Mr Whippy

[FetteredChinos]

hi guys, what do you think about this strat hot off the chinos production line?

Trading the Dow:-

results from Jan 1990 - Feb 2005 (yup, thats right) 15 years...

Trades 779
Points 24,362
Pts per Trade 31 (this includes trades when the Dow was sub 3,000)
Wins 686
Win rate 88.1%
Max Drawdown 1465 (hmmm)
Profit Factor 2.63
Average Win 57.3
Average Loss -160.6

Profit since Jan 04 is 1260 points (50 wins from 59 trades)


yes, i know the ave win/ave loss ratio isnt too great. but the equity curve is pretty stonking.


it is ludicrously simple.


the rules in a nutshell are...

calculate the average WEEKLY range on the Dow for the past 10 or so weeks

go long at the close every friday.

set an order to take profit at 0.2 x average weekly range.


if this is hit, then bag your profit and wait for the next week. if not, then close out on the weekly close.

current weekly range is approx 200-230 points or so, so each week we are looking to target 40-45 points.

oddly enough, even during the raging bear market of 2000-2002, it breaks even. any time apart from that it makes a killing.

smart.

very smart.


i have also, in the attached sheet, coded up a SHORT-ONLY Version, where the rules are identical, except we go short every friday.

that still makes a rather nice profit, though is currently in drawdown.

bearing in mind the long term trend of the markets is usually upwards, then Long-only is probably the way forward.

seems like a pretty good way to "scalp" 20-30 points a week. probably best not to watch the screen when trading it, as the negative weeks might be a little hairy.

Data is that taken from Livecharts.co.uk i would appreciate it if someone could double check that its data is approximately correct.


cheers, all.

FC

 

[FetteredChinos]

oddly, it only seems to perform best on the Dow. S&P isnt that great, and FTSE is only marginally profitable.

hmm..

either the data is wonky, or the Dow is whippy enough to make this work

works on monthly candles though

time to start working on my Buy-and-Hold strategies now.....

 

[barjon]

fc

You certainly dig them up must have a traders treasure trove somewhere that you tap into.

I've got a business venture for you. A range of Christmas crackers with no jokes or mottos but an FC system inside instead. With 30k+ members on T2W there must be a ready market and I'm sure sharky would play

good trading

jon

 

[FetteredChinos]

might as well stick two in each cracker then.

 

[barjon]

no - people would just have to buy plenty of boxes if they wanted the full fc repertoire

 

[FetteredChinos]

incidentally, knowing how i like to hold multiple positions etc...


there is approximately a 95% chance that the 0.2 target will be hit within two weeks, so if the target hasnt been hit in the current week, you roll it over to the next week.

that produced 26000 odd points, and an even smoother equity curve..


if you reduce the target to 0.1 then we're getting into 97% territory again....

 

[gcb01]

FC

another of your bags of cash to start with schemes?

I've tried adding a stop but it not only reduces the profit but also the max drawdown.

I think 0.25 gives the max profit and only worsens the max drawdown marginally.

The gap between achieved profit and missed profit (diff between week's high and profit made) is about £18k. Some of this could be picked up during the week with stops.

Profit per week is about 34 pips. Scaling up makes the drawdown heavy unless you have bags of cash to start with.

Look forward to the next!

 

[FetteredChinos]

dont think you need that much cash..

the maximum weekly range was approx 1000 points (correct me if im wrong)

thats not too hefty a drawdown.

and if you are trading sensible stake size..say £1 per £2-3000 capital then its a fair enough return..

better than a poke in the eye...

 

[mjarvis7]

Maybe because the DOW is only 30 stocks?

The 100 or 500 stocks in the FTSE or S & P might 'smooth out' the volatility via negatively correlated industries.

I'm in Aus so might try this on the ASX20 or ASX50 to test the theory out on the weekend.

 

[FetteredChinos]

not sure about that..

just tried it on the Eurostoxx 50 and the results were negative.. and evening changing the parameters i couldnt get it to outperform buy and hold.


either the Dow is unique in this phenomenon, or the data is wonky..

ive done spot checks on it, and it seems to match that of downloadquotes, give or take the odd point.

strange..

i'll look fwd to your findings though..

 

[FetteredChinos]

french index is a load of CAC as well


hmm..

 

[jmreeve]

Hello FC,

I have some reliable data so wil try to find time to run it through TS2000 a bit later.

What is the idea behind this strategy? Why Friday in particular?

If your figures are correct you would be looking at a drawdown of $14650 just trading a single YM.
A better exit is needed to make this tradeable.

 

[FetteredChinos]

friday is the weeks close..


i was trawling through the old Fool boards and had a look at some of JT's old strats..


his first grail strat of only trading short, and looking to scalp 0.8% of the index got me thinking.

his strat didnt work as he was testing on yahoo data which we know is a bit rowlocks.

but i thought, why not try it on weekly data. the market tends to weave around the previous close for most weeks..

thought it would be worth trying to "scalp" on a weekly basis, rather than daily.

using an ADR rather than a fixed % seemed to improve things.

i agree though, the exit strategy is a bit hardcore at the moment..

BUT, trading short only, the drawdown (but also returns) are diminished somewhat, as the down spikes tend to be longer than the up spikes...

 

[MGBRoadster]

I don't know how you'd backtest this, but if you used two accounts and traded one Long only and the other Short only then that would reduce the drawdown?????

 

[FetteredChinos]

indeed you would, and that is part of the basis of the systems i trade at the moment.

currently i bracket the market at about 70points either side of the friday close.

only problem is that it is currently flatlining, which isnt too great

could do with getting that equity curve going up again..

 

[ddunne82]

Hi.

JM, i dont think you can compare using YM as effectively you are putting on £6 per point ($10). If you then relate that to the rough estimate of £1 per 2.5k capital for safe-ish spreadbetting, well...you do the maths!!!

i think the system has a lot of merit for it, but its what i would call a bonus system - you have your regular systems that pay your salary and then you have a few longer term systems that give you your yearly bonus.

Based on FC's/JT's scalping idea, there is a book called "How markets really work" by LArry Connors that contains a large amount of statistical data from 1989-2003 for SPX and NDX. It contains some real cute little scalping ideas - i will post some of them this evening to give everyone a heads up. I cant do any kind of programming but they might give you guys who can some ideas to work off and share with us non-programmers.

Thanks

David Dunne

 

[FetteredChinos]

cheers DD, that would be useful..

 

[jmreeve]

JM, i dont think you can compare using YM as effectively you are putting on £6 per point ($10). If you then relate that to the rough estimate of £1 per 2.5k capital for safe-ish spreadbetting, well...you do the maths!!!

I generally prefer to trade futures and $10/point is a small position.
However, 1400 points draw-down is just excessive.
If the strategy does have merit then at the very least an exit needs adding to get you out before the losses mount so high.

 

[FetteredChinos]

agreed Jm.

BUT.. that 1465 drawdown came from a two 700 point down weeks in july 02..

with some sort of catastrophe stop, we could limit that somewhat.

the question is, how...

 

[ddunne82]

JM,

I can see your point on the point size and yeah 1440 does seem a tad OTT. I was thinking in terms of £1 a point so wasnt too taken aback but that amount aint gonna get too far.

Attached is a copy of 2 of the studies - for higher highs and higher lows, and for short term highs and short term lows. They have done other studies on:

1) up days in a row vx down days in a row
2) market breadth
3) volume
4) large moves
5) new 52 week highs vs new 52 week lows
6) put/call ratio
7) VIX

Kinda busy at the moment but if you want them i can do them at a later date.

I have just posted the major signifcant moves but i can also scan in each of the pages and upload them so you have the whole table of results. Let me know and i'm sure we can arrange this.

Theres definately something in this as you can see a LT statistical edge - we just gotta find a way to incoporate it into a system.

With the initial system FC, what if you put a max stop loss in - either an absolute value or as a % of previous week(s) range? I imagine you've tried various stops (400/350 etc) - a wide stop surely would take some of the ehat out of those trades but be unlikely to get hit?

Thanks

DD