project J

[sharktooth]

I pay dutch tax so the cost of trading qqq is about the cost of trading nq for me. Taxmax here is 1.2% on average yearly
capital. How about having tax for an edge??

Anyhow, my purpose now is not to chat about taxadvantages,
but to find solid feedback.

I speak dutch, so no long mails in English. To keep it focused I will reply a.s.a.p. and only to mails that provoke fruitfull further development of project J.

Goal:
develop, papertest and realmoneyexecute a positive expectancy (mostly) mechanical daytrading method for qqq before 12/13/03.

I want to base my plans on two central concepts;
1] prices tend to fall faster then they rise;
2] only test money when the papertrades are consitent enough;

Expectancy:
-15 breakeven or better days on any 20-day period;
-%winining trades=> 50%
-netRisk:netReward=> 2:5

Setup options:
-3 charts: qqq, nq and tick.nq;
-only 2 minute candelsticks;
-max 3 ema's on a chart;

Money management:
-max. risk =0.8% capital;
-no overnight trades;

Trade management:
-enter position with stoplimit order;
-protect position with (manual) traling stopmarket order;

 

[dis]

I want to base my plans on two central concepts;
1] prices tend to fall faster then they rise;

Yes, but so what? How do you plan to translate this observation into trading profits?

 

[Lobster]

Quote from dis:


Yes, but so what? How do you plan to translate this observation into trading profits?

I believe this is a very interesting question. Let's seriously look for an answer! Let me think about it while I post my reply to the other suggestions.

 

[Lobster]

Why is it called projet J?

Quote from sharktooth:


1] prices tend to fall faster then they rise;

IF we believe it, how can we take advantage of it? Have you already given it some thought?

Are you just guessing that prices fall more rapidly than they rise or have you done some sort of test to prove it?

Also, what time frame are we talking about?

Quote from sharktooth:



2] only test money when the papertrades are consitent enough;

Expectancy:
-15 breakeven or better days on any 20-day period;
-%winining trades=> 50%
-netRisk:netReward=> 2:5

Forget all that. If you make money you make money no matter how your risk/reward ratio or your %winning manifest themselves.

I would suggest to replace those rules with the following: If you average more than 1 NQ point per trade on 500 trades, then we're good.

Quote from sharktooth:



Setup options:
-3 charts: qqq, nq and tick.nq;
-only 2 minute candelsticks;
-max 3 ema's on a chart;

Money management:
-max. risk =0.8% capital;
-no overnight trades;

Trade management:
-enter position with stoplimit order;
-protect position with (manual) traling stopmarket order;

OK, no objections there. I think it's as good as anything.

 

[sharktooth]

Dis and Lobster,

Thanks for posting.

Why is it called project J?
I have a regular job and my salarygroup is "I".
My ambition is to create a winning daytrading method
that will provide income equal or even better then "J".
"J" is the next salary group.

Although i have a normal daytime job, the time difference with the US will allow me to monitor the US markets realtime:
-monday -thursday: 12:00-16:00.
-fridays -whole tradinday

Prices tend to fall faster then they rise.
I have no proof for this statement. It is based on my observations of prices developing over a period of 2 years.

How will I translate this into $?
I will develop a system that is biased to capitalize on this.
Im looking for a time frame of 15 min to 2 hours. The idea is to build a mechanical system that is biased to be a bit more agressive on short entry signals then on long ones.

Im planning to have method that is ready for papertesting on monday. Based one your comments, more (chart)studies and thougts, I have updated the initial Project J into :

Goal:
develop, fronttest and run an +Expectancy (mostly) mech. trading method before 12/13/03.

Instrument:
qqq

Main concepts:
1]price tend to fall faster then it rises;
2]avoid entry in choppy market, smart stops;
3]test +Expectancy, then test $

Expectancy is + if:
1. at least 45 profitable tradingdays out of 60
tradingdays;
2. daily netProfit/netLoss => 2.0

Evaluate entire Method:
if 13 losing tradingdays, occur within 60 tradingdays;

Setup Charts:
maximum 2 trendlines a chart;

-nq & qqq:
daily from 8:00
3 min. chandle
ema(9) of closing prices
ema(38) of local high prices
ema(50) of local low prices

-tick.nq
daily from 9:30
1 min. chandle
ema(9) of closing prices
Bollinger(2,10)

From the "short skirts" thread I got an elegant entry idea:
http://www.enthios.com/trading/parallel_ma_band.htm
I have modified the entry rules of this system to fit my biased
ema setup:
-Do not trade when prices are within the band of the 38 High and
50 Low ema's.
-Do not trade when the 9ema is flat or changing direction.
-Look for longs when above the blue bands
-Look for shorts when below the blue bands
not enter

Risk management:
-max. market risk= 0.75%*capital;
-no overnight positions;
-no margin

Pos.Management:
-enter : stoplimit order;
-protect: (manual)traling stopmarket order;

Check each trading day:
begin day : cash
end day : trades, cash

Check order:
-new, fill and canceled:
Long/Short, quantity , price, position, result;

Accountsize:
Fund untill >26000
rebalance if > 50.000 to =45000

In have to think about more explicit protection/exit rules before Im ready to (paper)trade.

I will keep you informed.

 

[thetraderprofit]

Shark,

First, I've been to Holland twice in the past two years and I find
it difficult to believe you don't want long emails in English, as EVERYONE there speaks perfectly, except when you need to use their bathroom.

But, in response to your comments about a QQQ strategy, I will be very succint:

It ain't gonna work.

There are hundreds of millions of dollars being spent on programming by full-time traders trying to exploit any of the potential edges you perceive. You are not going to think up a new one in your spare time.

If any system will work on QQQ, it's the one between your ears. And I don't believe it will work when applied to a mechanical system.

Now, tell me how I can lower my 38.6% tax rate to 1.2% !!

 

[sharktooth]

thetraderprofit

Thanks for posting.

Your story about the bathroom made me produce a little chuckle.
You know the place. I dont know if moving down here is enough to get your tax to 1.2%, but if you love beeing in Holland, maybe some proffesional taxadvice will help.

Your stating that a pure mechanical system on qqq wont work.

I have to agree, that the chances of me finding a 100% mechanical system are not that great. Me calling my method
"Mostly mechanical" was not an effective choice of words.

Luckilily im not married to mechanical systems. Im willing to apply any ammount of discretionary that is needed to achieve my goal.

That being said, the main intention of the chart setups I described is to give me a starting mechanical structure to work with. At best, it will help me to decide if it's better to go long, short or wait a while.

If it works as intended, this particular structure will help me to address the flaw of being too discretionary. Being too discretionary can be very very very costly.

I dont expect the structure to tell me exactly when or where to enter with what size, where to place a stop and how to trail the stop.

I will have to develop profitable tactics for these issues during the journey.

Entry Long: up or above higher blue band
Long stops:
-Initial stop=0.07 cents below the bid.
-Traling stop
=I will tight stops using low last 2 completed candlesticks.
=max. 0.07 below last, if price spikes in my favor;
=breakeven, If last> breakeven and direction turns unfavorable

Entry stops Shorts: mirror above

Quote from thetraderprofit:

Shark,

First, I've been to Holland twice in the past two years and I find
it difficult to believe you don't want long emails in English, as EVERYONE there speaks perfectly, except when you need to use their bathroom.

But, in response to your comments about a QQQ strategy, I will be very succint:

It ain't gonna work.

There are hundreds of millions of dollars being spent on programming by full-time traders trying to exploit any of the potential edges you perceive. You are not going to think up a new one in your spare time.

If any system will work on QQQ, it's the one between your ears. And I don't believe it will work when applied to a mechanical system.

Now, tell me how I can lower my 38.6% tax rate to 1.2% !!

 

[jaan]

Quote from sharktooth:
Prices tend to fall faster then they rise.
I have no proof for this statement. It is based on my observations of prices developing over a period of 2 years.

i suggest that you do a simple backtest before you begin. that'll save you a lot of time, because i don't think your claim is true for QQQ.

- jaan

 

[AAAintheBeltway]

Sharktooth,

I'm not sure how well your strategy will work but I do think you have some sound ideas. First, I think it is a good approach to use a filter to indicate if you should be using long or short trades. I think it is better to use a longer timeframe chart like the 60 minute or daily for this however. You can often spot obvious support and resistance levels on the longer timeframe chart.

Second, I think there may well be something to the observation that markets fall more rapidly but you may find it difficult to use this. One reason is individual stocks often fall rapidly on negative news but they tend to do it with a gap opening. What I have found is that the stock market often will go up all day without a significant retracement, but it rarely will go down all day without a significant bounce. The reason probably has to do with shorts covering. I think it is true that most short sellers in stocks tend to be short term traders, while most longs are more long term oriented.

Third, I think it is prudent for you to be flat at the end of every day, particularly since you cannot watch the market all day. In a trending market this would be a serious handicap, but it may be a better strategy in this choppy market.

Since you can only trade in the afternoon most days, I would try to identify a pattern in which the activity for the first part of the day is strongly predictive of the way the market will close. Fortunately, we know that markets that trade strongly all day in one direction tend to go out at the extreme high or lows. If you can discipline yourself to trade only on days when you have an edge you should do well. If you try to chase every moving average cross, you most likely will not.

Finally, you should learn how to interpret the tick, trin and prem. They are fundamental to trading the stock indexes.

 

[sharktooth]

Net Paperprofits:
Monday 12/16: 3 one lot trades -11
Tuesday 12/17: 1 one lot trade +12
On tuesday a tight traillingstop made the difference.

Jaan
Because only the filter part (long, short or wait) is pure mechanical, I think the best way to test my ideas is to frontest(papertrade).

AAAintheBeltway
Valuable feedback. Keep it comming.

Besides the 3 min intraday charts I use now, I will also use a 3 day 30 min chart to spot the recent strong support and resistance.

I do use tick.nq. Trin.nq seems to be lagging. Have not used prem
till now. Besides the necessary charting of price I prefer to use as little indicators as possible. AAintheBeltway, if you could choose to use tick.nq or trin.nq or prem, which one would be first
on your list?

I will try to indentify patterns in which first part of the day is strongly predictive of the way the qqq will behave towards the close. All suggestions/directions are appreciated.

Clyde