Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
I have been studying the open source trading platform TA4J, Java, because it is a fully extensible collection of trading rules, signal generators, and backtesting tools. I believe it's possible to build entry and exit signal generators for the Lichello algorithm, and write trading rules to use those and/or in combination with other signal generators, such as the long EMA signal I added years ago.
Hi Tom,
I guess I'm looking for documentation on how it works.
I confess, I'm pretty rusty using this forum.
Thanks
Jack
Been moons since I was here. Lots of them. In the interim, I did 5 years at SRI International in the AI labs, and found myself working with a new skill; I've been thinking about where AI fits in A.I.M. Way back when, I used a bit of "AI" in the form of the "genetic algorithm" to discover some improvements to the raw Lichello algorithm. Maybe it's possible to push further into the space of tuning the Lichello "springs". Over.
Don't have your email. Seems to have changed. Been a while. Mine is what I use here at gmail. I've been looking at some new ideas around AIM but want to has them out in the background before going public.
Cheers, Jack
Hi Tom,
Oddly enough, iHub doesn't remember me. Can't send private messages. Would like to talk about ETFs and more using my name and gmail if possible.
Cheers
Jack
Synchrovest. I'll look at that. Thanks.
BTW: I have been playing with something that is the equivalent of a stochastic based on the highest high and lowest low -- as it evolves -- ever since I began playing with trading commodities. A trading rule I discovered there, an obvious one at that, is "don't go long if you are bumping off the highest high". Of course, the long term bubble we just ended suggests that a modification to that rule, which takes into account the slope of a long term trend line (slope of a long moving average, say), is in order. There is probably some correlation between that stochastic and the Williams %R.
While we are at it, if you happen to read the book (blocking the title, it will come eventually) about the guys in Santa Fe who are doing well using chaos theory to trade financials, you will learn that they are playing with what turns out to be one of the first AI techniques ever; it was called Pandamonium by Oliver Selfredge, and it was basically a bunch of "shouting agents" with a learning algorithm that teaches who to listen to. _Predictors_ by Thomas Bass -- damn, I'm good, sometimes. Incidentally, read _The Eudamonic Pie_ also by Thomas Bass which is about the same guys when they used chaos to bet roulette by putting a transmitter in one shoe that transmits signals to someone elses shoe who is betting the table. Um. The other shoe has a receiver that pokes a pin into the trader's toe. Read how that pin went crazy once and...
Back to shouting agents, Tom's BigDummy Wave is one such agent. There are many. It seems reasonable to think in terms of lots of agents -- the more the merrier -- and some way to mix and match their signals. AIM BTB is also one such agent. Vealies make it clear that you don't always listen to certain signals.
Cheers
Jack
DeepDiver, indeed!
I'm not sure here, but in a definitely trending down market, the hedge seems reasonable; sometimes you might ride the short position even when things go against you given the odds are the short will return to the money. I'm thinking it's not like a margin call or somesuch. Or is it? Thanks. Jack
Thinking about deep divers.
I've been pondering graphs and this thought occurred to me. I betcha it's not original -- wouldn't be surprised at all if The Tom or someone else already thunk of it. But, here goes.
Think in terms of a down-trending market as evidenced in a long term moving average of the ticker being traded. Imagine using the trading rule I once spoke about before: buy only when below that moving average and sell only after rising above that moving average. In an uptrending market, where most of my optimization studies occured, that's way cool, I think. But, it might be a loser (my hunch -- haven't returned to testing yet) in a down market. Particularly where the ticker takes a deep dive, perhaps never (in a useful sense) to return to your first buy.
So, I've got this idea. I call it the Covered Short Position. (catchy name, what?). When you go long a ticker (below the moving average), in a separate account go short the same amount. If the ticker dives deeply, at least you're covered. No time limits as with hedging with options.
I confess, I haven't put all that much think power into this; I'm mostly shooting off my mouth over something that, on the surface, looks really useful. Comments welcome. Cheers, Jack
FWIW: Congress is working on a bill that lets naturalized citizens be president. There could, indeed, be an Ahnold fer president some day. Cheers, Jack
There's really a whole set of issues regarding things like modern portfolio theory *MPT* and optimum trade size and so forth lurking behind AIM which I have yet to get my head around. There is this issue of jumping in and out of various stocks when you create a portfolio. It seems to me that AIM doesn't work like that, that you select a stock, then ride it because you are making bets that take a long time to resolve, as compared to, say, day trading where your bets resolve in hours. Kelly's optimum betting strategy does work for things like commodities, but only if you are taking positions larger than one lot. Maybe also it works for stocks where you are a large lot trader. I'm going to be very interested to see how MPT plays out. My first guess is that you use MPT for creation of a portfolio that you intend to stick with for a long time. MPT may then be useful to see if conditions change such that your initial portfolio is nolonger appropriate.
Thanks, Tom.
Jack
Yes!
I have AIM 2000. If there is a tape of AIM 2001, then I'd like to see it.
Thanks. Jack
That's almost as much as I know about Java!
Cheers
Jack
Hard to imagine that Cheeseland was harmed by the massive economic downturn that befell hightech land. Best I can say at the moment is that it's nice to return to this group of great people. Is there a film available from the last AIMCon?
I'm not dead yet.
Jack
Hi Conrad,
Thanks for replying again. Gads it's been moons. I guess I wonder just what the market for software for AIM investing these days. If I recall, you said your program is written in c++, which, in a former life, I used to do. These days, I do pretty much everything in Java to guarantee cross platform. Cheers, Jack
Many thanks, Mark!
I suspected there was something like that going on.
The evolutionary programming I do is, indeed, a variant of the GA you have added to your product.
Slick!
That's all I can say. Slick!
Jack
Hi Mark!
Yes, I remember pretty well, for the old fart I am becomming. I cannot say I have a latest and greatest wrt AIM but I am working on a latest and greatest wrt (with respect to) online collaboration systems. I think that there is much room for improvement in the way we hold these online discussions; I've mentioned it before back in the days of SI, and I'm building such tools now.
While surfing a few of the recent posts, I noticed that even Tom is using the "O" word (optimize). Gads! What a little time will do to people. Now, I need to learn what he's talking about (mean-variance optimization, or somesuch).
Overall, I'm looking for some idea of what the bear market has done to AIMers. When I was last here, the bull market had just ended.
Cheers, Jack
Conrad!
If you're still around here, sorry for the nearly one year delay in responding. Talk about lagging the market!
My optimization technique is one from the class of software patterns called evolutionary programming. That's where you mimic what we think we know about how biological evolution works, by tricks like splicing genes and performing random mutations at various points along the gene. In my case, the gene is a list of "coefficients" used as multipliers in small routines which do things to AIM parameters. The idea is to "breed" a population of AIM parameters and run them in simulation, taking the winners and "breeding" them (combining their genes).
It has been several years since I did that work and all I really remember from watching it run was the giggles as the overall performance of the winning individuals in the population actually began to rise.
More later if there remains interest. My thinking is to make the Java project an open source one, meaning others can join in on coding, testing, owning, whatever.
Cheers
Jack
After a whole bunch of moons, I thought I'd jump back in. I was asked about my Java AIM engine. It's been well into a back burner mode for a long time -- mostly as long as the dotBomb unemployment lasted. Now that I'm employed again, I'll likely pop it to a mid-level burner (mid-level because there are a couple of other major projects running in the foreground besides my job).
Time, it seems, to relearn all the names of my friends here.
Cheers
Jack
Thinking about open source for my Java AIM engine.
It's an unfinished transliteration of my c++ engine used in the experiments I ran and mentioned at http://www.thinkalong.com.
Any interest here?
Jack
I'm Baaaack...
Hi folks. Don Carlson sent me a private email that got me here. I found out it's free, so, what the heck!
Cheers
Jack