Well the bot is officially done for the day, total= +$345 So with the running total as of yesterday at -$232 It puts the current progress of the bot at +$113 overall since gone live... So I am very happy about being green overall once again... the losses hurt, but its not as bad when I know I am up money overall.. In the next 20 min or so I'll run a backtest of the origional method on today and see how it would have faired if it was used instead for comparison sake
FrostEngine, Please don't take this advice as anything but well-intentioned. I think you need to get rid of these kind of emotions if you're going to be successful at any kind of trading, discretionary or algorithm-based. By turning your system on and off and altering the strategy while your are in the midst of trading live money, you are actually gaming the system and not giving it a chance to produce the statistics you see. I think in your case, if you haven't done so all ready, a good account blowup would serve you well, actually. When it happened to me on my first $20,000 account, I was devastated, but it left me with more determination. But beyond that, losing $20k forced me to realize that only the strictest of discipline in following and executing a strategy will yield positive results over time. Moreover, if you blow up your account, you will most definitely be significantly desensitized to the ups and downs of your account equity as you trade the second time around. One can hear about other traders' blowups, but the lessons can only really be drilled deep inside once you've blown up your own account. I believe that it's really a healthy thing for any serious trader to blow up a small account once (but not too many times, of course). RoughTrader
Roughtrader, I have blown up once already on a discretionary account. I think its because of this blow up that I am so worried about all the moves.. I don't want to experience another blow up.. I wish I can get to where I don't even think about what the bot is doing... when I ran for several months on the sim account I wouldn't look at how the bot was doing all day.. I would turn it on in the morning and when i got home from work in the afternoon I would see how it did..... now I find myself checking every 20 min or so.....
Now I have something to truly be concerned about.... I made a lot of changes to the backtesting engine last night while trying out some new strategies and changing the way i was doing a few indicators..... so that MAY be what happened....... BUT something happened.. because I just backtested today using the method I used today and I got I should have had a lost of -$80.. not a gain of $345.... granted i am really happy I had a gain of $345 today, but now I am really concerned because its the first time in 4 months where a backtest did not agree with what really happened live that day.... My only explanation is either #1 I changed something last night in the backtesting engine only that screwed it up.. or #2 the market internals I am using may not be being the same as I am seeing it live.. which I highly doubt that and don't see how it would even be possible... a #3rd possibility is the backtesting and live mode both have separate settings on how many ticks things are running on for bars and different things like that... perhaps one of the new pieces of information I am using is not set the same on both spots.... either way very concerned and worried.... I'm sure this post is going to bring a lot of heat from some of you who jump to conclusions and say that perhaps my backtesting is always off from the real results.. but this is the ONLY day that has happened... and coincidentally the only day i traded this method live... so it has to be something specific to this method and not the previous one or the system as a hole..... Talk about ruins my day.. I was all excited that things were looking up
sounds like a faulty backtest. do you use an implementation shortfall algorithm in your backtest (commissions, taxes, b/a spread, market impact of trade, & opportunity cost)? real trades almost always underperform paper, because the price moves against you before the order is executed. There are some relevant points in this thread: http://www.elitetrader.com/vb/showthread.php?threadid=39461 also (sorry if I missed this) but are you using market or limit orders?
I am figuring in bid/ask spread, and commissions into the backtest...... I am using market orders. Also as for as the market going against me when I enter my trades... typically I get slightly better fills than what was estimated.... Part of this could be by virtue of it being a somewhat counter trend strategy.. so sometimes the market is still moving away from my actual entry point, thus giving me a better fill than what I thought i would get.. for instance when I enter shorts, the market is normally moving up, and sometimes there is still substantial buying pressure when I enter my trade so I actually get to sell higher than I thought.... One plus of counter trend trading I guess... Still no idea what actually went wrong on the backtest for the day..... the thing that scares me the most is all of my backtesting on method 2 may be flawed for some reason..... which might mean I have to go back to my original method.... I would hate to do that because I actually really think method 2 is better........ can't wait to get this bug figured out to know where I stand.... a LONG weekend of work is in front of me..
Frost, not trying to lecture you or being mean, but I don't think you should be running ATS with live money yet. Not because you had a bad day, good day or... But because it doesn't seem your system is ready yet. From my backgroud of Software Engineer, to develop a successful software project you have to follow rigourous steps. the traditional way is: 1- Functional specification 2- Design specification 3- Implementation/Coding 4- Testing/Fixing bugs 5- Deployment/Maintenance It seems you are doing everything from 1 to 5 at the same time. I understand that in trading you might not be knowing exactly what the system should be doing when you are in 1. The solution is then to stick between 1 and 4. Do it in a loop as many times as necessary, and when you are happy with the results, ONLY THEN go to 5. It is like having a prototype, testing it, learning from it and building a new prototype, testing it, learning... It is called RAD process (Rapid Application Development). Just google it and there is a ton of info about it. Good luck,
StocksSniper, I appreciate the comments. However, I have been a professional software developer for some time now, and I know how the development process works. The software that runs this has been thoroughly developed and has gone through plenty of testing time.... roughly 4 months of testing in the sim account. I felt it was ready to go live so I took it live. Bugs always appear, that is a fact of software development. It appears a bug in the backtesting portion has surfaced with respect to a certain type of data being used. The bug will be corrected and things will move forward. I do however realize that if I was treating the actual STRATEGY as you would in typical software development that I am not following procedure. But my strategy is what it is, and will continue to evolve over time as I get new ideas and try to adapt to market conditions.... I probably should perform more testing on the strategies themselves before putting them out there..... but I always have been a bit impatient.. That is something I need to work on.. Again thanks for the comments, I welcome anything you guys have to say no matter how frank it may be. Sometimes that is what a person needs to hear to do the right thing.