I have a few moments to spare this morning so here goes. First it seems like most, the OP has been focusing on accuracy (trying to be "right")...That won't work. Instead a developer trying to find a profitable strat should be looking at "risk- to-reward" scenarios, specifically he or she should be trying to find a way to cash in on the most rewarding (profitable) opportunities in the markets they want to trade... This general rule can be applied to any scenario, but especially for those wanting to automate it is clear that this is the way to go. Why? well, take a moment to think about this. What is the point of automating a strategy, if not to remove the element of human emotion? Why try to remove emotion from the equation? Inability to control emotion is what gets a trader in trouble. We know this intuitively or if we don't we find out by getting an ass whipping in the markets. The most profitable opportunities in the markets today are those that cause the most emotional turmoil to traders. These opportunities exist because humans don't like to (or cannot) tolerate the emotional ups and downs that are required to get into a postion where they can profit. This is why the majority of volume on the Dow is programmed trading.....AND this is why the majority of that volume is backed by instituitions that have enough capital to tolerate the drawdowns and the expertise and experience to hang in there while the system works. Finally this is why when done correctly automated strats bring in big dollars, and conversely when done badly, they are subject to significant drawdowns. Some of these opportunities are obvious. Capitulations for instance offer big reward, especially when traded on the larger time frames. I can think of several on each time frame and they vary by market depending on whether you are looking at trending or mean reverting scenario. Applying this principle to my own commentary I would guess that the odds are poor that the original poster (or anyone for that matter) will "get it" and profit from the concept....LOL
I have been on vacation since Friday so I did not post the end of day P/L. Friday was a very slow day as part of my new filters were in. It only took 2 trades for a loss of: -$18.91 Looking forward to tomorrow. I think I will have a much better week.
All 12 of the stocks have greater than 2m average volume. The strategy can trade from 9.30-3:45 eastern. My slippage depends on the instrument. But I am incurring at least 2 or 3 cents a round trip. Some of the higher priced instruments more. The 600 shares is spread out of several trades. For instance the day I lost $140 i believe it made 11 trades that day. Some of the higher priced stocks it may only trade 20 to 30 shares a trade.. cheaper stocks 100+ shares. The number of shares for each trade is calculated by diving $3000/price of stock. So that all trades have the same $$$ value in play regardless of stock price.
I just reran the past 9 trading days using market replay one day at a time using the new filters, trailing stops.. etc.. I noticed of the 12 stocks the strategy watches only 9 of them ever traded on any of the past 9 trading days. AAPL traded 3 days for a total of -$19.24 XOM traded 3 days for a total of -$25.74 RIMM traded 4 days for a total of +$86.79 DD traded 2 days for a total of +$68.49 MRK traded 3 days for a total of +$36.29 MET traded 3 days for a total of +$145.69 JPM traded 4 days for a total of +$82.81 AMZN traded 1 day for a total of +$58.04 PFE traded 1 day for a total of -$10 None of the stocks traded more than half of the available days. Which is fine, the setups will not occur in every stock everyday. Since 3 stocks never traded, I think I will remove those 3 stocks from the list. So, I need to come up with some replacements. I have several stocks that are "on the bubble" to be removed. Those include AAPL,XOM,PFE, and AMZN due to lack of trades or lack of winning days. For example AAPL and XOM have lost everyday those have traded. Which might be a sign this strategy does not work well with the dynamics of those stocks. Thoughts? Do you think its a good idea to be rotating stocks in and out of the main lineup like this? I think its the only way I will ever end up with 10 to 12 stocks that the strategy excels at. EDIT: The 4 stocks on the bubble will remain the rest of this week to see if they can redeem themselves. 9 days may be too small of a sample to see if those stocks have potential with this strategy or not. The 3 that have not traded at all will be removed if I can think of some good replacements
For the 3 that have not traded at all, yes I do. When there are hundreds of stocks out there that meet the criteria needed for this strategy, then it doesn't make a lot of sense to stick with 3 of them that couldn't find the setup even once in almost 2 weeks. For the ones that are "on the bubble", I don't feel those have had enough time. And I still don't think they will have had enough time after this up comming week as well, but at the same time there are 100's that I could possibly have in its place.
frosty, Serious opinion. You'll bust within 3 months. I think I've written this in your last journal... your approach towards trading continues to go around in circles, never improving. The reason why you quit backtesting simply states that you have no idea about systematic trading yet. Everything you are provided with are a bunch of tools. It's how you use and conclude from the results of the tools. Read back some of the advices you got from the last journal. This is getting old. ------------------------------------------ stevesbg, I started automating my orders because I was manually executing 4 models and my fingers could not keep up. I never had all those emotional problems manually executing orders from a system. I can definitely say that people who use emotions and discipline as a reason to automate trades will end up as losers. No one has to agree, I don't expect anyone to deny their own intentions.
Very possible. I don't think it will be because I stopped backtesting. There are many flaws with backtesting that I saw caused significant differences when live. I actually thought ninjatrader's "Market replay" would serve me better than the normal "backtesting". But what I am finding is that is coming out different as well. For instance run the strategy live one day, then at end of the day re-run the strategy using market replay for that day and I am seeing some different trades between the two. I dont mean different prices/slippage. I mean in some instances a trade not taken at all.. With this reason I am switching to sim until I can figure out with them what is going on. If I can not trust what I see when running it with market replay, then I can not trust any changes/filters I put in.