Using a trading system to daytrade

I agree with dozu, your results might become more stable with a 4pm instead of 4:15 exit. They may or may not "improve" depending on how many post-market spikes went your way during your test period.

The problem is how to exit at 4pm using a 135m bar. You could try using Robert Linders' Pushpop code. Have the 135m chart push its data over to a 15 or 5m chart on which the system runs. Then you could exit at 4.

This would also allow you to monkey with different stop methods if you choose. Are you positive the 2pt stop is really working? Sometimes TS has problems with trades that are entered and then exited or stopped on the same bar.

Anyway, pushpop has lots of potential uses. You can find it at http://www.traders2traders.com Just do a search for pushpop.
 
Good suggestions.

I'll quickly download the SPX and run a backtest on it. Since it closely mimics the SPooz and ends at 4, the current code will work.

I've checked the trade report $500 stop and it seems to work (losses reported at $500 or $625, depending upon whether comm./fees are figured in).

While the performance isn't stellar, it does have a fairly straight equity curve. Using a daytrading only system would allow a higher margin/leverage situation. One broker I spoke with will allow trading the SP full ctx at $5.5k for the day.
 
Don't want to get the thread off rcreal's contribution.... but took a quick look at the oddball code:

Inputs: RL(7), BZ(3), SZ(1);
If ROC(Close Data2, RL)>BZ Then Buy;
If ROC(Close Data2, RL)<SZ Then Sell;

Can someone explain why 3 and 1 are selected for the trigger levels? I thought as long as ROC > 0, the market breadth is still positive, then how come ROC<1 is a bearish signal ?

What am I missing?
 
Originally posted by dozu888
Don't want to get the thread off rcreal's contribution.... but took a quick look at the oddball code:

Inputs: RL(7), BZ(3), SZ(1);
If ROC(Close Data2, RL)>BZ Then Buy;
If ROC(Close Data2, RL)<SZ Then Sell;

Can someone explain why 3 and 1 are selected for the trigger levels? I thought as long as ROC > 0, the market breadth is still positive, then how come ROC<1 is a bearish signal ?

What am I missing?
More...
 
Below is quoted from the link posted by m_c_a

One way to construct an oscillator-based system is to trigger trades when the indicator crosses above and below the “zero” line (the median line that represents neutral momentum, when the current price is the same as the price n periods ago). But a better alternative is to use two separate indicator levels, or zones — one to initiate long trades and another to initiate all short trades.

A good initial setting is to set the buy level to 3 percent, and the sell level to 1 percent. That is, you buy as soon as the rate of change of the advancing issues is 3 percent higher than it was seven periods ago and sell as soon as it falls below 1 percent higher than it was seven periods ago. (See “Strategy snapshot,” below, for the precise formula for the indicator.) This means the system will always be in the market, either with a long or short position.
[unquote]

when ROC is less than 1, say ROC=0.5%, that means the ADVN is still 0.5% HIGHER than 7 hours ago! how would you go short when the market is still bullish?

it would make more sense to me if it goes long when ROC>3 and goes FLAT when ROC<1, but not going SHORT when ROC<1.

Inputs anybody ??????

:confused: :confused:
More...
 
Remember, oddball is "always in"...it only switches from long to short and back to long. So there is a tradeoff between being having more confirmation of your next trade being in the right direction, and letting profits slip away from the current trade.

"when ROC is less than 1, say ROC=0.5%, that means the ADVN is still 0.5% HIGHER than 7 hours ago! how would you go short when the market is still bullish? "

It is a question of momentum. You are shorting a market because bullishness is fading, in an attempt to get a better entry. Besides, the parameters aren't set in stone. Perhaps a zone of 4 and -1 will give better results. Or no zone at all, just use a break of 0. Test it and see.

Plot a chart of SP and ROC(ADVN) and you will see what is going on. Is it a grail? Certainly not. It did a fantastic job of picking the tops and bottoms of the past 4 swings. But other times it has ugly problems.
 
Dozu, when speaking w/the author of Oddball, he disclosed his theory on the 1 and 3 buy/sell points.

Basically, he said there was a shift in the data point distribution to the positive side since the market has a tendancy to go up. Hence, the buy/sell points are a little "off center".

Hope this helps.
 
Originally posted by rcreal
Dozu, when speaking w/the author of Oddball, he disclosed his theory on the 1 and 3 buy/sell points.

Basically, he said there was a shift in the data point distribution to the positive side since the market has a tendancy to go up. Hence, the buy/sell points are a little "off center".

Hope this helps.
More...

Ok, that answered my question.... (sort of).. I've been wondering that with the 1/3 set as sell/buy points, the system actually has a short side bias (only need 1 to go short, as supposed to something like -3) and if it was curve fitted to the down year in 2001.

Since I dont have tradeStation nor access to ADVN historical data, I can't test how oddball performs with a NEUTRAL bias, i.e. set buy/sell points at 1/-1, 2/-2 or something like that. I think 3/1 setting might run into trouble in a bull market.
 
some dude on Chuck LeBeau's site back tested oddball using 4 years of data.. the system got killed in the bull market of '98-99, exactly confirmed my concern of it's curve-fitted bearishness in my post above.

So it's a little BSish to say the 3/1 setting is based on data-point distribution.

However, I still think rcreal's logic is valid... basically an intraday trend-following on higher high/close, confirmed by market breadth.
 
dozu888,

I'm not here to argue. I'm just stating what Mark Brown told me when I spoke to him a few months ago.

The conversation was about 1.5 hrs, there was a lot of information he discussed so the details may not be 100 correct.

Any way, as I recall, he said it's his theory or understanding. Did he optimize it? I don't know.

People are welcome to speculate, hypothesize, criticize, whatever. I'm only interested in the facts. MB put up some code for free, so people got their money's worth.

I just presented some code for a *possible* daytrading system for the SP along with some preliminary test results. I wouldn't complain about $30k in one year, even with $125 slippage/comm per trade.

It seems to work well now. Will it work tomorrow? Next month? In two years? I don't know. I've got several other good, robust systems to use so I won't be exploring this one.

If you have any questions about the oddball system PM me.

Cheers,

Rich
 
You must log in or register to reply here.
Back
Top