-
Please read our AutoTracker policy on the IFT deadline and remaining active. Thanks!
-
$ - Premium Service Content (Info) | AutoTracker Monthly Winners | Is Gmail et al, Blocking Our emails?
Find us on: Facebook & X | Posting Copyrighted Material
-
Join the TSP Talk AutoTracker: How to Get Started | Login | Main AutoTracker Page
-
The Forum works well on MOBILE devices without an app: Just go to: https://forum.tsptalk.com ...
Or you can now use TapaTalk again!
fabijo's account talk
- Thread starter fabijo
- Start date
fabijo
Well-known member
Well, that was a quick one day move. Going back to S tomorrow. My fingers are getting tired of all these IFT's. I really need to make this automated. Now they got the new logon process, with all the Agree and Okay buttons. I wonder when they will require a new password.
fabijo
Well-known member
It might just be easier for me to split between funds or just create a weekly monkey.
fabijo
Well-known member
You and the Monkey are just trying to beat my record of 166 IFTs!!
Probably! I think this thing will try to make an IFT every other day.
The_Technician
Active member
Sounds like a lot of monkey business to me.....:nuts:Probably! I think this thing will try to make an IFT every other day.
fabijo
Well-known member
Well, yesterday I had a relaxing day at my house, so I picked up a pen and paper. I'll be working on another program again (I don't think I'll ever stop). Now that I've been taking Visual Basic .NET and Oject-Oriented Programming with C++ for my college studies, I've been looking at programming a little differently. That object-oriented approach is definitely handy.
I mentioned some of the basic logic for my next program in ebbnflow's account talk thread. I've been writing the logic out and making little diagrams. My wife just looks at the paper and thinks I'm a mad scientist or something.
Basically, I'll create different investors (or monkeys). Each monkey has a set of rules that they follow to make their trades. Some are bullish, some are bearish, and some are neutral. From those, some are day traders, some are weekly traders, some are monthly traders, and so on. What I will do is have all of these monkeys look at the data and predict what tomorrow's price should be according to their bias. They will then make a decision to either buy, sell, or hold based on their rules.
So far, it just sounds like a little game. Let's say I have 15 different types of monkeys making price predictions for tomorrow. If I averaged their price predictions, I would have a general market prediction for tomorrow. The problem is that all of these investment styles to not have equal weight, so I need a way to give more weight to monkeys who have a tendency to succeed more than other monkeys.
That's where the pompous monkey comes in. He just keeps track of all the monkeys. He knows how successful each monkey has been over time and will decide how much weight each monkey's prediction has on the overall prediction.
What I plan on doing is having these monkeys participate in the historical market, so that I can play with how much weight each monkey has in the beginning. This is called training the pompous monkey. Once it is trained, I will let all the monkeys run loose, predicting tomorrow's price in their own way, with the pompous monkey averaging each of their predictions to make a prediction of its own. I will then feed that predicted price into the monkey market to have a prediction for the next day. No more days after that - just two days predictions. As chaos theory shows, even though your predictions look good in the beginning, the tiniest fractions in the beginning go wildly out of proportion far into the future.
I only really need to "predict" two days, because of our delay in the TSP. Even if I can't get a real price prediction out of this, I suspect that I will get a general idea of the market movement, which could be enough to know if they will be up days or down days. That's all that really matters, right?
Gotta go, Wall Street is knocking on my door. :worried:
I mentioned some of the basic logic for my next program in ebbnflow's account talk thread. I've been writing the logic out and making little diagrams. My wife just looks at the paper and thinks I'm a mad scientist or something.
Basically, I'll create different investors (or monkeys). Each monkey has a set of rules that they follow to make their trades. Some are bullish, some are bearish, and some are neutral. From those, some are day traders, some are weekly traders, some are monthly traders, and so on. What I will do is have all of these monkeys look at the data and predict what tomorrow's price should be according to their bias. They will then make a decision to either buy, sell, or hold based on their rules.
So far, it just sounds like a little game. Let's say I have 15 different types of monkeys making price predictions for tomorrow. If I averaged their price predictions, I would have a general market prediction for tomorrow. The problem is that all of these investment styles to not have equal weight, so I need a way to give more weight to monkeys who have a tendency to succeed more than other monkeys.
That's where the pompous monkey comes in. He just keeps track of all the monkeys. He knows how successful each monkey has been over time and will decide how much weight each monkey's prediction has on the overall prediction.
What I plan on doing is having these monkeys participate in the historical market, so that I can play with how much weight each monkey has in the beginning. This is called training the pompous monkey. Once it is trained, I will let all the monkeys run loose, predicting tomorrow's price in their own way, with the pompous monkey averaging each of their predictions to make a prediction of its own. I will then feed that predicted price into the monkey market to have a prediction for the next day. No more days after that - just two days predictions. As chaos theory shows, even though your predictions look good in the beginning, the tiniest fractions in the beginning go wildly out of proportion far into the future.
I only really need to "predict" two days, because of our delay in the TSP. Even if I can't get a real price prediction out of this, I suspect that I will get a general idea of the market movement, which could be enough to know if they will be up days or down days. That's all that really matters, right?
Gotta go, Wall Street is knocking on my door. :worried:
ayla
Member
I've been mulling over some comments I've heard (can't remember if it was here or somewhere else) about investment styles working better according to the type of market as well as other things. I would suggest that your pompous monkey first decide what the nature of the market is, bull or bear (or trending or in a trading range, etc.) and THEN decide what the success rate is for each monkey. I don't think this is the same as giving more weight to a "bullish" monkey in a bull market and more weight to a "bearish" monkey in a bear market. I could be wrong (won't be the first time).
I've been thinking more and more about this, that when I do any back testing, I shouldn't apply the same logic (or program) all the way back. That I should have a different program depending on the type of market. I have a lot to learn about the different types of markets so I can't explain it any better than this.
I appreciate you sharing your thoughts about your programing. Thanks.
p.s. And one last thought, I have entertained the idea of learning about "neural networks" to use in developing a program. If you're just starting a new program, and you haven't thought of this you might consider doing a google search for "investing" and "neural networks" and I think you might find some interesting stuff. (Maybe you could work the "neural networks" into a school project for some kind of credit...)
tspgo_com
Active member
Calculated Returns.
Fabijo:
I have been looking at your spreadsheet because TSPGO! sell and buy signals are generated by short term Simple Moving Averages and the difference between them, etc. Reading some of your posts I found some similarities and I got the impression that you were using a more refined method to get your Sell and Buy signals. I am not an expert using Excel formulas, therefore I use simple math formulas to calculate returns. For example to calculate the total return in your spreadsheet as of December 15, I use something like "=T896+S897" where T896 is December 14 accumalative total since June 2, 2003 and S897 is December 15's gain. When I use this formula to your spreadsheet the return as of December 15, 2006 is 96.53% instead of your carculated return of 157.64%. What am I doing wrong?
Thank you
Fabijo:
I have been looking at your spreadsheet because TSPGO! sell and buy signals are generated by short term Simple Moving Averages and the difference between them, etc. Reading some of your posts I found some similarities and I got the impression that you were using a more refined method to get your Sell and Buy signals. I am not an expert using Excel formulas, therefore I use simple math formulas to calculate returns. For example to calculate the total return in your spreadsheet as of December 15, I use something like "=T896+S897" where T896 is December 14 accumalative total since June 2, 2003 and S897 is December 15's gain. When I use this formula to your spreadsheet the return as of December 15, 2006 is 96.53% instead of your carculated return of 157.64%. What am I doing wrong?
Thank you
fabijo
Well-known member
Re: Calculated Returns.
Hey, thanks for checking out the spreadsheet. You're right, I just use moving averages. That spreadsheet uses the MACD, where you can change the parameters of the MACD. I see why you are getting 96.53%. When you add percentages like that, it does not give you a true return on your original starting value from June 2, 2003. Your method might work for calculating small changes in a small time frame, but those small differences work out big in the long run. The easy answer is to say it's because of compounding. Here's a quick example.
Let's say that today you had $1,000 in the TSP. Now if you had a .5% gain for tomorrow, your amount in the TSP is $1,005. If you then had a .5% loss for the next day, your TSP amount would then be $999.975, because you are losing .5% of $1,005, which is $5.025. Just adding those two percentages together would make you believe you had a 0% return (no gain, no loss), when really you had a .0025% loss. So, if you continue adding like that for 908 days, those little differences would add up to make a big error in calculating returns.
Does that help?
Hey, thanks for checking out the spreadsheet. You're right, I just use moving averages. That spreadsheet uses the MACD, where you can change the parameters of the MACD. I see why you are getting 96.53%. When you add percentages like that, it does not give you a true return on your original starting value from June 2, 2003. Your method might work for calculating small changes in a small time frame, but those small differences work out big in the long run. The easy answer is to say it's because of compounding. Here's a quick example.
Let's say that today you had $1,000 in the TSP. Now if you had a .5% gain for tomorrow, your amount in the TSP is $1,005. If you then had a .5% loss for the next day, your TSP amount would then be $999.975, because you are losing .5% of $1,005, which is $5.025. Just adding those two percentages together would make you believe you had a 0% return (no gain, no loss), when really you had a .0025% loss. So, if you continue adding like that for 908 days, those little differences would add up to make a big error in calculating returns.
Does that help?
fabijo
Well-known member
I've also found that different methods work in different types of markets. That's how my newest monkey is working. It's decision for bullish or bearish markets is decided by seeing if the 75 day EMA is more than the 180 day EMA. In a bull market (75EMA > 180EMA), it just sticks to the C,S,I. In a bear market, instead of hanging out in G or F, it will go to the C,S,I if 3EMA > 19EMA. That way, it doesn't just wait for the 75EMA to catch up to the 180EMA.
As far as the pompous monkey, I still have a lot of planning and programming to do to create all the different types of monkeys. I randomly think of a different set of rules for another monkey. The more I think of, the more I realize that the market is made up of all kinds of investors at any given time. Consistent through all markets are the buy and holders. I just might have to do research to see what general rules/methods a bunch of us monkeys go by. Some people like to pump more money into the market while it is going down, then slowly peel off a little at a time while in a bull run. There are so many monkeys to create.
Once I have all these monkeys, I'll be playing mostly with the pompous monkey's rules for giving weights. I might make it accumulative, or I might use your suggestion by having it keep track of the success rates during different types of markets.
If anybody wants to share a simple method that a monkey should follow, feel free to let me know. Some examples of monkeys:
Buy when Slow Stochastics crosses to the upside, sell on the downside.
Day trader monkey: Buy on a down day, sell on an up day.
and the possibility of monkeys are endless.
fabijo
Well-known member
Of course, the monkey I created in December is beating the monkey I'm following. Who to go with? I'm sticking with the masochist monkey and holding my 100% S fund position.
For those interested, here are where the monkeys are going tomorrow:
Masochist Monkey: S Fund
Scared Monkey: F Fund
For those interested, here are where the monkeys are going tomorrow:
Masochist Monkey: S Fund
Scared Monkey: F Fund
fabijo
Well-known member
The monkey market is planned on being a kind of neural network. I got the idea by reading this article: http://www.cprogramming.com/tutorial/AI/perceptron.html
It is a general description of a perceptron (meant to emulate one neuron). I was thinking of the pompous monkey as the one neuron. All of the market monkeys will serve as its input. I could create a bunch of pompous monkeys to act as a neural network, but I'm planning on starting with one to see how it does.
ayla
Member
Thanks for the link! That seems to provide a good foundation to start from. Definitely a keeper!
fabijo
Well-known member
No problem, ayla. Sharing is definitely the point of this board.
One thing I forgot to mention about this monkey market. It is mostly meant to test a little theory I have. Those sudden market moves to the down (or up) seem to catch everybody by surprise. My little theory is that there may be some moments in the market where all these different strategies happen to merge together causing a strong force in one direction. It seems like the entire market is acting in unison to that direction and it seems like some coordinated effort, but I think that there are just times where a bunch of investing styles accidentally meet in the same direction. I'm hoping that my monkey market will stumble upon a similar scenario. Even if I give it some random starting data, I'd like to see if the monkey market would produce a market that looks real with its sudden drops out of the blue.
Perhaps some of convergence has to do with institutions and hedge funds buying and/or selling at preconceived or predetermined inflection points, when (as an example) indices reach support or resistance at the 50 or 200 day moving averages, and also when targets regarding overextended or undervalued stock prices are reached.
fabijo
Well-known member
Yup. I think the monkey market big guys will emulate institutional buying and selling.
Similar threads
- Article
- Article