FTSE system

[new_trader]

I agree JT

Test test test is my motto!

I'm afraid I can't help out with the data though, why don't you look to open an account with someone like www.interactivebrokers.co.uk , that way you'd be trading the exact instrument you are testing.

I don't have an account but am looking at opening one. It seems quite daunting to be honest so if anyone on this thread does or has experience of them please dive in and tell us if it's as complicated as it looks!

Otherwise if anyone has a good source for buying the data please let us know, I'm not against buying it it's just that I've never found a supply I'm happy with.

Thanks

Stephen McCreedy

Hi Stephen,

I purchase my tickdata from

http://www.tickdata.com/

They have a proprietary program (tickwrite) that you can download and try with about 3 months of free data. (US Data)

The company has been very prompt with delivery, usually available for download within 24 hrs of placing your order.

 

[smccreedy]

Thanks will go and have a look although (and I know everyone is sick of hearing this!).

I'd love to have enough capital to be able to make a nice 30% a year looking at end of day or even end of week data!

Imagine the bliss of £1,000,000 to trade with, £300,000 profits, living of as little as you can and watching it grow ten fold in a decade.

All whilst somewhere hot of course!

Stephen McCreedy

 

[smccreedy]

Here is a pic of the equity of a system trading the same rules on several different markets.

They span different time scales but generally from between 1990/2000 to the present day.

I think most people could agree that the last 15 years or so has seen it's fair share of bullish, bearish and sideways action and I do feel it is a decent time to test against for equities and indexes.

I generally look to systems with a mild bullish bias as markets and indexes generally speaking are in an uptrend overall. (if only because in the case of FTSE for example losing companies are cut out each 1/4 and replaced with winning ones giving a bullish bias over time).

I feel this mild bullish bias means that over longer periods these systems would continue to do well or even improve.

Now before the cynics get involved in saying how these are hand picked markets, time scales etc don't bother, they were the only instruments I could think of the tickers for off the top of my head, Barclays and BP were thrown out as stock splits rendered the data no good for me.

The very easy basis of this 'system' is this.

Using only closing prices, when the market closes at a 9 day low, buy, when it closes at a 9 day high sell. It's stop and reverse so you are always long or short.

I think on balance the drift on the equity is to the upside over time.

I'd be interested that if you covered say five markets, perhaps indices followed each signal how you'd perform over time. My gut feeling is that maybe around 10% pa compounding with no gearing.

The FTSE took 4000 points from a contract that started at 2000 in 16 years.

From my dodgy maths that makes a return of around 6.8% compounding.

Just an interesting exercise, I wonder how if can be improved without curve fitting.

Maybe use a lag filter so in bull markets it only took the long trade and in bear markets it only too the shorts.

Also there are no stops at all so using stops might be able to get the return up.


Stephen

 

[RunTheNumbers]

Here is a pic of the equity of a system trading the same rules on several different markets.

They span different time scales but generally from between 1990/2000 to the present day.

I think most people could agree that the last 15 years or so has seen it's fair share of bullish, bearish and sideways action and I do feel it is a decent time to test against for equities and indexes.

I generally look to systems with a mild bullish bias as markets and indexes generally speaking are in an uptrend overall. (if only because in the case of FTSE for example losing companies are cut out each 1/4 and replaced with winning ones giving a bullish bias over time).

I feel this mild bullish bias means that over longer periods these systems would continue to do well or even improve.

Now before the cynics get involved in saying how these are hand picked markets, time scales etc don't bother, they were the only instruments I could think of the tickers for off the top of my head, Barclays and BP were thrown out as stock splits rendered the data no good for me.

The very easy basis of this 'system' is this.

Using only closing prices, when the market closes at a 9 day low, buy, when it closes at a 9 day high sell. It's stop and reverse so you are always long or short.

I think on balance the drift on the equity is to the upside over time.

I'd be interested that if you covered say five markets, perhaps indices followed each signal how you'd perform over time. My gut feeling is that maybe around 10% pa compounding with no gearing.

The FTSE took 4000 points from a contract that started at 2000 in 16 years.

From my dodgy maths that makes a return of around 6.8% compounding.

Just an interesting exercise, I wonder how if can be improved without curve fitting.

Maybe use a lag filter so in bull markets it only took the long trade and in bear markets it only too the shorts.

Also there are no stops at all so using stops might be able to get the return up.


Stephen

Stephen

This system on these instruments are untradeable.

Regards

Ben

 

[smccreedy]

Thanks for your opinion but would go further and explain why you believe this to be so.

Do you mean the specific instruments I've picked or indexes in generally or pretty much any instrument?

When you say untradable do you mean for physical reasons of execution or something or do you mean unprofitable rather than untradable?

What are your reservations with it, do you feel I've messed up the back testing?

Any input would be great

Thanks

Stephen McCreedy

 

[rtsaur]

The very easy basis of this 'system' is this.

Using only closing prices, when the market closes at a 9 day low, buy, when it closes at a 9 day high sell. It's stop and reverse so you are always long or short.

Hi Stephen,

I can understand the logic of how your systems (1) and (2) work but sadly not the system (3). Please could you explain how do you get the idea of it.

Many thanks.

 

[smccreedy]

Hi everyone

Getting back to the original system this is a tweaked version.

Nothing amazing but have moved the buy and sell fraction of the PDC to 0.5% from 0.25% this might address the problem of not getting filled and markets gapping on the open.

Also I have included a double check where by the low of the day must exceed my sell stop by 0.1% (around six points) or the high of the day must exceed my buy stop by 0.1% before the spreadsheet assumes the orders got a fill.

This means the argument that some s.bet prices are skewed etc should matter less.

For example if the orders for today are buy at 6477 and sell at 6412 the market high and low must exceed these levels by around 6 points to assume the trade happened.

Of course these means there might be times that in reality I am stopped out for my maximum loss when the system doesn't think I have been which means some losers could be bigger than they are recorded as, however this problem seems to be smaller than the issue that people felt the system worked well on paper but was untradable for real.

Performance on the Dax and Ftse looks good and with sensible bet sizing of say 5% loss if both orders are hit I think it might be a good system to have ticking over in the background.

Stephen McCreedy

 

[JTrader]

Hi everyone

Getting back to the original system this is a tweaked version.

Nothing amazing but have moved the buy and sell fraction of the PDC to 0.5% from 0.25% this might address the problem of not getting filled and markets gapping on the open.

Also I have included a double check where by the low of the day must exceed my sell stop by 0.1% (around six points) or the high of the day must exceed my buy stop by 0.1% before the spreadsheet assumes the orders got a fill.

This means the argument that some s.bet prices are skewed etc should matter less.

For example if the orders for today are buy at 6477 and sell at 6412 the market high and low must exceed these levels by around 6 points to assume the trade happened.

Of course these means there might be times that in reality I am stopped out for my maximum loss when the system doesn't think I have been which means some losers could be bigger than they are recorded as, however this problem seems to be smaller than the issue that people felt the system worked well on paper but was untradable for real.

Performance on the Dax and Ftse looks good and with sensible bet sizing of say 5% loss if both orders are hit I think it might be a good system to have ticking over in the background.

Stephen McCreedy

Hi Stephen

Sorry but I'm now lost.

Constructive feedback time - I appreciate that you have gone to considerable effort to produce these modified system parameters. However, if i had gone to this much effort, i would have looked to add headers to the columns within the spreadsheet.
As it stands, I do not clearly understand your last post, or exectly what the spreadsheet is showing me.

Thanks for adding it though, maybe if i study it further, I will be able to make sense of it all.

 

[smccreedy]

Hi

From left to right

Date / Open / High / Low / Close / Space / Prev close -.5% / Prev close +.5% / Buy? / Long Profit / Sell? / Short Profit / Daily Profit / Running profit.

So prev close -.5% is the level at which we sell

Prev close +.5% is level at which we buy

Next column adds a Buy if the high level was exceeded by .1% and then next to it shows the profit or loss made on that trade buying and exiting on close.

Next column shows a Sell if low level was exceeded by .1% and then next to it shows the profit or loss made on that trade from shorting and exiting on close.

Next column shows total daily profit (notice that if both buy and sell are shown the daily loss will be the max loss ie the distance between the two signal levels)

Next is the running total ie previous day's running total plus today's results.

Hope that makes sense any more questions feel free to ask again.

Stephen

 

[JTrader]

Hi

From left to right

Date / Open / High / Low / Close / Space / Prev close -.5% / Prev close +.5% / Buy? / Long Profit / Sell? / Short Profit / Daily Profit / Running profit.

So prev close -.5% is the level at which we sell

Prev close +.5% is level at which we buy

Next column adds a Buy if the high level was exceeded by .1% and then next to it shows the profit or loss made on that trade buying and exiting on close.

Next column shows a Sell if low level was exceeded by .1% and then next to it shows the profit or loss made on that trade from shorting and exiting on close.

Next column shows total daily profit (notice that if both buy and sell are shown the daily loss will be the max loss ie the distance between the two signal levels)

Next is the running total ie previous day's running total plus today's results.

Hope that makes sense any more questions feel free to ask again.

Stephen

Thanks

does this mean that these spreadsheet results can enter you into two trades in the same direction -
1. A) long at close/open + 0.5% (B) long again at high + 0.1%.
2. C) short at yesterdays close/open + 0.5% D) short again at low - 0.1%

If so, where is your logic in entering two trades in the same direction?

Or am i still misunderstnading you?

Thanks again.

 

[smccreedy]

The system only enters once at +/- 0.5% the previous day's close.

The 0.1% is a level I added to only enter if the target is surpassed by that amount.

ie if the entry level is 6500 and the high of the day is 6500 the system will assume you didn't buy. The high of the day would have to be 6506.5 for the system to assume you got in.

There is a small explanation as to why this is done on an earlier post where the spreadsheet is attatched.

Personally I think that this added variable means you know you would get a fill when you think you have but also means that you might get a fill when you don't think you have.

I think it is no better than without it buy many people complained that the system had bought at levels that the spread bet companies wouldn't offer them, this means every level that is supposed to be hit on the system would have been tradable.

Stephen

 

[JTrader]

The system only enters once at +/- 0.5% the previous day's close.

The 0.1% is a level I added to only enter if the target is surpassed by that amount.

ie if the entry level is 6500 and the high of the day is 6500 the system will assume you didn't buy. The high of the day would have to be 6506.5 for the system to assume you got in.

There is a small explanation as to why this is done on an earlier post where the spreadsheet is attatched.

Personally I think that this added variable means you know you would get a fill when you think you have but also means that you might get a fill when you don't think you have.

I think it is no better than without it buy many people complained that the system had bought at levels that the spread bet companies wouldn't offer them, this means every level that is supposed to be hit on the system would have been tradable.

Stephen

Thanks

i understand now.

However, this still does not fully account for the days when the open is beyond either trigger level (yesterdays open/close - high/low), when the spreadbet price may not be close to the cash index price, but the spreadsheet results assumes that you would have been able to enter via SB @ the cash price........however, with 0.5% entry levels, these days will be much fewer and further between.

As stated, theoretically, i favour placing the orders at 0810-0815, once the prices have come back in line (cash and SB), rather than entering immediately at the open.

 

[android8]

Android Trading

The system only enters once at +/- 0.5% the previous day's close.

:cheesy: Hello Mr S McCreedy

I think your wonderfull, and it's at least it's nice that eveyone has lotts
to say about your system.
I too have a theary, but I'm to tyired and I can't spell at this moment
so I'll let you know soon, bi the way, I'll soon put my money on it,
I don't think it will lose as much as I did last year.....????
Thanks a bunch... A McColl

 

[JTrader]

So far we have all assumed that the difference between the actual cash index price, and the price of cash bets offered by the SB co. will only work against us -

1. The sometimes big difference in these prices from 0800-1815, as the market opens.
2. SB quotes entering into trades when the cash market does not trade at that level.
3. Stopping us out of trades when the cash market does not trade at that level.

I agree, these are all negatives.

However, on the plus side, the difference between the cash index and SB prices can -
4. Prevent us entering a trade that would become a loser, as the cash index would enter us into the trade, but the SB price never reaches this level.
5. Prevent us exiting a trade at the stop-loss, when the cash index would exit us, but the SB price keeps us in as it never reaches the SL, and then price then moves back in our favour towards our entry price.

There is no reason to assume that scenarios 2-3, will occur more often than scenarios 4-5. On balance, perhaps we would expect the effects of these to cancal each other out, leaving scenario number 1 as the only ongoing issue/problem.

 

[smccreedy]

I agree JT I don't see why it should work against us over all.

In Jan there was a time that the yahoo data said I was stopped out but in reality I wasn't, the market reversed after that and my loss was lessened.

On another occasion though I was sold short when the market didn't actually get that low giving me a loss that might otherwise not have happened.

I think people's reservations are that as it can't be tested it takes away the security of it and makes it harder to stick to when this things crop up.

It can lead to you second guessing the system at times.

I am in the process of opening a direct market access account to trade the futures on Liffe and hope I will be able to get better data when I do. It will then be interesting to test it on proper liffe futures data.

Stephen

 

[shorty]

I had a look to see how the system worked over the recent volatile period....and it's done pretty well.
At the start of the volatility on 27/02 a short would hv been triggered not long after market start and held on for +116 points. If we stick to the plan and place orders for 28th at 4:30 on 27th, then we r short around 18:30 on 27th. This is where we r at danger to overnight volatility but in this occasion the upper long target of 6318 isn't hit so we r ok. Holding till the 4:30 close on 28th reaps a nice profit of +83. Incidently, if we tried to avoid the overnight volatility and enter market at 8am then CMC was around 6200, so no chance to enter near the 6255 short target. The wild overnight volatility before 1st of march means that both our targets are hit in overnight trade before market even opens for the maximum loss of -61. Only just a sneaky handful of pips above and below our targets.....a very nice night for CMC i'm sure ;-) Again, if we'd waited till 8am then we'd have hit long target around 9:15am and the subsequent plunge would hv taken us out at short target for max loss of -61 around 12:40, so no difference in the points total, just hitting the stops in a different direction.
So, in summary, holding the positions overnight has proven more profitable. My only concern is that the spread betters are definitely aware of breakout traders and any chance to knock out both short and long targets overnight could be a nice little earner for them. On the other hand, the volatility is very high at the moment and during a normal night I don't think they'd get away with a 1% swing in the index out of hours.

 

[shorty]

Hmm, not so good today again. CMC had another stop-hunting wild swing after 4:30 yesterday. Just narrowly within our range of 6058 to 6147, but if neither of these orders were triggered then we were definitely long just after 8am at 6147. The short trigger was 6085 and CMC just touched 6083 around 13:20, so we are taken out for the max loss of -61 again unfortunately :-(
Still, 77 points profit from those 4 days is pretty good.

 

[smccreedy]

The other alteration you could make that this system originally had was to use a fraction of the 20ma of ATR as the signal and that way in this kind of market you'd find the orders would be further away, the max loss would be wider and as such the position size would fall. Effectively it would normalise the system to this added volatility.

Stephen McCreedy

 

[JTrader]

I agree JT I don't see why it should work against us over all.

In Jan there was a time that the yahoo data said I was stopped out but in reality I wasn't, the market reversed after that and my loss was lessened.

On another occasion though I was sold short when the market didn't actually get that low giving me a loss that might otherwise not have happened.

I think people's reservations are that as it can't be tested it takes away the security of it and makes it harder to stick to when this things crop up.

It can lead to you second guessing the system at times.

I am in the process of opening a direct market access account to trade the futures on Liffe and hope I will be able to get better data when I do. It will then be interesting to test it on proper liffe futures data.

Stephen

Hi Stephen

are you still trading the original system?

How has it performed since the start?

Are you still placing your orders the night before, based on 0.25% calculations?

Many thanks .

 

[smccreedy]

Hi JT

I am not currently trading the current system but am in the process of doing some more homework on it for the futures market.

I move from spreadbet to direct access and would have to say I'd suggest to a lot of people to do the same if you are happy to trade at the minimum £10pp.

The costs in terms of spread and comm. are a lot less and the prices are purer with less skew which should help get a system off the ground.

I am trying to choose a source of futures data and once I get this sorted I want to run the same sort of calculations I ran on the index before and see how it gets on, I'll then look to start running it again if the back testing goes well.

I can know see a lot of the limitations that people pointed out like prices being skewed by the spread bet companies at the open which made getting in difficult.

When I get my hands on some futures data for the ftse i'll so some test and see if anything interesting comes up.

Really I ought to bite the bullet and buy some but I'm still trying to blag some free for now!

What about you have you been using it and if so how have you been getting on?

Stephen McCreedy