Stock technical analysis software

  • Guest, it's time once again for the hotly contested and exciting FoH Asshat Tournament!



    Go here and fill out your bracket!
    Who's been the biggest Asshat in the last year? Once again, only you can decide!

Soriak_foh

shitlord
0
0
Fulorian said:
Blah blah academic bullshit. It"s not hard to beat index averages by doing a little homework. Those statistics are compiled by PhD"s who scurried into academia after being failures at Wall Street. They make up shit like the efficiency frontier to feel superior about their failures. Trust me, I know: I live it every day (MS Finance & CFA 1 almost complete).
I hate data, too. It"s obvious we shouldn"t actually compare the performance of actively managed funds, because it"s always thoseotherfunds that drag down the average.

Here"s a few stocks I 500% guarantee exceed their benchmarks over the next 5 years (on both a risk adjusted and total return basis). I"m not saying you should own them (although I would, and do), but at least study their characteristics:
I know 3 months is a pretty short time, but I was intrigued by your choices and figured I"d look at how they"ve been doing so far.

Ford (F) - Trades at 5 times trailing earnings, has amazing earnings growth, has the best management in the United States. Has pulled off a miracle, and an equivalent valuation of it against the GM IPO puts the fair price at $30/share.
November 1: $3.80
Current: $3.37
-11.3%

Goldman Sachs (GS) - Best financial company in the United States. The VERY best and brightest, pulls off a 20% ROE like clockwork but is priced at less than 1.5x book.
Nov 1: $161.52
Current: $161.77
+0.1%

Hyundai Motor (005380:KS - you have to have an international broker to touch it, but you can still find financial statements and study what they"re doing/valuations) - Best profit margins in the business, market share is growing at the fastest rate any automaker has seen since Japan entered the US market in the 70s. 3x US market share in less than a decade.
Nov 1: $19.70
Current: $21.44
+8.8%

KKR Financial (KFN) - Personal favorite. Difficult to understand their financials, but they"re trading very cheap for what they are. Mainly only held by institutional investors, less predictable day-to-day movements as a result, but a long term solid company still stuck at a discounted price from the crash.
Nov 1: $8.78
Current: $9.58
+9.1%


If you had equally distributed $100k in your 4 stocks ($25k each), this is the performance:

+$1,675 =+1.675%

Every single one of these will have doubled by 2015. Some will have doubled by the end of 2011. The S&P won"t. And academics will, 4 years later, tell you that a fund composed of these 5 stocks and nothing else are outliers, not statistically significant, and were impossible to predict in 2010.
Will check back at the end of the year.

By the way, to compare your portfolio to some benchmark:
DOW
Nov 1: 11,124.62
Current: 11,989.83
+7.78%

But as I said, 3 months is a very short time. I"ll make a note to check back at the end of the year. I presume you meant these to be buy & hold stocks, so won"t be making any changes to your portfolio.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,320
73,376
Soriak said:
But as I said, 3 months is a very short time. I"ll make a note to check back at the end of the year. I presume you meant these to be buy & hold stocks, so won"t be making any changes to your portfolio.
There"s a reason why people are reluctant to tell you what they moved on until after they see how it does, if Fulorian is wrong he will undoubtedly be corrected for some time because this thread isn"t going away.

What"s not fair though is that last trading day F took a massive dip because of their earnings missing their estimates. If you had run those numbers on Thursday the picture would be much different.

Hebbin said:
Be sure to take broker fees/fills/taxes into account with your model.
Thanks. I"ve taken into account the broker fee, but I haven"t taken into account taxes yet. The broker fill (I"m guessing that"s the deviation between the price at which you place the order and the price at which you buy/sell) is one of the harder parts of the simulation for me to emulate at the moment, and without intra-day information I don"t know of a good way to do it.
 

prescient

Silver Knight of the Realm
97
5
Tuco said:
What"s not fair though is that last trading day F took a massive dip because of their earnings missing their estimates. If you had run those numbers on Thursday the picture would be much different.
That"s fine and all but stocks that miss their numbers tend to drift downward over time, and tend to miss their next estimate as analysts don"t tend to full incorporate bad news. Interestingly, on longer time horizons there tends to be little correlation between past earnings and future earnings as companies on average are mean reverting. That"s why contrarian investing actually works.
 

Soriak_foh

shitlord
0
0
Tuco said:
There"s a reason why people are reluctant to tell you what they moved on until after they see how it does
Of course - people never mention the trades that didn"t work out, only the ones that did. Which is why there isn"t much value in such posts. Anyone can recognize a pattern after it worked out, and simply not mention the cases where it didn"t. But he"s saying current research is invalid, and that he"s guaranteeing (500%) that these stocks will double by 2015... that"s a bold claim. Of course it"s not scientifically valid (n=1) and it"s quite likely his picks may work out. If not his, then someone else on the internet assembled a perfect collection of stocks and will make the very same claim at the end: "I always knew those would work out". Since we know there are people who perform worse than the market (that is worse than average), it follows pretty much from definition that there must be people who beat the market.

My point is merely that you cannot tell in advance who"s going to beat the market, and that people don"t do so consistently. Prescient mentioned mean reverting, which tends to apply to funds as well.

What"s not fair though is that last trading day F took a massive dip because of their earnings missing their estimates. If you had run those numbers on Thursday the picture would be much different.
But would that be a better reflection of their expected performance?

prescient63 said:
That"s why contrarian investing actually works.
Then again, there"s a reason some companies just perform badly. Anyone who thought, for example, that airlines would eventually make up for their years of losses would have been waiting for a very long time. And I"m sure there are many investors who are doing just that and are convinced it"s the smart play. It may be, it may not be...
 

Tuco

I got Tuco'd!
<Gold Donor>
45,320
73,376
Soriak said:
But would that be a better reflection of their expected performance?
Nah, I just thought it was "unlucky" for someone to come in here and take a look at Fulorian"s advice the day after his first pick drops 10%.
 

prescient

Silver Knight of the Realm
97
5
Soriak said:
Then again, there"s a reason some companies just perform badly. Anyone who thought, for example, that airlines would eventually make up for their years of losses would have been waiting for a very long time. And I"m sure there are many investors who are doing just that and are convinced it"s the smart play. It may be, it may not be...
There are quite a few academic papers out there exploring this. The one I can think of off the top of my head has an alpha of 8%. I"ll dig up the paper later today.
 

Fulorian

Golden Knight of the Realm
104
46
3 months is not nearly a long enough sample size. Ford"s results, for example, are a result of their most recent "results" - which is to say, a panic sell at the bottom-line number. An idiotic move, when you consider the fact that the bottom-line number is useless in this instance. The company pre-paid a bunch of debt, which was accounted for as losses. Backing that out is impossible as of yet, because the 10-K has not yet been released to show the full performance of the quarter.

There"s nothing to stop F from being above $18 again by March, once the full results have been released and analyzed.

It"s still not quite long enough, but I"d be happy with a Nov 1, 2011 comparison of results.
 

The Ancient_sl

shitlord
7,386
16
Soriak said:
I know 3 months is a pretty short time, but I was intrigued by your choices and figured I"d look at how they"ve been doing so far.
Not only is it a pretty short time, it"s a complete waste of time. Focusing on 3 months of stock activity is exactly why most people can"t beat the market.
 

Fulorian

Golden Knight of the Realm
104
46
The Ancient said:
Not only is it a pretty short time, it"s a complete waste of time. Focusing on 3 months of stock activity is exactly why most people can"t beat the market.
Any why Wall Street is broken. There is no foresight or time horizon beyond the next quarter any more. Analysts aren"t analysts any more (at least on the sell side - hint, they aren"t on your side with their recommendations), they"re salesmen and pump & dumpers, who parrot the guidance they"re given and find a way to turn it into a Buy rating (not hard).

If you aren"t focused on at least a 2 year+ horizon, you aren"t an investor, you"re a trader/speculator.
 

tyen

EQ in a browser wait time: ____
<Banned>
4,638
5,164
Saw some of these pop up. Posted for the lulz
 

rinthea_foh

shitlord
0
0
Fulorian said:
Any why Wall Street is broken. There is no foresight or time horizon beyond the next quarter any more. Analysts aren"t analysts any more (at least on the sell side - hint, they aren"t on your side with their recommendations), they"re salesmen and pump & dumpers, who parrot the guidance they"re given and find a way to turn it into a Buy rating (not hard).

If you aren"t focused on at least a 2 year+ horizon, you aren"t an investor, you"re a trader/speculator.
you make it sound like there was a time when it wasn"t like this

wall street is doing its job fine, extracting money from its clients
 

Elerion

N00b
735
46
It"s gotten worse, and will continue to get worse. The reason is that it"s nearly impossible to get paid properly for research alone.

- Digital distribution and the increased speed of information means that it"s almost impossible to restrict information flow to reach only your clients. All your research is publicly available very quickly.
- Reduced reliance on good brokers to place orders means the trades don"t necessarily go through the banks that prepare the research. The research reports are read, and then the trades are made through low cost online brokers that don"t spend resources creating research.

If it was possible to restrict your research to only those clients that actually pay for it, and all trades based on that research had to be made through the bank that wrote it, there would be a niche for pure, unbiased high quality research companies.
 

rinthea_foh

shitlord
0
0
I don"t disagree that it is not good. But it was like this 1 or 200 years ago since markets began.


Sure discount brokerage changes the way research is deployed. But there is still a market for it and it is still able to be monetized easily. From a buy side point of view, is it better buying your research through a broker/dealer relationship or an independent point of view?

There are many ways of keeping information confidential. Subscription based models work fine. Leaks serve as advertising. Delivery first of information is all that counts anyway. After one has the information and acted on it, its in ones interest to have it out there as widely as possible.
I talk with other traders and we share information. But I guarantee, this sharing increases subscription numbers to research providers.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,320
73,376
Quick update: I"ve finished the unit testing framework for all my existing analyses () and am now working on real signal processing work with trend detection and patterns (triangles/pennants/headandshoulders etc). I"m getting less and less motivated with this project because I"m getting less and less convinced in its validity, but I"ll still commit to finishing this part, which was really the original goal of the project.

If anyone knows any software packages (open source or no) with head and shoulder (Or other complex patterns) detection let me know.




I"m being increasingly aware of a certain need for statistical analysis to be incorporated into my system. One challenge is determining a confidence value for a signal (and then incorporating several confidence values from different analyses together). I haven"t really seen anyone calculating or measuring these, so if anyone knows of models for this let me know.

The two approaches I"d have to take are obvious: Assume that a very "pure" pattern is more statistically significant or measure the pattern and determine how closely other examples of the same pattern have predicted movements with similar measurements.

For example if we measured apennantand found it had 5 bounces in its consolidation period that lasted 3 weeks and the deviation in slope was X with a increase of Y amount of price in Z amount of time in the previous trend and a volume pattern of ABC, once a breakout occurs we could then detect pennants in symbols over D time period and pick out ones that matched the same measurements and see how confident we are that it will increase in price enough to make a purchase.

I"m not convinced that either of these two approaches is valid, but later this year I will probably try the second approach. Once I finish the pattern detection in this next block I will probably put this project down for an extended time and relearn matlab/statistics (And beat Dragon Age 2 and the Witcher 2)