Lets all join the Millionaires CLub

FetteredChinos

Veteren member
Messages
3,897
Likes
40
Afternoon all,

Welcome to Chinos' trading strategy mark 6. Thought i would throw this one out to you guys for some critique.

V6 is again simple to operate, and the usual caveats apply about using yahoo historical data, but i have been simming it for the last few days on paper and its looking quite promising.

Looks like we are getting in the region of 35 points a day. recent performance hasnt been too hot - down nearer 25 a day, but still pretty solid!! and best of all, it is ludicrously simple, and is based upon support and resistance levels. curiously though, and this is a big worry, is that it only seems to work on the US markets, as the FTSE and the Nikkei prefer the opposite entry rules. ho hum.

Still, im fairly confident that this isnt a statistical aberration, and the difference is instead due to the time zones etc. as you can see from the attached spreadsheet, i have gone back about 7 years and 2000 trades to verify this.

rules on this one are... wait for it....


1) in the morning check out the previous days DOW range.

2) set up limit orders to BUY at yesterdays low and SELL at yesterdays high.

3) often, owing to futures movements, the cash price is above/below these levels, in which case, say thankyou very much to your spreadbet company and place the bets there and then.

4) close all trades at market close. D4F dont do this, unless via telephone dealing, but more than a few of the other SB's have daily DOW bets.

5) there is no 5.


in the yellow box i have included a variable trigger figure (sounds good that doesnt it!!! :cheesy: ) seems as though something in the region of -10 returns the greatest number of points. ie SELL at yesterdays high -10 , or BUY at yesterdays low +10. this results in more trades (obviously) but at a lower return/trade. overall points though are increased.

now i dont know if im going totally mad on this, and obviously the exit strategy can be improved to improve total points further, but we seem to have a bit of a goldmine on our hands.

now i am aware that yahoo historical data is massively inaccurate in places (just look at the data for the past week. im 110% certain we havent hit some of those levels on a daily basis) - as a result i took the liberty of using some of the data that TBS had posted back on the Fool website a while back. seems as though this is meant to be the "definitive data" by all accounts. The data is only 100 days worth from 2001, but the returns from back then are even better!!! (higher volatility as we all know!!)

now where did i leave that speedboat catalogue??


FC
 

Attachments

  • reverse breakout.xls
    2.6 MB · Views: 1,005
Balls!!

if i had stuck to my "modified" version of selling at HIGH-10, then i would have been triggered on the big sell-off that is about to occur. as it is i went for the simple SELL HIGH and BUY LOW limit orders and as a result i am flat at the moment, and the BUY point looks a long way off!! always tomorrow i suppose!!!


FC
 
and there she goes... down to 10,672 so far...

oh well, markets will always be there, and retirement will have to be postponed by another day!


FC
 
lol cheers mate, look like the cat is coming back..

here kitty kitty.

watch it all go horribly and hideously wrong now!
 
Not sure if I understood.

SELL limit yesterday's HIGH and
BUY limit yesterday's LOW?

This assumes that each succesive day's trading range is within the previous?

I'm sure I've got this wrong as we'd have had an awfully flat line for an awfully long time...
 
Not sure if I understood.

SELL limit yesterday's HIGH and
BUY limit yesterday's LOW?

This assumes that each succesive day's trading range is within the previous?



Well not really..if u look at rule 3)

if the market is outside what would be the normal sell rules premarket, the u just place the trade and close it at market close..

maybe an example would be good...

the orders i set up for today..

i looked at the High/Low for the Dow on monday . it ranged 10704 - 10561 (give or take a few points. havent got the precise data in front of me) so this morning i set up my trigger limits..

i put a sell order in at 10704 and a buy at 10561

as it was, we had an inside day today/tuesday, and so neither trade was triggered. no matter, fingers crossed we get one on wednesday.

however, say, for example that the out-of-hours price of the DOW had climbed to 10,720 when you were putting on the trade limit orders. as the market is already outside the original level of 10,704, you would just hit a sell order there and then, and place the buy order in the same place at 10,561.. basically they are giving u an extra 16 points (in this example) for free.

as you can see from the spreadsheet, if the previous days high/low is exceeded, then there is about a 70%-odd chance of the market closing with the trade in profit.

going back to the example... say having put on the order at 10,720, the SB price then dives before the offical market open, and then never trades above the previous days high. in normal circumstances the trade wouldnt have been triggered, and we wouldnt be in profit at all. but in this instance, we have got at least 16 points, plus a few more (had the trade been triggered today with the market closing at 10609, we would have yielded 95 points, or 111 points if we had been triggered out of hours)

therefore the actual results obtainable via a SB company would probably be in excess of what is on my spreadsheet!!

curses. @£10 a point, that could have done me nicely


typically enough. neither trade was triggered today..

however, seeing that the DOW closed near the low of the day tonight, there is a pretty good chance we will see this in action tomorrow.

im not advocating u lot "following my trades" to quote another contributor to this board. but its just something that seems to have statistics on its side. and a 70% hit rate averaging 35 points a day is not to be sniffed at for a mechanical system that only needs a few mins a day.. My god where have i heard that before? maybe I should freelance for the Motley Fool lol!!

right i may well have to edit this post and turn it into something resembling proper English tomorrow, but this will have to do in the meantime.


take care guys, and enjoy the snow.

stay safe!

FC :)
 
TheBramble said:
This assumes that each succesive day's trading range is within the previous?

By using... =IF(OR(D13>C14,D14>C13),1,0)

you will see that there are no gaps in this data
Each day overlaps the next

Nice work FC
.
 
ok boys and girls, we've got a live one today..


have just placed my orders before another thrilling day at work..


yesterdays range was 10,609-10701...


am now using the limits+/-10 method after the shambles yesterday lol..

so ive whacked in my sell order at 10691 (10,701-10)

this morning the daily DOW cash was as 10582 when i logged on. normally i would have placed my order to buy at 10619 (10619+10) however the price was 37 points in my favour so of course i took that.


both orders close at market close.. stop set at 150 points..


fingers crossed chaps. i think with FOMC meeting today (is it??) that i may need a bit of luck.. i fancy another sell off personally. but a mechanical system must be obeyed!!!


lets see how this one pans out.


FC
 
Oh Dear

I've tested this with futures data, taking a day to be 09:30 to 16:00 and for the Sept 03 contract it lost massively.

JonnyT
 
??

...and on EOD since beginning of Dec '03 I make it a 46% win rate and overall loss - not particularly representative, but it doesn't bode well - see jpwone's thread for discussion on his mechanical system and the merits of using futures data vs EOD.

For the same period jpwone's system has generated a 3.5% return. I'm testing using capitalspreads' daily DOW, which has a 5 point spread.

Good luck!
 

Attachments

  • dowrevbodec.xls
    2 MB · Views: 385
Last edited:
I may have goofed!!!

I've modified my code slightly and now get a profit of 350 points.

Given its from 70 trades this is only 5 points per trade.

The code currently doesn't look at the intraday picture which is also a failing of the current spreadsheet (it actually must over estimate the real return as you have no way of knowing which came first the HOD or LOD)

I'm going to write some code to carry out a full intraday test with stop losses and profit takes etc to see if we can actually develop a tradeable system.

5 points per trade simply isn't worth the risk for such a system.

JonnyT
 
Err jonny, i dont see how it matters which came first.. if the high comes first, then we're short until either we hit the buy point which closes the trade, or the market closes.

if the low comes first, then we are long until the sell point closes the trade or the market closes. whichever comes first.

forgive my lack of comprehension, and i appreciate your considerable skill in system design, but i think we may have our wires crossed on this. mine is a blue one. is yours a red one?

Chow.. agreed that we had a few problems since Dec'03. but as you know, we have been trending like a beastie since then. climbing parabolically. the nature of this system is that it is there to make the most of choppy periods, when the market is dithering left right and centre. so i expect it to lose at the moment, but soon recoup its losses when this current trend loses steam.

there is only one way to find out if a system works, so im gonna stick my neck out and find out the hard way!

FC
 
Btw chow.. where did you get your data from??

i really could do with some clean data for once!

cheers


FC
 
Right I have some definitive results.

The system logic is as follows:

1) Data used is Dec 03 Mini dow future between 09:30 and 16:00 US time. i.e normal market hours as provided by MyTrack. 1 minute bars.

2) The system goes long when:
a) The low of the prior day is breached during normal trading hours.
b) The market at 09:30 is below the low of the prior day

3) The system exits a long at either 16:00 normal market close or when the high of the prior day is reached.

4) Reverse the criteria for short trades.

This net result is around 10 points per trade. In reality I expect 8 points per trade taking into account slippage and costs. I should point out this is nowhere near the published results.

Here are the resultd


Open High Low Close HighBreak LowBreak
15/09/2003 9447 9456 9403 9422
16/09/2003 9426 9537 9426 9529 -72
17/09/2003 9538 9557 9504 9524 14
18/09/2003 9527 9627 9502 9625 54 54
19/09/2003 9622 9649 9573 9608 20
22/09/2003 9526 9535 9462 9505 -21
23/09/2003 9512 9547 9470 9526 10
24/09/2003 9547 9556 9383 9389 78
25/09/2003 9408 9424 9302 9305 -77
26/09/2003 9314 9327 9268 9277 -24
29/09/2003 9301 9365 9257 9343 60 60
30/09/2003 9320 9320 9195 9252 -4
01/10/2003 9288 9438 9286 9428 -107
02/10/2003 9411 9481 9395 9468 -29
03/10/2003 9580 9635 9524 9546 34
06/10/2003 9550 9594 9529 9557
07/10/2003 9530 9620 9505 9613 66 66
08/10/2003 9626 9641 9564 9623 3
09/10/2003 9696 9738 9615 9662 34
10/10/2003 9659 9681 9626 9658
13/10/2003 9700 9750 9697 9733 -33
14/10/2003 9726 9781 9700 9778 -27
15/10/2003 9826 9828 9735 9776 50
16/10/2003 9708 9798 9703 9766 58
17/10/2003 9777 9794 9676 9699 -3
20/10/2003 9714 9746 9680 9740
21/10/2003 9741 9757 9707 9723 24
22/10/2003 9634 9651 9556 9583 -51
23/10/2003 9529 9599 9527 9580 51
24/10/2003 9502 9570 9469 9569 67
27/10/2003 9603 9637 9566 9575 28
28/10/2003 9615 9725 9594 9725 -87
29/10/2003 9697 9760 9690 9749 -23
30/10/2003 9811 9815 9730 9762 49
31/10/2003 9784 9815 9767 9771
03/11/2003 9805 9870 9804 9841 -25
04/11/2003 9817 9848 9798 9817 14
05/11/2003 9790 9844 9754 9790 0
06/11/2003 9793 9851 9750 9838 91 91
07/11/2003 9883 9886 9776 9789 94
10/11/2003 9789 9805 9717 9734 -41
11/11/2003 9723 9746 9700 9724 8
12/11/2003 9726 9850 9724 9841 -94
13/11/2003 9801 9845 9783 9825
14/11/2003 9826 9890 9727 9760 63
17/11/2003 9701 9725 9619 9698 -3
19/11/2003 9620 9707 9613 9694 76
20/11/2003 9628 9722 9605 9612 95 95
21/11/2003 9639 9641 9578 9635 31
24/11/2003 9668 9747 9665 9745 -77
25/11/2003 9740 9790 9718 9759 -11
26/11/2003 9786 9795 9699 9778 73
28/11/2003 9765 9799 9759 9799 -3
01/12/2003 9818 9899 9815 9891 -73
02/12/2003 9870 9887 9831 9847
03/12/2003 9885 9939 9867 9874 14
04/12/2003 9872 9930 9866 9927 61
05/12/2003 9890 9912 9841 9860 -5
08/12/2003 9848 9964 9848 9955 -42
09/12/2003 9982 9995 9908 9926 56
10/12/2003 9922 9953 9877 9920 13
11/12/2003 9935 10020 9934 10013 -59
12/12/2003 10014 10051 9970 10038 -17
15/12/2003 10143 10143 10018 10018 125
16/12/2003 10045 10133 10038 10120
17/12/2003 10106 10145 10089 10145 -11
18/12/2003 10168 10257 10149 10249 -81

264 516

JonnyT
 

Attachments

  • dow.xls
    2.1 MB · Views: 377
aha cheers mate. ta for the link, i'll give it a go...

in the meantime, using the data ive got already (for what its worth) ive knocked up an equity curve..(see attached)

this is based upon points only, and doesnt take into account actual poundage.

i know its the wrong way round - the most recent trades being on the left, and the oldest on the right, but im kinda busy at work at the moment, and i didnt have the time/motivation to reformat the chart.

still, its a fairly smooth curve with no major drawdowns. in the long run, we look like we have a winner. data notwithstanding.


btw, current bet is 40 points to the good so far. fingers crossed it follows through!

FC
 

Attachments

  • equity curve.xls
    83 KB · Views: 305
thanks for that Jonny. much appreciated.


is there any chance now you have the logic set up, of running your backtest on a couple of other contracts selected at random?

as i stated earlier, trending markets dont really help this system at all.

FC
 
hmm, now maybe some sort of filtering could be applied to fine tune this further.. i didnt want to end up using indicators or MA's to do this, but it seems that from Jonny's data that each LowBreak returns 20 points/ trade whilst the HighBreak returns 7/trade..

we are clearly in an uptrend, so perhaps filtering out only the breaks that are counter-trend could yield decent results. eg in a downtrend, we only look for false breakouts to the upside.

food for though.


FC
 
JonnyT,

Out of curiosity what would you say is an acceptable pts/trade rate for a daytrading system on say ES or NQ?
 
Top