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.
Big Site Update
I have used my position as CEO, Chairman of the Board (as well as the only member of the board) and Director of Research to make some major executive decisions without consulting anyone at all, LOL.
There are major changes to the site. You can see a summary here:
http://www.market-geeks.com/news.php
I can give reasons (not necessarily sane ones though) for the various decisions in case anyone's interested.
That XLK trade ended the week up $117. However that will be the last update on the hypothetical account.
Big changes coming.
Gonna shake things up.
Research Update
Here is the Pending Research List from the Market Lab. The actual Pending Research List page on the site has more detailed descriptions of each research project. This list is subject to change as other more pressing projects come up and bump some of the others down toward the bottom.
1. Define and explore the concept of "Redundant Patternsets."
2. Create and test a new group of patternsets to include classic TA indicators.
3. Explore all top-ten patterns at the event/condition level.
4. Examine the "best patternset vs. best pattern" question.
5. Explore methods of expanding BPA to include multi-bar directional projections.
6. Can BPA be used for fundamental indicators?
7. Is there anything to Fibonnaci retracements?
8. Is there anything to "pivot point" analysis?
9. Create and test a new group of base-3 patternsets to use with penny stocks.
Here is the detailed description of the first research project:
==============================================================
Define and explore the concept of "Redundant Patternsets."
Added: 5/24/06
Example: Does Patternset #4 "include" Patternsets #1 - #3, making them redundant? Or is there unique information in those three patternsets that is not captured by Patternset #4? I expect that this project will be exclusively analytical as opposed to experimental, meaning that I will be defining some terms and attempting to find a proof of one of the propositions above. An answer to this question will shape how we choose future patternsets.
===============================================================
I'm beginning work on this one now.
Cap, I agree. Googledy-Dunderhead-Blinkety Theorem will definitely have a lasting effect on how people trade sometime in the future.
It's just a matter of figuring out what it is.
Hypothetical Account Update
I was sick last night so I couldn't post, but the account purchased 450 shares of XLK at 20.50 on Monday. That was the worst price of the first full hour. The rest of the trade info is available on the homepage in the usual place. The trade is down about $104 so far.
I hope our research starts to bear some fruit before the account runs out of money! I'm working on the Pending Research List part of the Market Lab right now to get it up to date. You know...just sweeping and tidying up the lab, polishing the test tubes and so forth.
Even when all that stuff is done, there will always be more research to do. Research into market behavior is the real raison-d'etre for the site. The Market Lab is really the central focus.
Eventually, if the research leads to consistently good predictions, people may join the site just to get the full range of analysis from Trading Operations, but that's not the real focus.
I've decided just to keep the one hypothetical account history going as it is now. This will continue to serve as sort of a history of how well our research has gone throughout the life of the site. So maybe in 2010 someone will look back at the history and say, "Wow, their methods really sucked for about 2 years, but then the performance seemed to turn around in the summer of 2008. That was right around the time when they discovered the Googledy-Dunderhead-Blinkety Theorem and started using it to trade."
LOL, at least that's one possible outcome.
Wow, may you be speedy at accomplishing that and not go blind from staring at a screen and crunching numbers.
Just reading that made me begin to have feelings of dread and memories of business stats class began to flow back into my vision.
OK, I've decided not to start a second hypothetical trading account because that will be just way too confusing. However, there are several issues I need to investigate:
1. Should I be using the best PATTERNSET found to date for a particular instrument or should I only be tracking the best PATTERNS found to date? This PATTERNSET vs. PATTERN issue is crucial to figuring out how to use BPA to trade.
2. I need to study the top-ten pattern data that I have, down to the level of the specific events and conditions that make up each pattern. I want to know if certain events/conditions (like breaking to new highs/lows, MA crossovers, price reversals or whatever) turn up frequently in high-scoring patterns. I also want to know if high scoring patterns are generally clustered in the same patternsets, but this is tied in with the PATTERNSET vs. PATTERN issue.
3. I need to create and test many more patternsets. These will include sets that make use of all the standard TA indicators like RSI, MACD etc. I should probably do this before I do #2 because then I'll have more data for #2.
Anyway, the place to do all these projects is of course in the Market Lab of http://www.market-geeks.com. So my next step is to update the Pending Research List with these projects, and then get to work on them!
Although the detailed research notes and data will be available to members only, I'll cerainly post the major findings here on the forum.
The research lab is open.
Let's go to work...
Hypothetical Account Update
This week's trade is posted on the homepage. It's a buy of XLK.
I may start a second hypothetical account which picks the best scoring trade from the trading alerts each week. I've been getting the impression that the trading alerts might be performing better than the standard weekly plans.
The trading alerts are based on the top ten performing PATTERNS ever found for each instrument as opposed to always using the same PATTERNSET for each instrument. I'm starting to think that the idea of a "best patternset" for a given instrument is misguided. However, it may also be just that the 14 patternsets I have right now aren't very good.
By comparing the performance of the "best patternset" (shown on the standard weekly plans) approach to the "best pattern" (shown on the trading alerts) approach, I can investigate this.
I won't abandon the idea of a "best patternset" yet though until I try out many more patternsets than just these original 14. I also want to study the elements involved in the top ten pattern lists using a database to see if patternsets are a factor in making a good pattern, or if it's really the underlying events and conditions of which the patterns are composed. Kind of a "top down" vs. "bottom up" question that I would like to study.
So much research to do....
Out here!
Hypothetical Account Update
That QQQQ trade stunk on ice...
The price didn't hit the stop or the target, but it closed on Friday at 39.35 giving us a loss of 219.48 for the week.
I'm updating the site and will post the trade for next week sometime tonight or tomorrow night.
Lot of net losing trades though. This week's was another one.
Hey, sounds ok. I guess your stopped out positions must have been the only ones I recalled.
Site Update
FOREX is here! Here's the full news story:
http://www.market-geeks.com/news.php
Hi Matt,
You're right, the performance wouldn't be nearly as bad in real life for a number of reasons.
1. The actual entries would be better than what I'm showing, as you just pointed out.
2. In real life I would try to avoid letting a good gain turn into a loss, so I would move my stop during the week to lock in gains.
3. I would use the stop and target prices as a GUIDE and actually set them at logical places on the chart instead of blindly following the BPA numbers.
Remember though, this is a test. I'm not trying to show that BPA works. I'm trying to see if I can prove that it doesn't work, so I don't want to give it any breaks.
On the stop issue, I just did a count of the completed trades shown here:
http://www.market-geeks.com/directorAperformance.php
There are 10 completed trades, but only 3 have been stopped out. The other losing trades have just closed on the wrong side of the entry.
I'm trying to poke holes in BPA in order to find as many flaws as I can in order to fix them. I think I've just figured one of those flaws out. The current QQQQ trade is going south, but today's action was driven by news (CPI was higher than expected I think). BPA assumes that prices are completely driven by TA. So I'm thinking that BPA may not work nearly as well on a weekly chart as it might in an intraday chart which is more TA driven. Another subject for future research!
Cap, I think you should take the average of the first 100 trades or something like that to determine your position. With only 10k your purchase/sell is not going to move the market and with the liquidity of these indices it wouldn't be a problem.
Considering what was presented as fact, historical data, in the article that I posted here a week or two ago I think that your method of testing BPA is flawed. At the very least you're not giving yourself enough of a chance.
If it sets the day's high in the first hour and you choose that as your fill and as the article suggests that is the highest high of the day you just shot yourself in the foot. I think this is why you get stopped out quite often.
Surprise Site Update!
I haven't posted about new classes being available, because I've been knocking them off at a pretty fast clip. For a lot of the market-related classes all I really had to do was put in some links to pages on Investopedia.com or Stockcharts.com and leave it at that. No reason to re-invent the wheel when their sites are already filled with great info.
So I'm happy to announce that the Geek Academy is now finished! At least until I see a reason to add more classes someday.
Although most of the market-related classes are probably pretty basic to many on iHub, I would recommend taking a look at the one on the Random Walk Model. I have a personal rant in there about the book "A Random Walk Down Wall Street."
I would also recommend taking a look at the class called "Survey of New Frontiers in Market Analysis". This is a tour through a "wish list" of ideas for market analysis that are kind of "out there".
Anyway, now I can move onto the next step which is to add FOREX currency pair analysis to the Members Area.
Hypothetical Account Update
OOPS! I forgot to post Monday's fill again. Using the "worst price of the first full hour rule" the hypothetical account bought 236 shares of QQQQ at 40.28 on Monday. Our stop is 37.38 and our target price is 43.18 as shown on the homepage. The trade is currently down about $99.
Hypothetical Account Update
Ouch! The XLK trade was stopped out on Thursday for a total loss of $229.55 on the week. Next week's trade is a buy of QQQQ. Details are posted on the homepage.
Also, one of the top-ten individual alerts in the Members Area this week is a long XLU trade with a PTRL score of 4.26%. That's the highest I've ever seen. I'll have to remember to see what the utilities do this week, but that high PTRL score is a result of the rarity of the pattern. It has only occurred twice in the past 200 weeks.
Hmmm...and that leads me to think that BPA will produce lots of statistical artifacts when the pattern is rare. So frequency of pattern occurrence should be a factor in the score somehow...
EDIT: Ok, I don't have to change the scoring system. What I have to do is choose trades based on pattern frequency as well as PTRL score. A high score doesn't mean much if the dataset consists of only a very few price bars. Soooo... I guess if I multiply the number of times the pattern occurred by the PTRL score, that will give a more realistic ranking of the trading opportunities.
In fact, that's actually why I put the information on the number of times that pattern occurred in the past 300 bars on the full analysis page. Users can look at that and say, "Great score...oh wait a minute, it's only based on four bars? Forget it. Let's see what else is available." I might as well start using that information myself. Starting next week I think I will choose trades for the hypothetical account in this new way.
That's excellent! This is a great example of the kind of market research that Market-geeks.com is set up to facilitate.
Thanks Matt!
TradingMarkets.com
Here's What You Can Learn From The First Hour of Trading
Monday February 27, 9:50 am ET
By TradingMarkets Research
A reader recently wrote to me asking a research question. It seems as though a market commentator asserted that the S&P emini futures contract (ES) either makes its high for the day or its low during the first hour of trade on 71% of occasions. This struck my reader as a potentially useful piece of information, but he wondered if it is accurate. So I decided to look.
ADVERTISEMENT
This is not the first time I've investigated the timing of daily highs and lows. I wanted to see the odds of the market making its daily high or low during the first and last hour of trade. The results were interesting. I found that the daily high occurred during the first or last hour three-quarters of the time. The daily low occurred during the first or last hour approximately 60% of the time. Equally interesting, the midday hours only saw daily highs or lows approximately 10% of the time, although they accounted for one-third of time in the market.
The way I've used this information is to think in terms of "candidate daily highs and lows". When we sell off sharply in the first hour or run up significantly, I treat the high or low from that first hour as a candidate daily extreme. I then monitor the day's volume as we go along to estimate the market's likely range for the day. That gives me some clue as to how much the market might move away from this daily extreme. On the other hand, when we do make new highs or lows for the day during the midday hours, I make the assumption that these will not be the ultimate day's highs or lows. I then look for entries that will allow me to ride the trend into the final hour.
Of course, a 36 session sample is not much to go on. Prodded by my reader's inquiry, I went back to January, 2004 (N = 543 trading days) and examined the number of occasions in which the day's high or low was registered in the first hour. It turns out that we saw daily highs in the first hour on 177 occasions and daily lows in the first hour on 179 occasions. All in all, we saw either a daily high or low in the first hour about two-thirds of the time. It's not quite the 71% quoted by my reader's source, but it's not far from it.
So what can we do with this information? It seems to me that, if a market makes its high or low in the first hour, it is rejecting that high or low price as value. That rejection should manifest itself as significant selling or buying following the candidate high or low. Once we see the market hit a price extreme in the morning and then reject this with conviction, we now have the basis for a trade idea. The first retracement from the rejection move (i.e., the first bounce following a selloff from a candidate high or the first dip following a rally from a candidate low) should provide a relatively low risk entry for a short-term trend-following move.
A great deal of fruitful research could follow from this tendency. For instance, we could examine the early signs of rejection of candidate highs and lows so that these can be identified as quickly as possible. We can also examine the risk/reward of holding trades that reject first hour extremes until the final hour. Perhaps most intriguingly, we can explore whether this pattern plays itself out in individual equities and conduct real-time scans for signs of rejection of first-hour extremes to pursue this pattern among volatile stocks. While historical patterns may not always play out in the present, this one seems particularly durable. Stay tuned.
Brett Steenbarger, Ph.D.
Site Update
The first class in the Financial Markets course is now available in the Geek Academy. It's the class entitled "Survey of the Financial Markets".
Most of the classes in this course will not have much new to say to veteran iHubbers who are familiar with the markets. They are meant for an audience that may be well versed in math and science, but is less familiar with the financial world.
Hypothetical Account Update
The account purchased 444 shares of XLK at 22.00 which was the high of the first full hour on Monday. Our stop is therefore at 21.48 and our target is 23.04 as per the plan I posted on Saturday. As of now, the trade is down around $22.
Site Update
Copied from the News Archives:
http://www.market-geeks.com/news.php
The entire "Key Topics in Science and Mathematics" course of study in the Geek Academy is now complete. I finished writing "Backtesting & Datamining" last night, which was the last of the nine classes in this course.
I'm thinking there will be two main audiences for the Geek Academy. One will consist of experienced traders who may not be as well informed about the "academic" stuff. This group won't find much new in the market related classes, but may be interested in the science and math classes. The other group will consist of academically oriented people for whom statistics and the scientific method are old hat. However, some in this group may not know a stop order from a stop sign, so they may find the market related classes more interesting. One purpose of Market-geeks.com is to bring these diverse groups together in order to combine their respective strengths.
Anyway, it's time for me to begin working on the course of study in "The Securities Markets." Hopefully these classes will take me less time to write as I'm more familiar with the subject matter.
Hypothetical Account Update
The trade for next week is posted on the homepage. It is a buy of XLK this time (go figure...) with a much stronger PTRL score for the pattern. The win/loss ratio is 2:1 on this one.
The account will attempt to buy 444 shares of XLK at 22.02 or lower on Monday.
Hypothetical Account Update
XLK didn't hit the stop or the target prices during the week, and closed at 21.92, giving us a loss on this trade of $13.41.
I'm doing the weekend update for the site. I'll post next week's trade tonight or tomorrow night.
That's good news.
LOL, it's only up about $54 as of today's close.
However, that constant string of losing trades stopped right after I made that second fix to the optimization logic; the one about setting the stop and limit prices based on the ACTUAL entry instead of the PLANNED entry.
I guess that must have been kind of important, LOL!
Weeeeeeeee! $100 bucks!
Site Update
The class on Probability and Statistics is available in the Geek Academy.
Wait until you see how lazy I was in creating this one, LOL!
Hypothetical Account Update
XLK opened at 21.85 this morning and hit a high of 21.96 during that period before heading back down. We make the conservative assumption that we sold short at our limit of 21.89 as opposed to a possible higher price. The hypothetical account is now short 447 shares of XLK.
Our stop is at 22.31 and our target is at 21.06 as shown on the homepage.
As of the close at the price of 21.67, the trade was up about $98.
Site Update
I have completed the class entitled "Introduction to Combinatorics". It is now available in the Geek Academy.
Hypothetical Account Update
The free trade for next week is posted on the home page. It is a short sale of XLK, the technology sector ETF at a limit of 21.89.
The hypothetical account will attempt a sale of 447 shares of XLK on Monday.
Hypothetical Account Update
XLU did not hit the stop of 30.07 or the target of 32.23, so we assume an exit at the close of 31.32. The profit on this trade was $52.53.
I'll be updating the site tonight with analyses for next week.
Nice computer..
I feel your excitement..
OT Just bought a computer through Dell w/ tax refund
Supposed to arrive on Thursday. I'm fairly excited since it will be like a home entertainment system in and of itself.
XPS 400;Pentium® D Processor 820 with Dual Core Technology (2.80GHz, 800FSB)
Operating System: Genuine Windows® XP Media Center 2005 Edition
Media Center Enhancements: Single TV Tuner with Remote Control
Memory: 1GB Dual Channel DDR2 SDRAM at 533MHz- 2DIMMs
Hard Drives: 250GB Serial ATA Hard Drive (7200RPM) w/DataBurst CacheTM
CD or DVD Drive: Dual Drives: 16x DVD-ROM Drive + 48x CD-RW Drive
Monitor: 19 inch UltrasharpTM 1907FP Digital Flat Panel
Video Cards: 256MB PCI ExpressTM x16 (DVI/VGA/TV-out) nVidia GeForce 6800
Sound Card: Sound Blaster® Audigy®2 ZS (D) Card w/Dolby 5.1, and IEEE 1394
Speakers: Dell 5650 5.1 100 Watt Surround Sound Speaker System with Subwoofer
Network Interface: Integrated 10/100/1000 Ethernet
lol, nope. I've been working in the insurance and financial services industry. Primarily life/health insurance, since I have an investment team that does only that.
Always looking for new opportunities though. My dream job used to be working on a trading desk in fixed income or equities. Never really considered arbitrage.
Hypothetical Account Update
I forgot to update the entry, stop and target prices on the homepage last night, but it's updated now.
XLU dropped today, but didn't get anywhere close to our stop of 30.07. We are now down about $19 on this trade.
I thought you were in the annuity racke... errr, I mean business! :=P You didn't get laid off too did you??
Well I'll make sure to send you my resume.
Site Update
The class on Binary (base-2) numbers is available in the Geek Academy.
Hypothetical Account Update
I'm assuming this week's trade is going to go sour soon because if it doesn't we would have two good ones in a row. Unheard of!
Anyway, the account purchased 309 shares of XLU at 31.15 during the first full hour of trading. The way I arrived at the prices is described on this page:
http://www.market-geeks.com/directorAperformance.php
Here's the pertinent rule in this case:
"I find the first full hour (thus ignoring the wild swings in the opening minutes) in which the trade could have filled based on the limit price, and make the conservative assumption that the order will fill at the worst price for that hour (so if we're buying, we fill at the high for that hour and vice-versa)."
The trade was up by about $121 as of today's close.
Stay tuned!
I appreciate your thoughts! Actually, you hit upon a longer term goal which is to create a full scale trading company...so maybe I'll be the one hiring the arbs someday, LOL!
Yep, that makes perfect sense and I hope you're right about that. It was just a thought that came to mind that I thought I'd share with you.
Hi Matt,
My degree is in Civil Engineering, and I had a Series-7 and 63 but they expired in November. But it's not for me...here's why:
I've been laid off three times and am sick of it.
I would have a boss. I'm sick of that too.
They would require that I give up my trading accounts outside their firm, so I wouldn't be able to have my Fidelity, Ameritrade/Izone and FOREX accounts.
They wouldn't let me sell short or take bearish option positions.
Why create all this fantastic stuff for some company when I can create and use it myself, right?
I'm hoping the job I have now is the last one I'll ever have...
Hey, just a thought here but I know that you have a degree in engineering and something else, I think. With your experience at Fidelity and now your strong grasp of statistics and modeling I was wondering if you might be interested in looking for a job at an arbitrage desk somewhere.
Those folks create and run software to arbitrage the tiniest amounts in the treasuries, etc. They also make some nice bank.
Site Update
The class entitled "Survey of Mathematical Topics" is now available at the Geek Academy. All of the general science classes are completed, and this is the first of the math oriented classes.
Hypothetical Account Update
The trade for this coming week is posted on the home page. It's a buy of XLU, the utilities ETF. The account will attempt to buy 309 shares on Monday at the limit price of 31.47.
Site Update
The class on "Experimental Design" is now available in the Geek Academy.
This one starts off with an example about radishes to illustrate the process of designing a good experiment, and then we use what we've learned to design an experiment involving moving average crossovers.
"From radishes to rallies in one easy lesson!"
I stopped completely and had planned on resuming in March:
http://www.investorshub.com/boards/read_msg.asp?message_id=9484591
The simple "back to basics" trend following system I was using for stocks seemed to be working ok, however, I decided to finish paying off my car early first and then start moving my part of my excess income into trading accounts. So now I'm shooting for June 1.
I'll post about my own trading over at Cap's Trading Diary, including trades that I make based on reasearch done at Market-geeks.com. I'm thinking basic stocks, some options and FOREX will occupy my trading plate, but I'll be putting a lot of $$$ into more conservative income instruments too to build up the nest egg.
Did you stop your personal trading or are you just too busy with the site updates to post trading updates?
Followers
|
4
|
Posters
|
|
Posts (Today)
|
0
|
Posts (Total)
|
187
|
Created
|
02/27/06
|
Type
|
Free
|
Moderator Capitalist | |||
Assistants |
Volume | |
Day Range: | |
Bid Price | |
Ask Price | |
Last Trade Time: |