Seasonal trades with Larry Williams style exits

[FetteredChinos]

Well chaps, its been a while since the last comedy idea, but what would you say to an average of 66 pts per month on the Dow for the last 5 years, only trading once per month? with a win rate of >75%

i have been toying with a few ideas on seasonality, and playing with the exits...

the basic rules for the system are.....

ludicrously simple....

go long on the last trading day of each month..

if in profit after 4 days, then close the trade, else close after 12 days....

i havent optimised the holding periods, just thought of giving the trades enough room to breathe...

nor have i coded in stops..this would probably improve the r/r ratio, as the average loss is marginally larger than the average win (1.2x)

Caveats:-
This is using Yahoo Dow data, so the highs and lows are therefore massively incorrect, but the closes should be pretty accurate.

i backtested this on 1928-2000 data, and then fwd tested 2001-2005...

it held up nicely, with a similar win rate, despite the recent bear market...(66pts/trade 2001-2005, 100+ per trade 1996-2000)

obviously there will be cost of carry to include, but this should be a max of 1pt per day based on recent data, and since most trades close out within 4 days, this shouldnt harm overall profitability.

all it does is try to capture the month-end mark-up that tends to occur fairly reliably..

spreads arent included...

please direct comments and criticisms to the usual place...

FC

 

[JonnyT]

I would say it is not a robust solution and the end result is pure data mined.

You trade it if you want to, this mans not for trading it. (to coin a phrase from the last decent PM we had)

JonnyT

 

[FetteredChinos]

ive tried it on the FTSE and also the Nikkei just now with only a slight change in parameters..

using the 4th to last day as the entry point, this works relatively smoothly with the same parameters on all indices..

surely 800 trades on the Dow is enough even for you JT?

 

[trendie]

nice one FC. will you forward test it here ?

There is another almost once a month trade, the old Moon River method.
Will be good to see them going head to head.

good luck with it.

PS: whats "Larry Williams style exit" ?

 

[Sharky]

Interesting, thanks FC. If anyone has problems opening the zip straight from Internet Explorer, try saving it to to your computer first, then open it - this method worked for me.

 

[FetteredChinos]

Larry williams was a big fan of the "first profitable opening" exit.. ie if there is a profit on the table, then take it as soon as it appears...

oddly enough it does appear to work on some of his other systems i have backtested.

i will fwd test it here i suppose, but bear in mind its new years eve on the 30th, i may well not be here, so it will have to be papier trading seulement..

i havent looked into the moon river method properly, as i havent been able to find a decent moon phases excel download as yet.. is on the list of things to do though.

fc

 

[heasymo]

and then fwd tested 2001-2005...

Don't they call this also BACKtesting?

 

[Tuffty]

nor have i coded in stops

Mr Williams noted that most systems get better with no stops. Problem is to have any risk control a stop is needed. The system may get worse with a stop.

 

[FetteredChinos]

tufty, i believe that was in reference to stop-and-reverse systems, i can double check that from my library, but im pretty sure it is the case. plus is also backs-up what i have found from my own S&R work...

i will try to code and report back the MAE for the trades, to determine the maximum intra-trade drawdown, and therefore account size required to trade it..

i still think it could be an easy way for 40%+ annual return though, which is certainly not to be sniffed at...especially when the equity curve is essentially a straight line ;-)

fc

 

[FetteredChinos]

ok, been doing some tinkering, and have come up with something slightly more robust

reasoning that certain days have their own bearish and bullish bias, it would make sense to capitalise on the days themselves, rather than just the fact that it is X days before the end of the month..

also, thinking that if it only trades 12 times a year, in order to reduce the chance of losing years etc, an introduction of profit targets to boost the win rate would be desirable...

correspondingly, the new rules for trading the dow are as follows....

go long on the 2nd last friday of the month. ie so you are always long during the last trading week of each month. you then hold for a maximum of two trading weeks.

the target is set at 1 third of the prior week's range before entry. but this isnt set in stone, as pretty much every profit target improves things over just holding normally.

if the target isnt hit, then you close on the friday of the first full trading week of the new month.

no stops.. i know this will scare people, but you could use 10% of the current index value as a catastrophe stop (1000 points)

anyway, the results are, since 1990 are:-

Points: 10148
Trades: 190
Wins: 164
Win % : 86%
Maximum drawdown : 1224 points (worst case scenario)
Pts per trade: 53

since Jan 2001 (trying to isolate the bear market)

Points:3614
Trades: 58
Wins: 50
Win %: 86%
Pts per trade: 62

Seems to hold up well in sideways and down markets..

only one losing year in 15 (1998)

again, could yield a fairly consistent profit beating by and hope by a large margin...

obviously, excluding the bearish periods over the summer months would increase the ROI figures....but reduce the # of points...if you get my drift.

now, if you would excuse me, i have a nasty hangover to negotiate..

 

[FetteredChinos]

as an example, we would be looking to go long near the close on friday 23rd december, and then hold until friday 6th Jan, or until limit order gets hit

FC

 

[the blades]

as an example, we would be looking to go long near the close on friday 23rd december, and then hold until friday 6th Jan, or until limit order gets hit

FC

Eyup FC,

Good to see you back on this "section".
Is this tested on futures, or cash data?

Cheers,
UTB

 

[FetteredChinos]

cash data im afraid...

but since we are using limits to exit, if anything the results should be improved by the more elastic prices that exist on them as opposed to cash..

 

[MGBRoadster]

as an example, we would be looking to go long near the close on friday 23rd december, and then hold until friday 6th Jan, or until limit order gets hit

FC

Seems to be worth some further research.

For forward testing purposes I bought one contract of YM last Friday at 21:00 GMT, at a price of 10939.

I think "the prior week's range before entry" means the week ending 23rd December in this instance. (Please let me know if I've misunderstood this FC). So target is (11007-10795) / 3 = 39 points. ie 10939 + 39 = 10978

Not much reward in comparison to the stop loss, but a system's a system.

Target was hit today at 14:50