Frosty's trading bot goes live part 2

Hey frosty, the good news is your bot is consistent. I would pay a lot of money for a trading system that loses huge consistently. No joke. The only bad systems are ones that break even.

The good thing about a system that "loses" huge is that you can fade it. You may want to switch those buy and sell commands.... see what happens?
 
It does trend following?
Intraday trends can be short lived.
If you'd post the entry times and direction, people might get some insight into how it finds trends and might be able to suggest improvements.
 
Quote from nonlinear5:

Yeah, but then you'd also have to code around PPI days, CPI days, GDP days, ISM days, Home Sales days, and maybe even Truck Sales days.

..
More...
I would draw the line at Truck Sales days but seriously he may want to have file of 'special days' so that the bot could handle them differently.
 
Man just when I thought things couldn't get any worse....... I don't get what has happened. I think I have the worst luck when I go live. It would be one thing if the results did not match the sim account's results, but they do....... so its not like something is different "because" i'm live. Its like I pick the worse times to go live, or something changed... i'm not sure what the deal is.. i'm not giving up....

Today -$1334.40
Week to Date: -$2056.8
Since going live: -$3379.6
Total days: 10
Avg per day: -$337.96

Up Days: 4
Down Days: 6
 
Quote from frostengine:
Man just when I thought things couldn't get any worse....... I don't get what has happened.
More...

I think what is happening a 2nd time frost is that you are proving in real-time with real money that designing a computer program as a retail trader that can make money in real time based on historical data is rather difficult to do. Once you add in that your personal emotions will eventually take over (as before) and for those that say designing a program is superior to being the trading yourself each day b/c the emotions are taken out of it and you realize quickly that emotions are in fact part of trading, regardless if your computer bot is doing it or not.

The question now becomes, how long are you willing to 'hold on' to your bot here as the losses mount, which is obviously a question of how your emotional state can take this. If you're confident that this type of a drawdown is to be expected and just bad timing on your part with real money, then you must stick with it. If you are not 100% confident though, that's when things can get tricky.
 
frost,

in order to reduce pain, can't you scale down your strategy a litte? Obviously you can't lower position sizes but you could consider tuning other strategy parameters (profit targets, stop loss, timeouts etc) to reduce risk. Sure, this would probably reduce profitability also but you could gain time to have a better feeling what the situation actually is - bad luck or bad market.

If you can not scale down your strategy this way because outcomes are too sensitive to parameter changes - this would be a sign of overfitting. (I'm sure this is nothing new for you though)

torel
 
Quote from frostengine:

Man just when I thought things couldn't get any worse.......
More...

Frosty,

Could you decribe for us how you created your system? I don't care about the system itself, but what steps you took to create/optimize it?

I'm not sure about you, but alarm bells are going off like crazy for me. Since the simulated results are matching your actual results, then the most likely cause of a problem (assuming that there is one) is that your system is over optimized to fit past data (and therefore might be worthless in trading with future data). Is this possible for your system?

Bad luck and Murphy's Law is common in trading; however, you should be certain that bad luck/timing is all that it is, and not an indication of a poor system.

Note: The second post of this thread contain some of my thoughts on proper 'optimization':
http://www.elitetrader.com/vb/showthread.php?s=&threadid=36083&highlight=activate/deactivate


-Eric
 
Frost...

Im sorry to see you having these difficulties. But in all honesty I have to say... I know you've stated many times this is a trend following system, but now Im questioning the method the bot is using for determining or defining a trend. Results speak for themselves... seems you are operating a counter-trend bot, or the bot is failing to recognize intraday trends, taking intraday trades based on longer term analysis instead. A good intraday entry can easily be converted into a profitable swing trade. A good swing or position entry can be much harder to convert into a profitable intraday trade. jmho

I wish you all the best.
Osorico :)
 
You must log in or register to reply here.
Back
Top