Jonnyt is absolutely right. The system has to be tested using intraday data and has to be tested using actual futures data. I did the same as Jonnyt a while ago, had a fantastic system based on the cash which went t*ts up in real trading due to the bigger swings on the futures. Or as some call it 'the spreadbet out to get me bias'!
As a novice to trading and TA, can anyone recommend free! sources of
1. Dow and Ftse Futures data
2. Intraday data (pref 5 min)
Currently learning to chart with free data from MSN (OHLC) but only get actual DOW and FTSE.
Any help greatly appreciated.
Rgds Stephen
1. Dow and Ftse Futures data
2. Intraday data (pref 5 min)
Currently learning to chart with free data from MSN (OHLC) but only get actual DOW and FTSE.
Any help greatly appreciated.
Rgds Stephen
Sidinuk
Maybe I'm misunderstanding the chat about the futures and it's effect on the cash price with SB companies as relates to the subject of this thread.
With stocks, they bias so bad it is often impossible to trade, but apart from the opening, throughout the day even D4F mirror the actual Index within a couple of points. This would not make much difference to the spreadsheet results and as any entry would be more than 5 mins away from the open, you would be able to enter at near as dammit the index price.
Darren
Maybe I'm misunderstanding the chat about the futures and it's effect on the cash price with SB companies as relates to the subject of this thread.
With stocks, they bias so bad it is often impossible to trade, but apart from the opening, throughout the day even D4F mirror the actual Index within a couple of points. This would not make much difference to the spreadsheet results and as any entry would be more than 5 mins away from the open, you would be able to enter at near as dammit the index price.
Darren
Darren,
It would and does make a difference.
The spreadsheet has not catered for gaps. Using a large figure like the 96 takes this out of the equation some of the time but not always.
The daily cash prices do not accurately reflect the cash index so you will also find sometimes the system will take a trade in reality when in backtesting it didn't. Add on the SB companies pivoting aroung there BID/OFFER and this is more likely.
The results from the spreadsheet are absolutely best case. Conservatively you would have to take 7 or 8 points off per trade and that reduces the returns very dramatically.
Add this to the fact that you would have lost at least £1400 per point trading this since April I have no faith that this is a viable long term (or short term) trading strategy.
I have in reality traded very similar systems to this and am speaking from bitter experience. The drawdowns will be large, the question is can you handle them?
JonnyT
It would and does make a difference.
The spreadsheet has not catered for gaps. Using a large figure like the 96 takes this out of the equation some of the time but not always.
The daily cash prices do not accurately reflect the cash index so you will also find sometimes the system will take a trade in reality when in backtesting it didn't. Add on the SB companies pivoting aroung there BID/OFFER and this is more likely.
The results from the spreadsheet are absolutely best case. Conservatively you would have to take 7 or 8 points off per trade and that reduces the returns very dramatically.
Add this to the fact that you would have lost at least £1400 per point trading this since April I have no faith that this is a viable long term (or short term) trading strategy.
I have in reality traded very similar systems to this and am speaking from bitter experience. The drawdowns will be large, the question is can you handle them?
JonnyT
darren,
Futures move around a lot more than the cash index and the futures data will show exactly what levels have been traded. For example the cash index might show the high for the day at 9950 and the low at 9850 but the futures might show a high of 9953 and a low of 9847 (ignoring the fair value adjustment). D4F prices will follow the futures so at the high the quote might be 9950 - 9955. Say you have a stop to enter the market at 9951, the cash won't have hit it and your back test won't include it but D4F would have dealt the stop based on their quote. You would, therefore, have to put your stop at 9955 to ensure that the cash index did actually hit 9950 before you got into the trade. 5 pts on a trade doesn't sound much but trade 200 times a year and it takes 1000 pts off of your 'cash index' back test results.
The only way to be sure of your backtest results is to use the actual prices from the instrument you are trading. This is easy if you are using futures as this info is available, with spreadbets it isn't. So if you are spreadbetting you need to make proper adjustments for it. It's easier to make those adjustments from the futures prices as you know that the SB quote will straddle those prices whereas you won't really know exactly how they relate to the cash index. An extreme example could happen on a futures spike. The futures could spike down from 9100 to 9000 then back to 9100 within a minute, the cash might only have time to go to 9050 before the correction back.
Futures move around a lot more than the cash index and the futures data will show exactly what levels have been traded. For example the cash index might show the high for the day at 9950 and the low at 9850 but the futures might show a high of 9953 and a low of 9847 (ignoring the fair value adjustment). D4F prices will follow the futures so at the high the quote might be 9950 - 9955. Say you have a stop to enter the market at 9951, the cash won't have hit it and your back test won't include it but D4F would have dealt the stop based on their quote. You would, therefore, have to put your stop at 9955 to ensure that the cash index did actually hit 9950 before you got into the trade. 5 pts on a trade doesn't sound much but trade 200 times a year and it takes 1000 pts off of your 'cash index' back test results.
The only way to be sure of your backtest results is to use the actual prices from the instrument you are trading. This is easy if you are using futures as this info is available, with spreadbets it isn't. So if you are spreadbetting you need to make proper adjustments for it. It's easier to make those adjustments from the futures prices as you know that the SB quote will straddle those prices whereas you won't really know exactly how they relate to the cash index. An extreme example could happen on a futures spike. The futures could spike down from 9100 to 9000 then back to 9100 within a minute, the cash might only have time to go to 9050 before the correction back.
continuing with the cash / futures arguement;
I developed a couple of systems with very similar trading rules to this one (ie, open a position once a preset level had been breached, close EOD) and was concerned about this very issue.
For several months I ran the systems with both the cash quote and the futures price (real futures NOT SB prices) and found that in general the difference in price made no difference to the overall profitability of the system.
It also made little difference whether I traded the futures using the cash to trigger the trade or used the futures as a trigger.
Trying to use the out of hours futures price action didnt help at all, in fact both systems were more profitable if I only included the futures prices from 1430-2100 in my calculations.
good luck all
Neil
ps, agree with other posts re the increased volatility of the futures.
I developed a couple of systems with very similar trading rules to this one (ie, open a position once a preset level had been breached, close EOD) and was concerned about this very issue.
For several months I ran the systems with both the cash quote and the futures price (real futures NOT SB prices) and found that in general the difference in price made no difference to the overall profitability of the system.
It also made little difference whether I traded the futures using the cash to trigger the trade or used the futures as a trigger.
Trying to use the out of hours futures price action didnt help at all, in fact both systems were more profitable if I only included the futures prices from 1430-2100 in my calculations.
good luck all
Neil
ps, agree with other posts re the increased volatility of the futures.
Last edited:
Going to play devils advocate here for a moment.
Does it really matter which instrument (cash or futures) we use for development if what we are doing is proving the concept? The concept will succeed or fail with either.
Does it matter that it has lost money this year? If someone offered you a system which had a relatively small loss once in every 4 years but was significantly profitable for the other 3 would you refuse/reject it? Ifsomone offered you a system which (on average) broke even for 3 out of 4 years but made a significant profit in the 4th year would you reject it?
Does it really matter which instrument (cash or futures) we use for development if what we are doing is proving the concept? The concept will succeed or fail with either.
Does it matter that it has lost money this year? If someone offered you a system which had a relatively small loss once in every 4 years but was significantly profitable for the other 3 would you refuse/reject it? Ifsomone offered you a system which (on average) broke even for 3 out of 4 years but made a significant profit in the 4th year would you reject it?
Bigbusiness
Experienced member
- Messages
- 1,408
- Likes
- 23
If it isn't working now and has only been tested for a few years then I would be cautious. If it worked over a 10 year period, then the last 6 months wouldn't matter so much. I am also a bit concerned about using OHLC data to test a system that trades during the day. It would be interesting to see how well it does using 1 minute data.
But it looks a lot better than some of the systems I have seen being sold on the internet.
jpwone
I suspect there is a danger while developing this system that you might get bogged down in what may turn out to be pedantic detail.
On cash v futures, there are enough abitrageurs to keep fair value well within bounds.
If you are concerned, just track it for a few days.
On Trading hours, once you move outside official cash hours, you might as well make it 24 hrs. Globex;SB's etc all trade outside official hours.
On Gaps, I only mentioned them as part of the stop loss issue.
It seemed to me that on gap days, the energy may have been taken out of the trading market by the gap and so prospect of reversal might be greater. A lot of markets like to fill their gaps.
Apart from that issue, I would ignore them. Anyway its mainly a cash market issue. (There may not be a gap in a 24 hr market ?)
So, for what its worth, I suggest you work around the cash market. HLCO is indisputable because of official hours.
And I agree with you, it doesnt matter if you have a losing run.
All systems have losing runs !
If only our forsight was as good as our hindsight ?
I suspect there is a danger while developing this system that you might get bogged down in what may turn out to be pedantic detail.
On cash v futures, there are enough abitrageurs to keep fair value well within bounds.
If you are concerned, just track it for a few days.
On Trading hours, once you move outside official cash hours, you might as well make it 24 hrs. Globex;SB's etc all trade outside official hours.
On Gaps, I only mentioned them as part of the stop loss issue.
It seemed to me that on gap days, the energy may have been taken out of the trading market by the gap and so prospect of reversal might be greater. A lot of markets like to fill their gaps.
Apart from that issue, I would ignore them. Anyway its mainly a cash market issue. (There may not be a gap in a 24 hr market ?)
So, for what its worth, I suggest you work around the cash market. HLCO is indisputable because of official hours.
And I agree with you, it doesnt matter if you have a losing run.
All systems have losing runs !
If only our forsight was as good as our hindsight ?
Hi Glenn
Reason for closing at the end of day was twofold.
1. Risk reduction. Was only in the market when it was open.
2. The luxury of an unbiased middle of the quote price for the exit
It took one potential spread payment out of the equation.
John
Reason for closing at the end of day was twofold.
1. Risk reduction. Was only in the market when it was open.
2. The luxury of an unbiased middle of the quote price for the exit
It took one potential spread payment out of the equation.John
JT - absolutely agree with u that the ss must use the futures prices - the opening price is essential to see if this system is going to work.
My ss doesn't enter a trade if the trigger is within the opening gap - at that point one has to decide whether to go for it or not.
It doesn't cater for intraday swings, but hopefully a large enough stop will sort that out - what that stop is, I don't know. Will have to look at BB's stats.
With that in mind, using average of 55 as trigger, the ss shows a profit over the period u mention.
I did a trial trade yesterday - trigger 55. Entry was 9908. Using CM's criteria, no way would I have gone long there. Anyway, was in profit for a while before dow dropped back - didn't hit the 9853 stop though.
Maybe a case of using a trailing stop here to lock in profits - but that would defeat the purpose of a mechanical system.
I think this system will have it's work cut out with current mkt conditions - it will need more manual intervention (such as trailing stops) to make it work.
Also, although it's essential to use the futures opening price (more often than not, the cash catches up), u r not trading futures - u r doing the cash thru spreadbets - that takes the spikes out of the equation - I know the spreadbetters base their cash on the futures, but it doesn't seem as volatile.
Glenn - 10 point triggers will lose u money very quickly
My ss doesn't enter a trade if the trigger is within the opening gap - at that point one has to decide whether to go for it or not.
It doesn't cater for intraday swings, but hopefully a large enough stop will sort that out - what that stop is, I don't know. Will have to look at BB's stats.
With that in mind, using average of 55 as trigger, the ss shows a profit over the period u mention.
I did a trial trade yesterday - trigger 55. Entry was 9908. Using CM's criteria, no way would I have gone long there. Anyway, was in profit for a while before dow dropped back - didn't hit the 9853 stop though.
Maybe a case of using a trailing stop here to lock in profits - but that would defeat the purpose of a mechanical system.
I think this system will have it's work cut out with current mkt conditions - it will need more manual intervention (such as trailing stops) to make it work.
Also, although it's essential to use the futures opening price (more often than not, the cash catches up), u r not trading futures - u r doing the cash thru spreadbets - that takes the spikes out of the equation - I know the spreadbetters base their cash on the futures, but it doesn't seem as volatile.
Glenn - 10 point triggers will lose u money very quickly
What I found looking at this system as it stands is that you will need an initial stop for when you enter the trade initially- I like to use 27 as I have found that if the trend is correct most of the time it will retrace to around 25 to 27 before continuing.
However, the times when this scenario is incorrect it would be wise to stop your losses before it gets out of hand.
If the trade goes into profit you can then increase the stop loss to a higher figure or break even as a worse scenario.
Looking at some of the past figures I think a good stop loss is definitely required to lock in profits.
I.e if the market move up 180 points most days you will find that it has dropped back to maybe 30 to 60 pts or more so having a stop loss would have lock in a fair amount of profits.
Whereas if you left it to close out at the EOD you would have lost unnecessary profits.
However, the times when this scenario is incorrect it would be wise to stop your losses before it gets out of hand.
If the trade goes into profit you can then increase the stop loss to a higher figure or break even as a worse scenario.
Looking at some of the past figures I think a good stop loss is definitely required to lock in profits.
I.e if the market move up 180 points most days you will find that it has dropped back to maybe 30 to 60 pts or more so having a stop loss would have lock in a fair amount of profits.
Whereas if you left it to close out at the EOD you would have lost unnecessary profits.
RRG,
It's horses for courses I suppose. My preference would be to have a mechanical system running in the background (building a pension fund) requiring minimal intervention.
When I'm able to watch the market closely I prefer to time entries/exits based on TA - currently studying volume analysis, which I'm finding more lucrative than the lagging indicators.
It's horses for courses I suppose. My preference would be to have a mechanical system running in the background (building a pension fund) requiring minimal intervention.
When I'm able to watch the market closely I prefer to time entries/exits based on TA - currently studying volume analysis, which I'm finding more lucrative than the lagging indicators.
Similar threads
