Automating Trading Strategies in Todays Markets

RoughTrader,

No, I am not trading live again yet. Although, I am gearing up for another go at it in about a month. Just about got the funds saved up needed to let it run without being scared every 5 minutes that i'll have to shut it off.

Shutting it off when I did was the biggest mistake I ever could have made. Since that day on IB's sim account it has done incredible. I think the lack of volatility during the holidays is what got me. Since then the volatility has been higher and the bot's profits have soared.

I'll start another thread here when I do go live again.
 
Quote from frostengine:

RoughTrader,

No, I am not trading live again yet. Although, I am gearing up for another go at it in about a month. Just about got the funds saved up needed to let it run without being scared every 5 minutes that i'll have to shut it off.

Shutting it off when I did was the biggest mistake I ever could have made. Since that day on IB's sim account it has done incredible. I think the lack of volatility during the holidays is what got me.
More...

Live and learn. Systems have good periods and bad. Tough part is building confidence to trust your system through thick and thin.

RT
 
RoughTrader,

That is so true. You can have the best system in the world, but it will experience draw downs. If you don't have the confidence to stick with it during those times then it will be of no use to you. It is just so difficult when in the back of your mind you keep thinking... is it really broken this time?
 
I know very little about autotrading but it is something I am working towards.

With drawdown, cant you backtest your system (in sample) and find the max drawdown so when you forward test (out of sample) or go live with it you will know the max drawdown expected and then if the system begins reaching larger drawdowns you simply shut it off because it is no longer performing as expected?
 
oh, sorry frost I saw your post in another thread in regards to your system reaching max drawdown. Another solution I have read about in regards to knowing when your system doesn't work is to use an equity curve with a moving average. When you equity dips below that moving average all trades turn to sim until your equity rises above the moving average where your trading moves to live
 
Quote from cipher4d:

I know very little about autotrading but it is something I am working towards.

With drawdown, cant you backtest your system (in sample) and find the max drawdown so when you forward test (out of sample) or go live with it you will know the max drawdown expected and then if the system begins reaching larger drawdowns you simply shut it off because it is no longer performing as expected?
More...

Maybe max drawdown is significant psychologically, but not statistically. It's only one outcome, and not very probable that the same drawdown will occur in future trading because drawdown depends on the order of trades!

One way of proving this is to run a MC simulation, taking trades in a random order. If you run it several times, you'll get a distribution of drawdowns - any of which are probable future drawdowns. Then observe the moments of the distribution (skewness, kurtuosis), mean, and standard dev...
 
Well now that people are talking about risk management in systematic trading...

Equity Analysis is fine, I do it to some extent. Though, you'll have start asking questions of whether you are curve-fitting the equity curves, as you are with the actual markets.

If your competency of system development is mainly on technical analysis, then there's not much to say but study. Do not try to trade the equity curves with TA.

...

One thing to remember about MaxDD is to also look at the distribution. If the MaxDD is a major outlyer, then you're going to have problems because you are using a figure that rarely happens and as unreliable. Literally, outlyers are uncommon / unexpected results within a distribution. Rationally, it doesn't make sense to use an unreliable figure as part of your risk management, correct?

That is why using Monte Carlo or VaR starts having significance. It smooths out "some" of the ironies involved in DD figures.

....

Anyways... you have to do your tests. Test if cutting systems based on Equity curve works. Test if using MaxDD works. Test systems. Test risk management models. Test measures. If you have a thought in mind, test it. TEST EVERYTHING.

If you don't have the ability to do so, well that's your loss. You just have to believe / risk everything you read from other people.

(I seriously wonder, now about newbies (including myself back in the days...) ...

how can you risk a large sum of your money based on some book that costs $50-60 dollars...

or some trading system that costs $100... even more...

I'm sure you won't let someone else trade for you with the same skill level as you, so what makes newbies think they can make money trading???

Newbies are the most arrogant, self-inclined, assumptious traders.)
 
You must log in or register to reply here.
Back
Top