT, show these people the results! |
| KSK8 561 posts msg #145848 - Ignore KSK8 modified |
1/2/2019 11:53:02 PM TAKE A STAB AT THIS BEAUTY |
| mahkoh 1,065 posts msg #145850 - Ignore mahkoh modified |
1/3/2019 5:57:41 AM OK, results. A few things to consider: Out of the 6000+ securities I only have reliable data for some 3500 symbols. Getting data for all is a hassle which I do not intend to go through as I don't trade stock below $10. I cannot filter out OTC/BB and biotech. Most pricedata goes back to 2013 -2014, but the further back I go less stocks are covered. That is probably why the position count graph looks more crowded towards the present. I would suggest using a portfolio size of 2, thus risking half your capital on a trade, but for fun we'll just evaluate going all-in on every trade. If there were 2 candidates Stratasearch would pick the ticker which came first alphabetically but that can be optimized. I have used a variable for rsi number, it turns out rsi(2) above 99 without having to cross above gives best results. https://imghostr.com/A8fw2mXa Position count: https://imghostr.com/fwgzkXqz Equity curve: https://imghostr.com/dXO6Bdx4 Statistics: https://imghostr.com/FgSe73tY Worst results: https://imghostr.com/DsDAPa4G Regarding the worst results, I don't think a trade like HTBX on 5/1/2018 would have materialized trying to short 1.7 million shares at the open. Take the equity curve with a grain of salt.. I found a way to eliminate biotech by feeding the results into Stockfetcher. Out of 198 results 91 were labelled Industry: Medical. Average win for those was 4.59% per trade accumulating to 418 % over the test period. Non medical: 107 trades, average 5.40% totaling 577% |
| nibor100 1,099 posts msg #145852 - Ignore nibor100 |
1/4/2019 2:08:18 AM @mahkoh, First, thanks for running the backtest in SS for us all. 2nd. From my review of your results it looks like SS shorted at the Opens and Covered at the Closes but all of the trade exits are dated as the next day, is that an SS convention or glitch? 3rd, A majority of the stocks in your worst performers list seem to match the ones picked by KSK8's corrected filter in SF, but some don't. I know you have a couple of things working against that, limited number of stocks with good data, and no way to filter out BB/OTC. However is there a way you can get a count of how many stocks of your 3,500 fit the general parameters of his filter on a given date so we can compare that to how many stocks in SF fit those same parameters on the same date? Lastly, CNET is on of your poor performers with a trade on 1/11/18 but it doesn't show up in the SF filter results for that date. it does show up as a winning trade on 1/5/18, any ideas what happened, as your open and close prices look right? Thanks, Ed S. |
| KSK8 561 posts msg #145853 - Ignore KSK8 |
1/4/2019 3:38:25 AM mahkoh, I appreciate the backtest you conducted even though I am not a fan of the stats displayed. But I wanted to verify for myself, so earlier today I downloaded Stratasearch and after 30 minutes of speedreading through the manual, I conducted a backtest using the same parameters as you posted combined with some of my personal preferences. And after running it I must say these statistics seem flawed. And I am not just saying that just to say it. For instance if you examine some of the stocks it gave in the backtest results such as the one you listed Nibor, "CNET", you can clearly see it does not match the criteria when checked. On top of that, some of the results of some days do not coincide with stockfetcher's candidates. Why is this? Plus I can not filter out otc stocks without having to manually select them out. -- I'd say the isolation of otc stocks with this filter is a pretty crucial criteria. If you were to facilitate otc's in this strategy, you'd run into so many short squeezes it'll make you despise the purpose of the filter. A 400%'er appeared earlier in this year hence why I have "market is not otcbb". -- All that aside, I am utterly astonished with some of Stratasearch's features. How is this not popular? |
| Lapre506 48 posts msg #145854 - Ignore Lapre506 |
1/4/2019 7:56:38 AM Not that I have much to add to the conversation here... because people much smarter then me are debating. But one thing I've noticed with Stockfetcher from all my manual backtesting when you are using not otcbb in the filter code. If a stock was once on a major exchange and then got downgraded to OTCBB it will no longer show up in the historical results on that filter when looking back. This tends to sway backtesting results in a favorable manor. Because had you taken a play on a result for the filter and it crashed and became OTCBB you'd have taken a major loss. But when looking back because it's now OTCBB it no longer shows up in the history of the filter. |
| mahkoh 1,065 posts msg #145855 - Ignore mahkoh |
1/4/2019 8:13:38 AM @Nibor: The signal date is the date SS labels as the date when the trade is opened. ref(open,1) in the order type means the trade is executed at the next open. CNET turns out to be a data error. As you can see in the linked file it misses data from 1/4/18 to 1/9/18. Without these days the 1/11/18 spike probably led to a rsi(2) value of 99 thus triggering the trade. https://imghostr.com/mcadZNcq Stratasearch has settings on when to exclude a stock due to missing days, the default is 10. Stratasearch uses yahoo eod data as a source but changed things to their website which prevents a smooth download. I have another instance with a 2500 stock database that is running fine, but 6000+ with a lot of obscure companies that may not even be listed anymore is a problem. Usually my download crashes at the letter "M" due to excessive connection failures, so there will be few trades for tickers beyond that. Here is a list for the trades going back to February 2018. Remember that only one stock is selected alphabetically and I use rsi(2) above 99. And I accidentally coded volume > 1mln instead of price*volume>1mln https://imghostr.com/hgP2zaNd |
| StockFetcher Forums · General Discussion · BEHOLD THE HOLY GRAIL | << 1 ... 6 7 8 9 10 ... 14 >>Post Follow-up |
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