10 months fulltime and still no edge

estim · Jan 8, 2009

From my savings I started out in marts 2008 fulltime trying to build fully automated systems. So far 10 months working on fully automated systems and I still have not found a stable solution. Living in Copenhagen, thatâs in Denmark (which is why my English a mess), the US market is open from 15:30-22:00 local time. Normally I program the models during the day and follow the market when it opens at 15:30.

My focus has been high liquid US stocks. I have an account with IB which I connect to from MatLab or NinjaTrader, depending on what types of models I have tested.

All my models so far are intraday. I have been working on different timeframes from 5 secs â 5 mins. These are some of the methods I have tested without luck:

Pairs trading: scanning the equities in the Nasdaq 100 index for correlations/cointegration and trying to bet on the spread. No luck. Statistical arbitrage models using principal components. No Luck. Typical indicator models: Crossover systems, channel breakouts, ect ect. No Luck

I am down to 3 models that I will give a go:

Sector/etf models. Scanning for leaders and laggards in the first 30 minutes or so.
Trading the gap.
Trading around High/low of the day.

Basically, I think it is my Trade management I need to work on and to keep the models much simpler than so far. I have also been day trading (not paper trading) a couple of months for getting a feeling of ordertypes/fills when going to be used fully automated.

Any comments are appreciated.

ehsmama · Jan 8, 2009

you can look at many things still -
1. trade the news
2. trade the events
If you are full time trader. It is not hard to make money.

chvid · Jan 8, 2009

We are in the Bermuda triangle of finance right now with all instruments spinning around.

So it is not easy now - it probably never was.

But:

Look at time frames that are longer than 5 seconds to 5 minutes.

Days and weeks.

If trade management / trade sizing is a problem then simply stepping down in leverage / exposure may help.

What is your performance measure?

Consider hedging any beta out.

H2O · Jan 8, 2009

Quote from estim:

From my savings I started out in marts 2008 fulltime trying to build fully automated systems. So far 10 months working on fully automated systems and I still have not found a stable solution. Living in Copenhagen, thatâs in Denmark (which is why my English a mess), the US market is open from 15:30-22:00 local time. Normally I program the models during the day and follow the market when it opens at 15:30.

My focus has been high liquid US stocks. I have an account with IB which I connect to from MatLab or NinjaTrader, depending on what types of models I have tested.

All my models so far are intraday. I have been working on different timeframes from 5 secs â 5 mins. These are some of the methods I have tested without luck:

Pairs trading: scanning the equities in the Nasdaq 100 index for correlations/cointegration and trying to bet on the spread. No luck. Statistical arbitrage models using principal components. No Luck. Typical indicator models: Crossover systems, channel breakouts, ect ect. No Luck

I am down to 3 models that I will give a go:

Sector/etf models. Scanning for leaders and laggards in the first 30 minutes or so.
Trading the gap.
Trading around High/low of the day.

Basically, I think it is my Trade management I need to work on and to keep the models much simpler than so far. I have also been day trading (not paper trading) a couple of months for getting a feeling of ordertypes/fills when going to be used fully automated.

Any comments are appreciated.

Especially when trading pairs, I think you should increase your time frame...

estim · Jan 8, 2009

Quote from ehsmama:

you can look at many things still -
1. trade the news
2. trade the events
If you are full time trader. It is not hard to make money.

I have also considered those options. However, if I cannot get the other systems to work I don't think I can get a fully automated newstrader or eventtrader to work...

estim · Jan 8, 2009

Quote from H2O:

Especially when trading pairs, I think you should increase your time frame...

Actually I think I will give this a go. One of my challanges was also the normalisation of the two time series, or when to have the initial spread. Eg. working on intraday spreads I guess you should align the series within the first 30 minutes or so??? or adjust for the gap (this is what I did).

estim · Jan 8, 2009

Quote from chvid:
If trade management / trade sizing is a problem then simply stepping down in leverage / exposure may help.

What is your performance measure?

regarding trade management it is not initial position size it is more trade management strategies I think I should optimise. Scaling in/out, multiple targets, stops vs no stops or stop and reverse.
I use several perfomance measures but most important I am always looking at the slope of the accumulated profit. I read in one of your posts that you were working on a automated system have you got it up and running?

Wayne Gibbous · Jan 8, 2009

Automated/mechanical trading doesn't work. You need to "read the tape" with your experience.

Tums · Jan 8, 2009

Four blind men checking out an elephant,
you know what they found...

also,

If you are looking for the wrong thing... you will find the wrong result.

if you are looking in the wrong places...
you will never find it.

if you are using the wrong tools...
you found it, but can't capture it.

if you are there at the wrong time...
you will miss it.

more later...

murray t turtle · Jan 8, 2009

Quote from chvid:

So it is not easy now - it probably never was.

But:

Look at time frames that are longer than 5 seconds to 5 minutes.

Days and weeks.

If trade management / trade sizing is a problem then simply stepping down in leverage / exposure may help.

What is your performance measure?

==================
Es-Ti;
Diversify in time, end of day, days;
tough business anyway, but intraday may be the toughest.

And while diversification in sectors, markets;
& larger time frame helps,
wisdom cuts back when losing.May indicate a trend change;
certainly did indicate a trend change in DEC

Also analyze errors[if not profitable, wisdom is smaller is better];
i mistakenly thought DEC 1st would be bear rally;
overdue,
bear rally ,DEC 1st, -NOT TRUE.

Repeat, when protitable, cut back when losses;
it may indicate a trend change. It did indicate a trend change ,DEC 2nd................................................................

Repat, DEC 2nd we know now .started a trend change;
and that trend is still in effect ,
it may change back ,if it takes out/closes $88 0n SPY