Dow 96 point entry

[jpwone]

purpose of the thread to discuss the 96 point entry system raised on the IK thread.

<i>Go long on the Dow when it is up 96 points and short when it is down 96 points in any day. Stop if it reverses back through the previous day close and always close at the end of the day (no trades carry over and there are plenty of days with no trades entries). If you back test this you will find it is very profitable. I have not forward tested it. It might not work in the real world but at least it wont break you finding out.</i>

Excel spreadsheet of backtest is here spreadsheet

<i>I have found that usually i will get the entry point within a point or 2. The exit is at actual close so no spread if trading daily cash. Today I missed the entry by 5 points (target entry was 9878, actual entry was 9883). On occasions you will beat the entry price by a few points so I'm usually not too bothered if I'm within a few points. To be honest you can enter a couple of points earlier and this will more than compensate for the spread and slippage. The danger with entering too early is that there is more probability of a reversal.</i>

 

[esiotrot]

Thanks for the reply John
I'm getting more interested in these mechanical trading systems everyday. I'll work through yours and get back to you.

 

[daps]

Hi jpwone

thanks for the spreadsheet, have just looked at it. looks very interesting.

so entry is if dow hits +/- 96. ..?

but on the spreadsheet you mention if high great er then ave high finish then buy and vice versa.

ave high is 96.80
ave low is -98.26

and entry levels are 96.....so dont understand that part.

do you have figs (spreadsheet) for 2002-2003.?

regards

Daps

 

[swagbag]

I have looked at your spreadsheet it looks very interesting, you have buy & sell points and buy & sell points with a stop loss. I cannot see what the stop loss is. It may be staring me in the face but I could not see it. Could you please clarify.

Thanks.

 

[daps]

stop loss is 96 once in trade, hence previous days close.

 

[Glenn]

John
Thx for your reply.
Fair enough on the spread.
Cheers
Glenn

 

[jpwone]

Daps

I settled on 96 as the entry as I found it made little difference to the backtest if I stuck strictly to the average but made implementing it in the real world a bit easier.

No figures for 2003 but I will see if I can get the data. Previous data source was Reuters and I had checked it for accuracy. If anyone has access to the Reuter data for this period I will add it to the spreadsheet.


Swagbag

Stoploss is at previous days close. It is , in effect, a 96 point stop. Seems extreme but it worked in the backtest so I've stuck with it. With a system of this type i suspect that you need a wide stop as whip saws around the entry point would quickly eat any potential profit.


HTH

John

 

[snip]

John

Did you develop this system yourself.? Just wondering why +/-96pts. I guess it must be the optimum point between trade continuing and reversing, but how did you find this, was it just trial and error back testing different values.?

Have you tried anything similar on the other index's.?

By the way, the order I placed this morning on Capital Spreads demo has just been executed at 9882... Lets see how it goes..

Cheers Snip

 

[snip]

John

Looks like you beat me to it with your post above..

I have Dow data going back to 1988. If you want it, PM me and I'll attach the spreadsheet. Not sure it's Reuter data though.

Cheers Snip

 

[jpwone]

Hi Snip

yes - all my own work. Don't really know why I'm giving it away as it represents 4 years of testing and refining and trial and error.

Not tried it with the other indices yet. Was going to look at the Dax as a starter but often the other indices are influenced significantly by the Dow. Might as well trade the whole dog rather than just the tail.

HTH

John

 

[sidinuk]

John,

It's great to see someone develop a system themselves and publish it on the board.

I think you need to make the entry criteria more dynamic to take into account different market conditions. You have chosen 96 as that is the average movement over the whole sample period.

This presents 2 problems. One is that the parameter is a factor of the actual sample data which reduces it's reliability going forward. Secondly, a fixed factor doesn't take into account varying trading ranges, the trading ranges are a lot tighter at the moment than they were during your sample.

The solution? Well I would suggest making the factor a function of recent activity. For example take the average of the last 20 days open-close ranges and use that for today's session rather than a fixed value.

I have altered the s/s to do this and the result become 7033 (no stop) and 8969 (with stop) against 5637 and 8612 originally.


Not only are the results better but it will make the system more robust in the future.


I suggest you play with different periods for the averages and see what works the best.

 

[esiotrot]

Hi John
Very interesting!
I've been working on something very,very similar but had not twigged using the average change in the Close. I was using Pivots but it got a little complicated. I also had not considered trading long and short on the same day.

Are your figures DJI or the Dow future? The only reason I ask is that to trade mechanically you would need two different instruments - for example long on Dec future and short on Jan future as Finspreads offer. (i.e. not the actual future but using SB) Working the spreadsheet on the Cash and then trading a future might distort the results.

Interesting to see what money management does to the gross result.

BTW - what version of Excel is that. I like the fixed rows 1-20 and the ones below scrolling, very neat. I just slide the splitter bar to get two windows.

 

[quint]

John,
Always interested in new ideas. Thanks a lot for sharing yours. Logic suggests the key to trading the indices is going to lie in a relatively simple system. Yours looks promising. Uses end of day? and a realistic stop loss (though a few of those in a row could be very worrying). One question. You didn't use Yahoo Finances' Historical Prices did you?

 

[jpwone]

Hi all

it is very much a work in progress so any changes comments etc. are welcomed.

Quint
I'm a simple kind of person. Seriously, I looked at all sorts of systems and this was originally an attempt to predict reversals. It lost money in the backtest so I reversed the trade direction and then refined it! It uses Reuter data which was cross checked against another reliable source.

esiotrot
The long and short on the same day is on the same instrument (DJI). It means the stop must have been hit on one trade and the reversal has continued far enough to put it into the opposite trade. Its the latest version of Excel. It should work in previous versions. Select a cell, select window (on toolbar) and select freeze pane.

sidinuk
absolutely agree about the need to make it more dynamic. My idea was to take two averages. One would be the overall average and the other would be a recent (3 month?) average. The average used would then be the overall with a weighted recent average. This would give the recent activity more emphasis. Be nice to make it better.


John

 

[mmillar]

jpwone,

I've been doing similar (though not identical) work on the S&P and Nasdaq. It's funny that I've come to the same conclusions via the same route - I was writing an anti-trend system but noticed that once a market is up by x during a day then it is very likely to keep going up. And vice versa for down. Very simple system that is easy to back and forward test.

Good luck.

 

[esiotrot]

John
Looks like you're on to something here.

Thanks for the tip on freezing panes - works fine.

Keep up the good work guys.

 

[bonsai]

do I understand the system ?
today you would have gone long at 9878 for +21pts

but if before the close it had fallen to 9782
you would close for -96pts.

I make no comment on todays performance, I just want to know if I understand the system.

edit
my arithmetic had gone walk about.

 

[jpwone]

Hi all

just a quick recap of the system so that we all are working from the same starting point.

The system is designed to clearly define:-

<b>The Trade Entry</b>
Long if actual plus 96 points from previous close.
Short if actual minus 96 points from previous close.

<b>The Stake Size</B>
The stake size is 0.05 percent of the trading pot. This is chosen to ensure that losing streaks do not break the trading pot. The stake size is kept very modest as the stoploss is relatively large.

<b>Stop Loss</b>
The stop loss (96 points) is chosen to ensure that reversals do not stop a trade prematurely.

<b>Trade Exit</b>
The trade is exited at the close of trading for the Dow on the day of trade entry if not already stopped out. Trades do not rollover.

If adjusting/modifying the system all elements should be considered. Eg. changing the stop level could allow a larger stake size.

John

Ps. mmillar, thanks for the encouragement . Good luck with your own system.

 

[jpwone]

bonsai,

we overlapped there. Honest, I was writing the summary as you posted

<i>today you would have gone long at 9879 for +20pts

but if before the close it had fallen to 9782
you would close for -117pts.</i>

Entry is correct but would have stopped out a losing trade for minus 96. Backtesting showed that the stop was not hit too often. Its over a year since I looked at the system in detail but if I get a chance I will do some more analysis ie. win to loss ratio, number of losing trades that are at stoploss level etc.

John

 

[bonsai]

that's Ok John
my arithmetic was appalling anyway.
had to go back and edit it.