Chinos' Simple S&R FTSE 100 System

[FetteredChinos]

Well guys, having done some serious research last night and modified my existing trading method to the extent that it still gives me an "edge", i thought i would share the strategy/method that started it all off many moons ago

as i have alluded to in posts elsewhere on the site, the "indicator" that i have devised doesnt take into account the close, only the highs/low of each day. the "indicator" doesnt lag at all - it is simply an expression of the latest bars action to that of the bar immediately preceeding it.

in the chart attached to this post, i have included a chart of the closing prices of the FTSE100 for 2003, coupled with the BUY points marked in GREEN and the SELL points marked in RED.

as you can see, in choppy action this system works well, but doesnt cope too well with impulsive action. I have chosen 2003 as my example year, partly because showing the points of the last few years would have made the chart unclear, and secondly that 2003 wasnt the greatest year for the system (owing to the long trend we had), when it performed relatively poorly, but still turned a respectable profit.

from my calculations, 2003 yielded the follwing stats

Total Points 1026
58 Trades
43 Winners (win rate 74%)
points per trade 17.7 (pretty poor actually)

this is a truly poor year. 2002 for example was:

Total points 1777
70 trades
46 winners (win rate 66%)
points/trade 25.3 (more in line with historical averages)


whether this system is now past its best is debatable, but hopefully it can give the rest of you guys inspiration for your own systems. it certainly got me thinking.

this one suffers from fairly large drawdowns, but as long as the stake size is kept relatively small (for spreadbetting i use £1 stake for every £1,500 in my trading fund - quite conservative) then there shouldnt be too much drawdown.

the rules are::

1) Calculate the Average of the High and Low for today's trading day

2) if today's average is 20 points HIGHER than yesterday's average, then SELL at the CLOSE

3) if todays average is 20 points LOWER than yesterday's average, then BUY at the CLOSE.

4) thats it - its a simple Stop and Reverse strategy. hold a position until you get an opposite signal.

i have tinkered with it somewhat to minimise the drawdowns etc, but at the moment, these methods shall remain somewhat secret lol.


comments as usual are most welcome.

it works to some extent on the DOW, but the points/trade level is almost the same, and we all know how the DOW whips around like a beastie, hence why i prefer to trade it on the FTSE (plus the more sociable hours!!)


FC

 

[anmolway]

Thanks FC for sharing your method. Will look into it in more detail in terms of drawbacks.

 

[FetteredChinos]

no problems mate.

drawdowns are always a problem with stop and reverse systems.

when i get a moment free i will attach my backup spreadsheet which shows the system "in action".


by doing all sorts of crazy auto-filter jiggery pokery you can learn to filter out the bad trades!

(edit - i mean you can filter out where the bad trades are "more likely" to occur lol)


FC

 

[rog1111]

FC

Out of interest, may I ask whether your results are based on the FTSE index or FTSE futures prices ? Big difference. I recently came up with a nasdaq100 system that worked brilliantly for the index values, but it was hopeless for the NQs or QQQ.

rog1111

 

[FetteredChinos]

based on the Cash index im afraid, so the usual caveats apply.

the system is based on EOD prices, at which point the fair value of the futures in relation to the cash would have settled down. besides, in some cases the bias will work in your favour, in others it will work against you.


having said that, in my experience of trading it last year, as the nature of the entry points is contrarian, you often find that you have the SB bias on your side. which makes a pleasant change

FC

 

[DUM TRADER]

With this type of system it does not matter if you trade the cash, because you are buying or selling at the close of trading and most spread betters allow for this. Anyhow the results would be the same as with the futures as all the trades are taken at the close.

I used to know a guy back in the states where he used something similar to fetteredchinos system. but he uses a moving avr and if price is higher or lower by. 050% of the moving Avr then he will trade the opposite direction. but he wouldn't take the trade at the close. he would wait for the open and use some kind of volatile breakout method to to back it up.
I will think up of some filters to see if we can make this work even better.

Byt the way what was you bigest draw down on the dow & ftse


DM

 

[FetteredChinos]

DM/ cj12 :cheesy:


thankfully i never experienced a mega drawdown, but according to backtesting, it amounted to 529 points (kaboom!)

however, this isnt as bad as it sounds. it only equates to about 33% equity if sensible position sizing is adopted (my £1 per £1500 rule)

i attach the backtest/forward-test spreadsheet for you guys to have a look through and play with the various triggers etc.

of course this can be improved etc, but this was what got me thinking about improvements etc, and hopefully it can trigger some ideas for the rest of you to come up with. i have since developed a new method (Chinos System 9 is its working title, and my god is the equity curve smooth!!! I may share this in good time )


FC

 

[rog1111]

With this type of system it does not matter if you trade the cash, because you are buying or selling at the close of trading and most spread betters allow for this. Anyhow the results would be the same as with the futures as all the trades are taken at the close.

DM

With respect, being the cautious trader that I am, and not wishing to detract from FC's results, before using the system, I myself would still check that it was valid for futures prices. You could still use the index values for signals, but you would have to use the futures prices for testing, making sure of course that if EOD data was used, then the time frames matched exactly (not always the case). Also there are plenty of other threads here that stress the importance of verifying the accuracy of the index data, particularly from free sources.

GL
rog1111

 

[FetteredChinos]

agreed rog,

but the fact of the matter is that i have actually traded this methodology with success in the past, and in fact the only reason i have changed it is that i have found better, more consistent methods, that are derived from it.

again, this method that i have shown isnt the most robust of systems, so i have displayed it only for "educational" purposes.

FC

 

[pogle]

thankfully i never experienced a mega drawdown, but according to backtesting, it amounted to 529 points (kaboom!)

FC,
I don't think this is correct, at least according to my idea of what drawdown is (which may not be the same as everyone else's!).
Your code seems to accumulate losses until the next win at which point you reset the drawdown to 0. I calculate drawdown as the difference between the current trading balance and its highest ever value. I've modified your spreadsheet and get a maximum drawdown of 673 points. Still not disastrous but it's a bigger KABOOM!

As I say I don't know if my version of drawdown is the 'correct' version but it's always worse that your version which is no bad thing in my mind.

Thanks for posting this (and your other ideas).

pogle

 

[wolfie]

FC

If I have followed your s/sheet correctly you would have made 117 points so far this year, would you mind confirming this?

You mentioned you've tinkered with the 'rules' to improve the returns, have you introduced a stop loss rather than wait for a signal in the opposite direction?

Thanks for the post.

Wolfie

 

[FetteredChinos]

wolfie, if you read my journal, you will see my last few entries and exits. on cloed positions the past month has yielded 129 points. (3 wins from 3)

i had a bit of a disaster in january with "The Dow Experiments" (see other threads on the tradings strats board) , so im only back level with my starting capital as at 1st Jan. curses. still, im clawing it back, and it finally looks as though the DOW is jumping off its cliff

as regards a stop loss, i tried all sorts of levels - fixed, absolutle levels, say 100 points etc, or even a trailing stop loss. they made little difference to the overall results, in fact, more often hindered that helped. which kinda makes sense - the trade only stops and reverses when the market has its "blow-off" (ahem) in your favour.

pogle - good point re:drawdown - i just couldnt think of a way to code it in excel - my brain has been very tired lately.


as regards rules to improve the system, instead of improving the exits, i sought to improve the entries.

a good place for the rest of you to look is to note the relationship between a current day's high/low points and that of its predecessors, and then note the course the market then followed.

still, if this method shows us one thing, it is that a high win% system is far easier to stick with through thick and thin, and also has a smoother equity curve, in my opinion.

as always, i shall be doing some more research over the weekend. amazing how i keep getting inspiration when hungover like a donkey

cheers

FC

 

[TheBramble]

I don't think this is correct, at least according to my idea of what drawdown is (which may not be the same as everyone else's!).
Your code seems to accumulate losses until the next win at which point you reset the drawdown to 0. I calculate drawdown as the difference between the current trading balance and its highest ever value.

investopedia agree with you pogle:-

http://www.investopedia.com/terms/d/drawdown.asp

"The peak to trough decline during a specific record period of an investment or fund. It is usually quoted as the percentage between the peak to the trough. "

 

[KIMMRUNNER]

I find all these automatic systems interesting.

But the most difficult question of all is how to define what is a 'better' system, and the answer is not easy

I tried to open a recent thread to debate this point: but didnt get much of a response.

Which is better?

A system which trades only once every 30 days but with 99% success, or one which trades every 2-3 days with 70% success?
Waiting for trades is frustrating, but there are patterns which are infrequent but more or less infallible.

Is a system which trading single point achieves (say) 20000 points profit allowing for spreads over the last 4 years, but at the cost of a dip of 1000 points, better or worse than a system which achieves only 10000 points but does so only dipping 100 points?

I have also pointed out that to avoid the accusation of over optimisation, there must be a set of dates when external events take over - eg 9/11 to which response must be random, since no system can predict 9/11!!!

It would be nice to produce a set of rules which assign a performance index to a system then have a competition for the best system!

 

[FetteredChinos]

eh up Kim, i didnt think you were still around!!

i have been trying to reverse engineer your system for which you posted the results up a few days ago. i thought i had it, but when i extrapolated it into the other data, it didnt work. curses.

could you perhaps share a bit of it? it seems to be a simple close at EOD system when the entry criteria have been set. but what im not sure about it that it seems to buy/sell at the extremes of the day. is it something related to a gap-trading strategy?


in relation to your questions here, i would prefer to trade a system that trades more often, even if it is necessary to sacrifice a bit of a win rate, and perhaps overall profit. maybe it is just in my nature, but i would prefer to have locked profits in, rather than have them on paper.

but it all depends on how long you want to spend in front of a PC. i dont like sitting here for any longer than i have to, so perhaps one system that trades every few weeks, but allows your to define the limit orders beforehand etc, is preferable , if you would rather be on the golf course.

on the other hand, why not trade both systems together, so you load up when you get a simultaneous signal from both?

FC

 

[KIMMRUNNER]

As regards the system for which I put forward data the answer is complicated: although there is nothing magic about it and there are many similar trading schemes: some of which work well for example on hedged indexes

I notice that people seem to have a scepticism about maths, but that I am afraid is what it is.

I started on the basis that most of what is written in tech trading books is complete and total tosh. The same tired old ideas that dont work in practise, as any amount of back testing of stochastics, rsi etc etc seem to prove.

So I was looking for methods which work consistently over 5 years, -ie work in bull, bear and uncertain markets.

So I had to go outside the books to find it.

I am probably about to confuse you but you did ask!

Moving averages are easy enough to understand

Weighted moving averages are more complex: but skew the average towards values with high weight - for example you could use weight of 1 for rising values of closing price, weight of zero for falling - which is then taking an average only of ascending closing prices.

The weight system used in that example is complex but related to the position of the closing price relative to the centre of the high low bars.

Weighted averages are straightforward to calculate on a spreadsheet

Stochastic is an example of the difference between two moving averages.

So finally to the punchline:
The example spreadsheet is essentially the difference vetween two weighted moving averages, with similar time constants.

***OUCH!!!***
It is not as bad as it sounds

The nice thing about the example shown is it is the difference frequency of the moving averages results in reversal every few days: so spread charges are small.

Finally the data is compared with another trading system ( also obscure!) to protect agains loss - which is why the system rarely loses much, preferring not to trade if the schemes dont agree

Obscure : perhaps.

Works well: dont know!!!

Thats what I am hoping people can tell me! by defining a performance index:

Certainly if you download the historic data from systems such as Goldline, this performance is way way superior by any definition & I wonder how such people have the nerve to charge!! : but whether it is the best that can be done with the FTSE? that is what I would like to know!

One metric is to take the absolute change every day: so that you can add up what a single point return would be for the FTSE if you had tommorows newspaper and called it right every day : expressing the system as a percentage of that.

On the FTSE I have achieved something like 25% of the maximum possible and on SP 500, I have achieved better than that.
Sounds poor, but you need 65% successful trades to achieve that. Which for prediction is not bad.

But the philosophical question is nagging me as to what is better?

The system that never loses has no capital employed therefore infinite gain. SO on that basis trading 95% certainty every 40 days must be better than 65-70% every other day - the capital cost is less.

Food for thought?

 

[FetteredChinos]

Kimm,

very impressive stuff. thankfully it doesnt seem as complicated as i first feared. I recall the good old days of Further Maths A-level 9 years ago, and by jingo THAT was confusing..


btw, i dont think you attached your spreadsheet to your post, or did you "example" sheet refer to the one on the other thread?


i havent looked at weighting averages in the way you suggest, as i didnt think there was much merit in them. your methods seem to have rekindled my curiosity though..you seem to use the divergance a bit like MACD n'est-ce pas?

seems a bit similar to the way RSI is calculated. ie counting moves up seperately from moves down?

as regards which system to trade.. even if u had a 95% system and the trades were yonks apart. how could you psychologically trade it if you were sitting on a loss caused by that missing 5% for months on end? not easily i can tell you.. but still, i agree about the long term picture.


FC

 

[DiscoPete]

Confused!!

FC

Just had a look at your journal (going okay at the mo!) and today you had a long signal. This means using your system that the average of the high and low of todays trade must be 20 points lower than fridays. However i make that the other way around surely (and not even greater than 20pts)?! Or is it the other method that you are working on on not your original system?

Todays average = 4552.75
Fridays average = 4541

Difference = 11.75

So how did you arrive at that descision?

Cheers

Pete

 

[KIMMRUNNER]

The algorithm may sound superficially like other schemes , but using the weightings: as I do, the behaviour of the schemes have nothing in common at all with this other tech trading stuff: indeed the underlying principle by which I arrived at this is chalk and cheese.

Have a playwith complex weightings and you will see it is an entirely different world that doesnt not play by any of the same rules. For one thing it seems to work, which try as I have and try as I may I cannot get MACD RSI or any of those other tired old schemes to work.

May be they work on a few months of back testing. But they all seem to reverse under long period hard testing.

In a way the scheme against which I am second checking is more interesting still: which is analogous to , but mathematically unrelated to candlestick pattern testing. It is out of that that many of these infrequent but 100% success schemes seem to occur.

The way to get rich would seem to be to back these infrequent positions heavily: but I am impatient. WHich has cost me dear on occasions.

 

[FetteredChinos]

pete,

apologies, but in my journal i am only trading a subtly different version of the method i have exposed on this site.. until i can come up with something different in future!!

as i speak, its not looking too grand, what with the Dow starting to correct, and its going to take the FTSE with it. still, a system is a system!


Kimm,

thanks for explaining further. im going to have a bit more of a play with weighed averages and see what i can come up with..

btw, what sort of periods are u looking at? short term eg 5 periods, or much longer timeframes? judging by the frequency of the trades, im suspecting something short maybe a 3 or 5 combined with a much longer one, perhaps 25-40 or so.. just my early hunches..

i'll see if i can come up with a similar method, and hopefully exceed your results lol.


btw, what system are u comparing your system too? is it not a bit cumbersome trying to follow two (or more) methods simultaneously?


FC