just a warning that if you are optimizing a 4 or 5 parameter
strategy and it's taking 24 hrs, you are probably optimizing all
the parameters together. This is going to result in curve-fitting,
or in other words a combo of parameters which produced great
results in the past but will do dismally in the future.
when you have a lot of parameters it is best to optimize them
one at a time or at most in twos. So for example in an MA crossover
strategy with a stop loss, a profit target and a breakeven point,
you optimize the MA crossover first, then use the optimized MA
values to optimize the stoploss, then use the optimized stoploss
and MA values to optmize the profit target and so on.
not only is this much faster but the risk of curve fitting is
much lower.
I could show you some incredible curve-fitted charts which show
a perfectly straight equity curve going up diagonally from left to
right until the day after I optimized them, at which point they go
straight back down again!
The few parameters the better. The state of the art in mech
system is no parameters. I know that sounds crazy but it's
true.
It can also mean that they are not being widely held in margin
accounts.
As I understand it, IB and other brokers use shares held in
margin accounts to provide to shorts. Therefore if the margin
account holders sell their SKF or whatever, it reduces the float of
shares that IB has available to short.
So clearly SKF is being sold hard either way, by longs
liquidating their positions or shorts initiating theirs.
ND, this is because modern portfolio theory is based on an
efficient market hypothesis which states that market timing is
impossible because 'price has no memory' and price moves are
serially independent.
In fact, it is trivial to show that the efficient market
hypothesis is bunk (the Paulos Random Walk Index is a direct
statistical measure of the serial dependency of price and is a very
nice indicator in its own right) but this would invalidate a great
deal of modern portfolio theory which an entire business model
depends on, both in the schools and in the brokerages.
Am out of the game for the moment apart from some SKF puts but
futures are up bigtime this morning. This is the first rally that
has the whiff of the real thing to me. Could be a buy-the-rumour on
M2M Thursday, maybe good for a run to the low-to-mid 700s, or could
just bounce off that 700-705 area.
Let me be real clear -- I think we NEED to violate 670 to rally.
I don't think if we violate 670 we have to rally. We could be in
free fall for a while, since there is now no meaningful technical
support.
have to go out now but looks to me like we are setting up for
the move that I anticipated yesterday, which would be a breach of
680, wash down to below 670, and then a big upward spike. Or
possibly a bounce off 680, which would be much less bullish
IMO.
Yeah I meant to post about it yesterday but didn't have
time.
It made that one disastrous trade based on the late-day spike,
then got stopped out the next day @ 13%. Then on the next 60 min
bar it got back in on the right side of the trade, which is where
it is now.
It was a bad trade but I sort of forgave the strategy because it
was just absolute bad luck that the 60m bar closed right on the top
of the spike. If it had closed 15m earlier or 15m later it wouldn't
have triggered. Normally the 60m period filters out those kind of
panicky market moves, which is why it is so successful. Still, at
least the 100% record is gone, which I am sort of relieved
about.
That trade was doubly bad for me because it went twice as short
as I intended it too, due to a setting I had wrong. I hedged it
over night with ES but it still wiped out my (quite large) profits
for the week.
I dunno. There are a bunch of targets. If we break though the
top of the channel and keep going, they are all in play. I posted a
bit about this yesterday with a price volume chart.
I was thinking of buying April puts on SKF, SRS, FAZ etc since
even if we continue to fall, once we start to rise the compounding
is going to knock the stuffing out of them.
But then I looked at UYG, URE etc and thought, man these are
pretty damn cheap right now.
Funny... I have a bad habit of imagining what people I only know
from email, or the phone, look like in real life. When I meet them,
I always turn out to have been comically wrong (eg imagining dark
curly hair when the person is blond and almost bald).
BUT
RP looks almost exactly as imagined he would.
Now that is scary.
Googling for images of unSane on the internet will only get you
pictures of the excellent band but if you would like a clue of the
kind of person I am, you can watch one of my helmetcam videos on
youtube.
I hedged the SKF position, which is already underwater, as I
think there is a decent chance of a gap down tomorrow. If there's
no gap I'll probably unhedge it.
The community is delayed by three days for non registered users.
Title: Mech systems a gogo I'm
Posted by unsane on 27th of Mar 2009 at 12:44 am
I'm still not quite believing my eyes on this one. I have not yet traded this system but it looks, um, promising. This is trading 100 shares of FAZ from about the beginning of the year.
I've been quiet over the last few weeks as I've been woodshedding some systems. Will share some results and indicators tomorrow or next week. But the bottom line is that simpler is better.
Matt was right a couple of days ago when he said you have to play the statistics. There's no point in taking the odd mech discretionary trade. That will kill you. The mech systems work because they grind away making profits, and because they have no bias. In the current climate that is a huge advantage. Anyone read xTrends lately? Yikes.
Repeat after me: the losses are part of the system. The losses are necessary to generate the profitable trades. If you do not take the losses you do not take the profits.
just a warning that if
FAS Mech System
Posted by unsane on 26th of Mar 2009 at 12:04 am
just a warning that if you are optimizing a 4 or 5 parameter strategy and it's taking 24 hrs, you are probably optimizing all the parameters together. This is going to result in curve-fitting, or in other words a combo of parameters which produced great results in the past but will do dismally in the future.
when you have a lot of parameters it is best to optimize them one at a time or at most in twos. So for example in an MA crossover strategy with a stop loss, a profit target and a breakeven point, you optimize the MA crossover first, then use the optimized MA values to optimize the stoploss, then use the optimized stoploss and MA values to optmize the profit target and so on.
not only is this much faster but the risk of curve fitting is much lower.
I could show you some incredible curve-fitted charts which show a perfectly straight equity curve going up diagonally from left to right until the day after I optimized them, at which point they go straight back down again!
The few parameters the better. The state of the art in mech system is no parameters. I know that sounds crazy but it's true.
nice way to direct liquidity
Barney Frank
Posted by unsane on 11th of Mar 2009 at 11:00 am
nice way to direct liquidity out of the equity market and into the futures and options markets, no?
(does this also apply to them? I'm assuming not)
the banks have a problem
C
Posted by unsane on 10th of Mar 2009 at 11:39 am
the banks have a problem with their balance sheets not their cashflow
most of them are highly profitable
the problem is they are carrying toxic debt and over-leveraged
It can also mean that
IB update
Posted by unsane on 10th of Mar 2009 at 11:05 am
It can also mean that they are not being widely held in margin accounts.
As I understand it, IB and other brokers use shares held in margin accounts to provide to shorts. Therefore if the margin account holders sell their SKF or whatever, it reduces the float of shares that IB has available to short.
So clearly SKF is being sold hard either way, by longs liquidating their positions or shorts initiating theirs.
My solution: buy puts!
ND, this is because modern
Where and how to get started
Posted by unsane on 10th of Mar 2009 at 08:24 am
ND, this is because modern portfolio theory is based on an efficient market hypothesis which states that market timing is impossible because 'price has no memory' and price moves are serially independent.
In fact, it is trivial to show that the efficient market hypothesis is bunk (the Paulos Random Walk Index is a direct statistical measure of the serial dependency of price and is a very nice indicator in its own right) but this would invalidate a great deal of modern portfolio theory which an entire business model depends on, both in the schools and in the brokerages.
Am out of the game
Posted by unsane on 10th of Mar 2009 at 08:15 am
Am out of the game for the moment apart from some SKF puts but futures are up bigtime this morning. This is the first rally that has the whiff of the real thing to me. Could be a buy-the-rumour on M2M Thursday, maybe good for a run to the low-to-mid 700s, or could just bounce off that 700-705 area.
Let me be real clear
Bang!!
Posted by unsane on 6th of Mar 2009 at 10:54 am
Let me be real clear -- I think we NEED to violate 670 to rally. I don't think if we violate 670 we have to rally. We could be in free fall for a while, since there is now no meaningful technical support.
here's the chart, spaced out
I still have SKF in my 60m mech but it ...
Posted by unsane on 6th of Mar 2009 at 10:29 am
here's the chart, spaced out a bit so you can see what happened. the pink down spike hit the reversal trigger level (the green dots), and the system went short on the next bar. Unfortunately that bar spiked straight up and kept going! System stopped out @ 221 on the first bar next day, then went long again on the following bar.
Nerves of steel!
have to go out now
Posted by unsane on 6th of Mar 2009 at 10:20 am
have to go out now but looks to me like we are setting up for the move that I anticipated yesterday, which would be a breach of 680, wash down to below 670, and then a big upward spike. Or possibly a bounce off 680, which would be much less bullish IMO.
Yeah I meant to post
I still have SKF in my 60m mech but it ...
Posted by unsane on 6th of Mar 2009 at 10:18 am
Yeah I meant to post about it yesterday but didn't have time.
It made that one disastrous trade based on the late-day spike, then got stopped out the next day @ 13%. Then on the next 60 min bar it got back in on the right side of the trade, which is where it is now.
It was a bad trade but I sort of forgave the strategy because it was just absolute bad luck that the 60m bar closed right on the top of the spike. If it had closed 15m earlier or 15m later it wouldn't have triggered. Normally the 60m period filters out those kind of panicky market moves, which is why it is so successful. Still, at least the 100% record is gone, which I am sort of relieved about.
That trade was doubly bad for me because it went twice as short as I intended it too, due to a setting I had wrong. I hedged it over night with ES but it still wiped out my (quite large) profits for the week.
It is now long from 227.20 and looking good.
I still have SKF in
Posted by unsane on 6th of Mar 2009 at 09:52 am
I still have SKF in my 60m mech but it is now hedged with Apr 250 puts.
All the mechs are basically poised to go full tits long above 700.
my view: trade the tape
adding to my longs here, but still small positions. with ...
Posted by unsane on 6th of Mar 2009 at 09:04 am
my view: trade the tape not the news
I dunno. There are a
One regression channel to rule them all
Posted by unsane on 5th of Mar 2009 at 10:00 pm
I dunno. There are a bunch of targets. If we break though the top of the channel and keep going, they are all in play. I posted a bit about this yesterday with a price volume chart.
Now see what happens when
One regression channel to rule them all
Posted by unsane on 5th of Mar 2009 at 06:46 pm
Now see what happens when you overlay fibs.
Reversal at 670 places every single important pivot of the last three weeks on a fib line.
It's so pretty it OUGHT to come true, even if it doesn't.
One regression channel to rule them all
Posted by unsane on 5th of Mar 2009 at 06:27 pm
This is the three-sigma regression channel on the 24-hour ES 60 min chart.
All the big reversals have required a move outside the 3-sigma channel (in other words, a move more than three standard deviations from the regression line).
We didn't get one today, but a move down to about 670 would do it.
I think that is where we get the snapback, and it's going to be a doozy.
I think I'm gonna have orders ready for URE, UYG, BGU, FAS plus some canucky stuff like HEU, HFU and HOU. Stop should be tight and obvious.
/EDIT also if we rally from here instead of outside the channel, it's dollars to donuts we reverse without breaching the channel on the other side, and ratchet lower still.
I was thinking of buying
Posted by unsane on 5th of Mar 2009 at 05:49 pm
I was thinking of buying April puts on SKF, SRS, FAZ etc since even if we continue to fall, once we start to rise the compounding is going to knock the stuffing out of them.
But then I looked at UYG, URE etc and thought, man these are pretty damn cheap right now.
Thoughts?
Funny... I have a bad
just for fun - some pics of me, RP, and DrO
Posted by unsane on 5th of Mar 2009 at 12:42 am
Funny... I have a bad habit of imagining what people I only know from email, or the phone, look like in real life. When I meet them, I always turn out to have been comically wrong (eg imagining dark curly hair when the person is blond and almost bald).
BUT
RP looks almost exactly as imagined he would.
Now that is scary.
Googling for images of unSane on the internet will only get you pictures of the excellent band but if you would like a clue of the kind of person I am, you can watch one of my helmetcam videos on youtube.
eg:
http://www.youtube.com/watch?v=6sVePjOwzHA
Spool forward to 2'30" for the scary part. Trading SKF has nothing on this.
I hedged the SKF position,
SKF 60m mech system
Posted by unsane on 4th of Mar 2009 at 04:55 pm
I hedged the SKF position, which is already underwater, as I think there is a decent chance of a gap down tomorrow. If there's no gap I'll probably unhedge it.
that's what I think is
Posted by unsane on 4th of Mar 2009 at 04:41 pm
that's what I think is the most likely scenario right now - with a reversal at the trend line
here's an interesting chart of price-volume going back to the last short term top at 780.
as you can see we bounced off the 38% retracement, but the supply there is extremely thin, which is interesting. We closed just above the big finger between 705 and 710 which should represent relatively strong resistance. This is the only area in this region of the chart which shows that buyers outnumber sellers.
This chart shows you exactly why gaps want to be filled, because there's no supply there at all.
Then as you can see the next decent finger is 735-740 and above that steady supply.
Anyway the area between 715 and 735 is very thin so if this rally continues it will probably work its way up there, before meeting an impenetrable wall of sellers above 735.