FTSE system

[smccreedy]

Hi

I've posted an incredibly easy FTSE system, the return since 01.01.96 is around 28,000 FTSE points.

The basis of the system is that it sets two levels for the day, one to buy at (above the market) and one to sell at (below the market). You enter the first to be hit with the other acting as a stop, you hold any position till the end of the day and settle it against the closing price.

An advantages of this system is that is take away a lot of decision making and takes stress from the operator.

That said watching large paper profits dissapear at the close can be hard to stomach.

This system can and has worked, order entry can be tough but there are few days I haven't been able to get entered at the price I want.

My theory is that like any system it is the ability to blindly follow it that will let it down and the trader behind it not the system.

This year alone it's run up to +155 points and back to+55 points, quite hard to take but over the long term no reason to think you won't see a return.

I've got a more advanced system that incorporates a 2% risk budget set to be lost if the position is stopped, growing the account this way is quite staggering if you can stick to it.

I'm interesting in feedback from those who think it can't or won't work and those who can think of problems with it (data is yahoo and in my experience pretty good on FTSE 100).

I'm also interested in how many people think easy systems that can work eventually will be ruined by the operator.

What would really interest me is someone with enough capitalisation to set this system up for a full twelve months to follow blindly, and the same system run by someone who can chose when to exit, ie not have to wait till close but not allowed to cancel the stop order of the day. So they can take a profit or take a smaller loss than the stop but can only enter initially at the system signal.

Any ideas which would come on top?

My feeling is the trader would take profits too quickly perhaps smooth the equity curve but overall miss some of the blow off 100 point moves that pay for all the stops along the way.

Thanks

Stephen McCreedy

 

[Hoggums]

Hi

I've posted an incredibly easy FTSE system, the return since 01.01.96 is around 28,000 FTSE points.

The basis of the system is that it sets two levels for the day, one to buy at (above the market) and one to sell at (below the market). You enter the first to be hit with the other acting as a stop, you hold any position till the end of the day and settle it against the closing price.

An advantages of this system is that is take away a lot of decision making and takes stress from the operator.

That said watching large paper profits dissapear at the close can be hard to stomach.

This system can and has worked, order entry can be tough but there are few days I haven't been able to get entered at the price I want.

My theory is that like any system it is the ability to blindly follow it that will let it down and the trader behind it not the system.

This year alone it's run up to +155 points and back to+55 points, quite hard to take but over the long term no reason to think you won't see a return.

I've got a more advanced system that incorporates a 2% risk budget set to be lost if the position is stopped, growing the account this way is quite staggering if you can stick to it.

I'm interesting in feedback from those who think it can't or won't work and those who can think of problems with it (data is yahoo and in my experience pretty good on FTSE 100).

I'm also interested in how many people think easy systems that can work eventually will be ruined by the operator.

What would really interest me is someone with enough capitalisation to set this system up for a full twelve months to follow blindly, and the same system run by someone who can chose when to exit, ie not have to wait till close but not allowed to cancel the stop order of the day. So they can take a profit or take a smaller loss than the stop but can only enter initially at the system signal.

Any ideas which would come on top?

My feeling is the trader would take profits too quickly perhaps smooth the equity curve but overall miss some of the blow off 100 point moves that pay for all the stops along the way.

Thanks

Stephen McCreedy

I did a lot of analysis with this very system, and sadly - it doesn't work - unless you keep your bet size (I assume we're talking SB'ing) constant throughout the entire exercise. Once you start to add in money management (i.e. the more money you've got the larger your bet) & buy/sell spreads you'll wipe out several times along the way. Also you have to contend with the "grey" out-of hours market the sbs operate that can't be back-tested against.

Be careful with what data you are testing with - e.g. if you are using yahoo then you cannot rely on the opening value - (it's always yesterdays close) as the SBs will have already moved their price up to where they think the market will be within a few secs/mins. Your start of day value should be something 5 mins into trading that day. It has the effect that your entry is passed before the market opens, so you get triggered at a higher value that you intended - thus increasing the distance between your buy/sell and reducing the amount of money that you expect to make.

 

[smccreedy]

Thanks this is exactly the kind of input I'm looking for.

With bet sizing I never wiped out even going as high as risking 10% of the account on each bet, were you using stops?

I've found entering the orders immediately after the close for the next day means that most day's I can get filled.

At one point I widened the buy and sell points, this hindered back testing performance but I think would be easier to copy exactly as you were able to get nearly all orders on.

When I trade CFDs with CMC you can get orders on the market as little as five points away from the market so if you were in at 7.45 you could get your orders on easily unless the US had moved a long way overnight.

Also as I stated before I don't think this system is the be all and the end all but I do feel it can provide the basis for more ideas.

One example might be. How would the account be altered if instead of using the other entry order as the stop you moved the stop to the mid price between the two orders? Perhaps you'd lose some winners but you'd halve all losers.

Another example might be, what if you move the stop to the entry price once you can, again you'd lose some winners on pull backs but losers would be cut dead quite often.

Another thought is the notion of adding at other values which would increase you winners and help pay for the losers.

All these can't be tested without good intra day data which I don't have.

My main idea would be how would this perform if you had discretional control over exit ie you can take a profit or smaller than maximum loss if you choose, this can't be back tested of course but would be interesting to see what impact the change had.

My principle idea stands that simple system can be built and with dynamic money management can generate decent returns for an acceptable amount of risk.

My belief is that it is the operation not the development that presents the largest hurdle ie following the rules is harder than writing them.

Stephen McCreedy

 

[pssonice]

telling ME how your system works isnt the key

the key is the answer on one question :

what is your mean, min. and max. initial loss divided by your account size ?
res. what might be the initial account size to run short term into profit ?

 

[smccreedy]

When I added betsizing to the test I made a bet that meant should both orders be triggered, ie my worst case loss cost 2% of the account.

By doing this from 02.01.96 to 31.12.06 the account grew from £10,000 to £980,000.

The worst draw down in this period was around 5% of the account.

This I think worked out around 50% annual return compunded for 5% risk.

 

[smccreedy]

Here are some performance stats for anyone who is interested (see stats.jpeg)

 

[pssonice]

great idea, absolutely, let s talk about this

can we find anything strange on your assumption
to treat the OPEN as a kind of key level ?

 

[smccreedy]

No in this data the open is just the previous close the open is irrelevant, the only reason I keep it is because when I download the data it's easier to keep it in than to get rid of it. It is not included in any calculations.

I tried this system on the DAX and must say it worked quite well but I guess has the same problems inherent in this system.

This system buys on the market crossing the previous close +0.25% and sells on the market falling .25% from the previous close.

I have tried many more complicated ideas like using ATR and a fraction of it to determine entry orders but it didn't work as well. Adding 0.25% to the close yields 28,000 versus 20,000 points using 0.3 x 20lag ATR added on and taken away from the previous close.

I guess only trading this system over time will show how tradable it is.

Perhaps some filters might help, ie use a 50 day lag of the market and only take buy signals when above it and sell signals when below it.

Just thoughts.

 

[theEdge]

If you're using the high/low prices from Yahoo you'll need to be careful. I found a profitable FTSE system that used the prior highs/lows for entries, based on the Yahoo cash prices.

But it didn't work because the SB prices are based off the futures prices. And the futures prices didn't have the same corresponding highs/lows.

 

[smccreedy]

Hi I take your point.

I use the 'Daily FTSE' product offered by Cantor Index that is settled against the cash market not the futures.

It does seem that most of the determining factors would be the ability to get filled at the prices you want.

How did you're previous system work?

Buy if previous high exceed by 'X' points of % or whatever, sell if previous low exceeded by 'X' points etc.

I have to say I haven't looked in to that yet but it's an idea I'll play with.

As I said before I feel simple systems can work but it seems the operators won't stick with them or let them through the bad times or make sure they follow every signal.

Stephen

 

[theEdge]

I can't remember the details as I was only doodling a few years ago in Excel.

But it was something amazingly naive such as buying or selling the prior day's highs/lows and then letting the daily cash bet settle at the closing value.

Good Luck!

The Edge.

 

[Don500]

There is a flaw in using the Cash FTSE rather than the Futures (or SB) price. On Cash FTSE (as you said earlier) the Open is always the same as last night's Close.....but that doesn't mean you can trade the futures equivalent at that price.

eg say Cash closed at 6200 last night
Cash opens at 6200 today (by definition)
Say fair-value on the future is 30 points over cash, and it opens at 6260....therefore cash will immediately jump to 6230 - but your system says you bought @ 6200 * 1.0025 = 6215.5 ie you've made 14.5 points (6230 - 6215.5) on your system that you couldn't have made for real, because you can't trade the cash as it is - you have to use the future, and that opened at 6260

I haven't explained it very well.....but does that make sense?

 

[smccreedy]

I'm not sure who you use but I can SB on the cash price and get it settled against the cash price.

I do not trade the FTSE fut.

Also even if I were to but used the FTSE cash reaching levels to enter my order although I'd be paying the premium over the cash when I opened I'd also be getting that back when I exited.

ie I want to buy at 6200 and sell at 6240, I'd buy at 6220 and sell at 6260, same difference.

Although this isn't want I'd want to do.

I appreciate everybody's input, all I can say is I have trade for real since 2nd Jan and the results are as follows:

Net 76.8 points up, not that great but it you can rely on it who cares?

Was up 141 in first week and this is what the problem is, to keep on pulling the trigger because you know the system works when you have rough weeks.

This is the premise of my thread, simple systems work, it's the traders that follow them that don't.

The best systems are hard to stomach, they might be wrong more than they are right and the payoff might not seem so good but if it works it works. Most traders want to be right 90% of the time, make a little, then lose a lot 10% of the time and tell themselves if they don't do that again the system will work. Of course they do it again and again and again.

I will keep on trading it. I think it works best if you're not around to watch as I have had several large profits wiped out over the day and that's when you get wound up and want to second guess the signals.

Yesterday is a good example.

Long at 6220, see a high of 6257.2, paper profit of around 37 points and take a loss on the day of 9.7. Hard to come in today and carry on but you have to stick with it if it works.

I've tested taking profits at % levels and absolute levels, ie if I'm 10 up take a profit but the system then fails.

You have to watch the days like yesterday to make sure you get the +100 points three times a year to pay for all the small losers.

Trading technically isn't hard, mentally it's very tough.

Stephen McCreedy

 

[pssonice]

it is strange but yet it s true.

your algorithme is profitable in FTSE.
u will lose continously in DOW an DAX

 

[smccreedy]

Oh thanks, I'm sure I tested on the Dax profitably but then I am never 100% on the data.

Dow worked alright too actually.

Still I can only speak from experience and all I can say is that at the close I know my signals for the next day.

I divide my account buy 20 to get my risk budget and then divide this by the distance between the two orders and round to get my position size.

so far this year it has worked, that may be fluke I don't know but so far I have been able to replicate my system exactly and importantly made money.

Thanks for your continued input.

Stephen McCreedy

 

[Don500]

I don't see how can guarantee trading yesterday's close on the open of today. If FTSE closes tonight at 6200 (cash), then the Dow drops 800 points tonight, war breaks out in the Middle East on Saturday, the Nikkei falls 1500 on Sunday night, are you saying that you can still sell FTSE on 6200 on Monday morning before watching it fall into the mid-5,000's ?

 

[Don500]

your system would say you had sold at 6184 and you were quids in, but in reality, if you sold on Mon morning, you would be filled at 5500 (or wherever mkt was).

Take it to extremes.....why not buy on stop 1pt above close, and sell on stop 1 pt below close, and then you MUST win unless mkt always closes at the same level !!!!!

I suspect we've misunderstood each other somewhere along the line - however, I do agree that simple systems can work (and aren't overly optimised).

 

[smccreedy]

You see the point I'm making, and perhaps I've not explained this properly is that this was designed to trade and instrument offered by cantor called the Daily FTSE, they also offer the Daily Dax, S&P and Dow.

It is priced against the cash market and becomes untradable at around 4.28pm and is cash settled against the closing price of the market, which is exactly the price quote on teletext, yahoo, Reuters etc.

This means I am never in over night, over the weekend etc.

Of course if it was the futures I was trading I think I might be able to put in an order to exit on close which would get me out at the end of the day but this happens automatically.

So this morning I went short at 6196 and got filled at that level. Tonight whatever the market closes at I will be settled at unless it's above 6226 in which case my buy order that is still on will act as stop and close me out first.

Thanks

Stephen McCreedy

 

[smccreedy]

Sorry I'm with you now, you mean how can I guarantee I get entry at that price, not close at the close price.

Well I set the orders after the market close as soon as the next day's market becomes avaiable to trade.

When I traded with CMC I could get the orders in 2 pips from the market even five or so out of hours so as most of the time the orders are around 16 above the market it isn't a problem.

With cantor it has been more so as you can't get as close but so far no problem and to be fair we've seen some large overnight moves.

As I said so far this year it hasn't been a problem and I'm sure soon it will be and when it does I'll be able to find how big a problem and hopefully find a rule to cover it.

It might be to sell at the best fill I can get and change the position size so I'm risking the same size on the account, ie wider distance to my stop so smaller size.

Sorry for the confusion I'm doing a few different things and even though I think I'm reading the replies when I re read I see I'm missing some of the point.

Thanks

Stephen McCreedy

 

[juanchi]

Interesting. I've been testing a quite similar system, and it works really well. I've only done in paper and I'm ready to use it but first I want to make more test. I back-tested several markets, Edollar, Bunds, FTSE, S&P, Euro, Jpy, Tnotes and all achieved good profits. The data is from Bloomberg and the in-sample periods are 10 years.
There are two differences with your system, first, I use the open plus and minus a certain amount, not the last day's close and second, my stop order is stop and reversal.

Two things bother me though, doing the back testing I assume (and that maybe it's too naive) that once the market hits my stop and reversal order it won't go again to the first level where I should stop and reverse again. I the spread between the two orders is quite large I don't think this double stop and reverse would happen many times. But still the system is quite profitable to overcome these volatile days. I should probably try your idea and just use stop orders to see what happens.

The second thing that bothers me it's that in a fast market I'm not sure I would get a fill at the stop level. Let's say that I have an stop order to buy the March future Tnote at open+7. if the next payroll number is -300,000, there's no a chance that I can buy it before open+25. But still, I think that this problem will only apply to markets that are ultra sensitive to economic numbers