Spread Betting Forum | Online Trading Talk-
http://spreadbettingbeginner.com/Forum/

A new spread betting strategy
http://spreadbettingbeginner.com/Forum/viewtopic.php?f=6&t=27		 Page 2 of 8
Author:  offthelip [ Thu Nov 25, 2010 3:40 pm ]
Post subject:  Re: A new spread betting strategy
I have just run a quick check using the parameters as defined at this moment, I back tested this against the FTSE 100 companies that were in the FTSE 100 all the time between Jan 04 and Aug 08. The results are interesting, the long condition occured 218 times, the short condition occurred 287 times. I recorded the closing price relative to the close on the day of the trigger for the next 30 days, the two plots below show the averages of the results. (Values across the top) There is definitely an edge there!!!!
Attachment:
Initial Test.jpg
Initial Test.jpg [ 175.21 KiB | Viewed 12943 times ]
Author:  SpreadBettingTrader [ Thu Nov 25, 2010 5:28 pm ]
Post subject:  Re: A new spread betting strategy
Hi offthelip. Where do you run your backtests? Do you do it in ProRealTime or have you written your own?
Author:  offthelip [ Fri Nov 26, 2010 12:21 am ]
Post subject:  Re: A new spread betting strategy
All the software that I use I have written myself. I use EXCEL with lots of VBA macros. I down load prices from yahoo,
MSN Money and digital look. Then I have written my own functions for lots of the standard things suchs as EMA, RSI , AVTR, ADX, Stochastics, so it is very easy (at least for me) to set up a spreadsheet with the historic data, a few functions, a little bit of logic , then iterate through it looking for the trigger conditions and copying the results to another spreadsheet. One thing I like about doing it all in EXCEL is that I am not limited by what the platform allows, one can do just about anything with VBA. I have looked at PRo real time but it is not as flexible as what I can do with EXCEL, e.g. it doesn't support pairs trading. I have thought about selling some of the software I have developed but, I have never got around to it and probably never will.
Author:  lotontech [ Fri Nov 26, 2010 9:52 am ]
Post subject:  Re: A new spread betting strategy
I would be more inclined to use back testing to try to disprove a strategy rather than to prove it works. It only takes a single negative proof to upset numerous positive proofs, which (by coincidence) I have alluded to today at http://tradingtrail.blogspot.com/2010/11/black-swans-and-turkeys.html

In Chapter 14 of my Financial Trading Patterns book I used back testing to show how a particular mechanical strategy based on trailing stops could massively outperform the FTSE in many cases. But a few minor tweaks to the trailing stop distance or time period could result in a massive under-performance :-(

If you can't break a strategy by back testing it, no matter how hard you try, it's a good thing. But be careful about concluding that "it definitely has an edge".

This is meant to be constructive, and I hope it helps.

Tony.
Author:  offthelip [ Fri Nov 26, 2010 10:46 am ]
Post subject:  Re: A new spread betting strategy
You absolutely right Tony, I have done a lot of backtesting and I have found it is very easy to find a strategy that produces positive results over a particular test period, however when testing it over an "out of sample" period the results are often completely different. This is particularly true if any optimisation of the strategy is done when testing over the initial period. In this particular test I was only testing a very small part of the strategy which was the setup, and it did generate a sufficiently large sample number to be significant. However 04 to 08 was mainly a bullish period so I would expect a positive return on longs with a random entry point. ( I haven't run this test but I might, just to compare). This does appear to be the case in that the short trigger goes positive shortly afterwards. This test was run without any optimisation of the strategy, so it might work in other periods. I think you are correct in pointing out that at this point all this test proves is that the strategy shouldn't be rejected at this stage. This sort of back testing is very easy to do for testing entry conditions, it is much more difficult to do for testing exit conditions. As you pointed out in your earlier post there is a lot more to a strategy than finding an entry condition that gives you at least a reasonable chance of making a profit.
Author:  SB_Beginner [ Fri Nov 26, 2010 2:55 pm ]
Post subject:  Re: A new spread betting strategy
Off the lip thanks for the back testing. To be honest I don't really understand quite what the charts you have posted indicate. While I think it's great that you have done this and I appreciate it, I think we are getting a head of our selves. We only have a very small part of our strategy so far.

I think we are probably happy that we have our setup criteria established.
Buy criteria:
Price > 200 day EMA
200 day EMA rising
RSI(14) < 30 (oversold)
Volume decreased for two successive days
Price decreased for two successive days

Sell criteria:
Price < 200 day EMA
200 day EMA declining
RSI(14) > 70 (overbought)
Volume Decreasing for two successive day
Price increased for two successive days

I have added these setup criteria to a new page on the site which you can view here. I've done this so these forum posts don't start getting too long. When we are happy with each section of our strategy I will add it to the page. We can name the strategy at a later date.

Next on the list is Entry triggers. Does anyone have any thoughts on these?

• Setup - DONE
• Entry triggers
• Stop placement
• % of funds to risk
• Position size
• When to move stops
• When to exit the trade
• Length of the trade
• Trade parameters to log
Author:  SB_Beginner [ Mon Nov 29, 2010 2:44 pm ]
Post subject:  Re: A new spread betting strategy
I've been thinking about the entry triggers for our new spread betting strategy. I don't really have that much experience with creating entry triggers so the only thing I can suggest is we use something similar to what Malcolm Pryor suggests in both of his books.

If we go long when the price exceeds the high of the previous bar and go short when the price drops below the low of the previous bar.
Or,
We could also say when the price exceeds the high of the last N bars or drops below the low of the last N bars. (Where N would be any number of bars)


What does everyone else think to this idea for an Entry trigger? What value of N should we use?
Author:  offthelip [ Mon Nov 29, 2010 4:55 pm ]
Post subject:  Re: A new spread betting strategy
I would vote for just one day i.e. buy if over the high of yesterday, my reasoning being we have identified a pull back with RSI < 30 and the price dropping for 2 days, I think if we wait for the price to increas back to where it was three days ago, we will have lost all the slight advantage we might have got from the pullback. "off the lip"
Author:  SB_Beginner [ Tue Nov 30, 2010 11:53 am ]
Post subject:  Re: A new spread betting strategy
Off the lip I think that makes perfect sense. So our entry triggers will be as follows:

Buy:

When price passes high of previous bar


Sell:

When price passes low of previous bar.

Next on the list is stop placement. I'm of the opinion that one should always use a stop when spread betting. Even if it isn't guaranteed, I still think it's a good idea. The question is where do we place our stop? (we'll get onto if we want to trail it later). Anyone have any thoughts. To quote Malcolm Pryors books again he suggests setting a stop below the low of the past N bars. (I've just noticed I've said bars everywhere. I'm assuming we will be using a daily chart for this strategy based on or setup. So one bar = 1 day). We could set a stop below the low of the past N bars or maybe just use some factor of ATR or something else. What are peoples thoughts on this?
Author:  lotontech [ Tue Nov 30, 2010 2:04 pm ]
Post subject:  Re: A new spread betting strategy
Where to place those stop orders might also be a function of your position size, in the sense that risk = stop distance x position size. On a four-digit index you would need a very small position size with a very tight stop distance if you wanted to keep your risk small, e.g. limited to £15. In contrast, as I blogged this morning, with a £1-per-point bet on Yell at about 15p there would be little point having a stop order at all.

This may be too much to consider at this stage, and I think you're asking about optimal stop placement (to reduce the probability of stopping out) irrespective of money management criteria (i.e. whether you can afford a stop at that level).
Page 2 of 8 All times are UTC
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/