Tag Archives: successful

Youtube demo on How to Learn HFT algo strategy development software forex database for successful profitable trading

Youtube demo on How to Learn HFT algo strategy development software forex database for successful profitable trading

 

http://quantlabs.net/membership.htm

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

What’s your biggest obstacle to becoming a successful quant trader?

What’s your biggest obstacle to becoming a successful quant trader?

Hi there,

Lately I’ve been getting quite a few questions from this list on various aspects of quant trading.

You’ve been wanting to know exactly how I can help you get started. However, it’s occurred to me that maybe the best questions have NOT yet been asked! So allow me make a request of you right now:

What’s your most burning question (or biggest pain point) with quant trading?

What is it that you most need to learn?

What’s your biggest hurdle to entering the quant world (or getting better results)?

What do you most want to learn from me that will save you time and frustration ASAP?

Reply  by commenting below right now and I’ll collect all your questions promptly. And I promise to answer your 5 best questions next week! Stay tuned …

Bryan
Quantlabs.net
–> http://quantlabs.net/dlg/sell.php?prodData=m%2C3 <–
–> http://quantlabs.net/quant-member-benefits/ <–

 

Hi there,

Lately I’ve been getting quite a few questions from this list on various aspects of quant trading.

You’ve been wanting to know exactly how I can help you get started. However, it’s occurred to me that maybe the best questions have NOT yet been asked! So allow me make a request of you right now:

What’s your most burning question (or biggest pain point) with quant trading?

What is it that you most need to learn?

What’s your biggest hurdle to entering the quant world (or getting better results)?

What do you most want to learn from me that will save you time and frustration ASAP?

Reply to this email at feedback (AT) quantlabs.net right now and I’ll collect all your questions promptly. And I promise to answer your 5 best questions next week! Stay tuned …

Bryan
Quantlabs.net
–> http://quantlabs.net/dlg/sell.php?prodData=m%2C3 <–
–> http://quantlabs.net/quant-member-benefits/ <–

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

What’s your biggest obstacle to becoming a successful quant trader?

What’s your biggest obstacle to becoming a successful quant trader?

Hi there,

Lately I’ve been getting quite a few questions from this list on various aspects of quant trading.

You’ve been wanting to know exactly how I can help you get started. However, it’s occurred to me that maybe the best questions have NOT yet been asked! So allow me make a request of you right now:

What’s your most burning question (or biggest pain point) with quant trading?

What is it that you most need to learn?

What’s your biggest hurdle to entering the quant world (or getting better results)?

What do you most want to learn from me that will save you time and frustration ASAP?

Reply to this email right now and I’ll collect all your questions promptly. And I promise to answer your 5 best questions next week! Stay tuned …

Bryan
Quantlabs.net

è http://quantlabs.net/dlg/sell.php?prodData=m%2C3 ß
–> http://quantlabs.net/quant-member-benefits/ <–

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

What are the main reasons a very successful backtested strategy means so much less than a medocre live strategy to potential employer?

What are the main reasons a very successful backtested strategy means so much less than a medocre live strategy to potential employer?

 

Could you be more specific? Which platform?, which instruments? (Are they Mifid regulated) which Broker, does your algo. include derivatives. etc.etc.

But generally, a backtested strategy is not the same as live trading – Quite simply said, “the proof is in the pudding”
As an example if, you use server controlled software like Metaquotes (MT4) you can have fabolous results using backtest. BUT as soon as you go live with your algorithm, it will fail completely, and the market-maker will make a lot of money on you.

Reason beeing that they will control the quotes they send to you, and those quotes are not as attractive as those send to you in a backtested environment.

In our environment (Not retail) we do not offer backtesting in our software, instead the clients test their algos. using real-time quotes and send their order to synthetic exchange(s) which will match accordingly.
This takes a lot longer, but you will have the same environment as live trading.

 

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

How long does it take to become successful at trading and investing?

How long does it take to become successful at trading and investing? –http://revolutionarytrading.wordpress.com/2011/09/08/how-long-it-takes-to-become-profitable-at-trading-and-investing/

revolutionarytrading.wordpress.com

As a human being, you will make mistakes.  None of us are perfect and, to me, that is something that is perfect about us.  The point being that if we were theoretically ‘perfect’ then life would …

—–

Enjoyed reading the post. Most importantly the way you have focused on our internal reactions to market actions and twisting them to positive experiences. Thanks.

 

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

How successful is automated trading when almost all the world’s major markets are falling? Can any strategy save you

How successful is automated trading when almost all the world’s major markets are falling? Can any strategy save you or you rather stay away from market?

—-

Can do both (save you and put you in cash).

If your strategy can go both long and short, then volatility can only help you.

If you are monitoring the performance of your systems, then there can be a signal to go to cash (ergo no volatility).

—-

Don’t think so. One of the person using tradestation lost about a million using tradestation and has to manually cut his trade…
still, could u explain more?

A short term trend following strategy could have done very well in this market. Here is the intial blog post in a series about a filter to avoid extreme markets:http://marketsci.wordpress.com/2008/10/10/when-adaptive-systems-fail-and-what-i%E2%80%99m-going-to-do-about-it/

Correlations often come closer to 1 when there is panic (Ralph Vince has some good books on that topic). People tend to not include in their analysis:

Correlations between asset classes are not static or linear. (and yes, cash is an asset and can be a position).

The nature of strategies is constantly changing. As more people know about a strategy, the less profitable is becomes (hence, the profitability of a SP-500 mean reversion intraday strategy vs the old Turtle rules from 30 years ago).

Those 2 issues are usually not included in strategy backtesting. In addition, software programs like Tradestation do not let you backtest a portfolio of strategies – hence an incomplete picture.

—-

Let’s try to explain a little more:

A trading strategy can take a short trade as well as a long trade. If your strategy can take short trades, then even if a market is trending down, you can make profit.

Something like 90% of stock traders don’t know how to go short.

If you are analyzing your strategy to see if it is making money, or losing money, when your strategy starts to lose money, you can turn off your strategy, which would be “putting you in cash.” Which Anthony pointed out — is also a position.

In other words, doing nothing, taking no trades, and keeping your money in cash, is a very valid position. Most stock traders do this when they sense trouble in the market, or algorithmic traders when their strategies start doing poorly.

If you need more clarification, please feel free to contact me.

Good luck!

 

Actually when the markets become saturated with “autobots”, I switch to manual. An over-crowded playing field creates a vacuum as everyone’s “Bots” are draining spreads as they keep trying to “one up” their price points. I turned my bot off this morning (it made me good money the last couple weeks) and grabbed the wheel myself. It is good to shift gears

 

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

You may want to read this: The Silicon Valley Way, Second Edition: Discover 45 Secrets for Successful Start-Ups

You may want to read this: The Silicon Valley Way, Second Edition: Discover 45 Secrets for Successful Start-Ups

 

This is a very serious old school you may actually learn from. It was also written before the internet was even born which takes out all the junky noise you get distracted by.

http://www.amazon.com/Silicon-Valley-Way-Second-Successful/dp/0982796110/ref=cm_cr_pr_product_top

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

If you would like to have a successful career in the hedge fund industry

If you would like to have a successful career in the hedge
fund industry you should watch our 90 minute hedge fund
career webinar:
http://hedgefundcertification.com/Free-Webinar.html

If you like the webinar you may also be interested in
checking out our Hedge Fund Career Library as well:
http://hedgefundcertification.com/Library.html

By the way, our Certified Hedge Fund Professional (CHP) program
opens to 200 new participants on Thursday July 7th, 2011 at 2PM EST. If you would like to review the benefits of completing this training program please see this page of our site: http://hedgefundcertification.com/Benefits.html

I hope you and your team enjoy the Hedge Fund Career Webinar
& Resource Library!

– Richard

Richard Wilson
Hedge Fund Group
http://HedgeFundCertification.com

Richard Wilson is head of the 50,000 person Hedge Fund Group
and the Certified Hedge Fund Professional (CHP) designation.

The author of 10 books, Richard has written the #1 most popular
book on hedge funds, and he is a global speaker who has provided
industry training in in New York, London, Singapore, Chicago,
Hong Kong, Boston,Tokyo, Brussels, Monaco, and Moscow. Richard most recently served as the opening day chairman at the largest hedge fund conference in the industry, GAIM International 2011.

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

successful end-of-day algorithms just don’t do well on 30-minute or hourly data

wonder why my successful end-of-day algorithms just don’t do well on 30-minute or hourly data. Anyone with same experience or thoughts?

6 days ago

 

MyTechTrades Home Page mytechtrades.com

Home Page for MyTechTrades.com technical stock trading website with backtested easy systems that…

 

—]

in my opinion, successful EOD algorithms normally translate well into intraday trading strategies as far as defining the LONG/SHORT channels, that’s about all I can say without knowing what algorithms are used and how you apply them at the 30 minutes level.

Thank for posting this discussion and good trading

From my experience, EOD can do well using particular models that might not work well for intraday trading in many cases. The reason could be that EOD movements based on daily averaged evaluation of all inputs during a long period while intraday decisions are driven by each news input and fast traders technical signals. So that there are some methods that work better for EOD, others for intraday

When you move from EOD to M30 or H1, not everything “scales down” appropriately. In equity markets [as opposed to forex], each and every daily bar starts at a discontinuity from the previous day’s bar. For intraday bars, however, there is a discontinuity every 7/17/etc. bars. Can you algorithms handle these properly?
Also, while the difference between a day’s close and the next day’s open remains the same in dollars, it is much larger (for hourly bars) in terms of average bar size (e.g. ATR). Finally, commissions and/or spreads may now “eat up” more of your trade size (read: profit), taking away from gains

Momentum has a larger influence in EOD trading as compared to day trading. While some systems can handle both time-frames, I haven’t found any one system that does both well.

When you increase trading frequency, you usually do more trades and have better trading statistics but you are more influenced by micro-effects (execution) and most of all bid-offer spread versus volatility of pricechange. If your strategy changes positions a lot intraday, this bid-offer versus volatility is critical.

Thank you all for your thoughts on this. I don’t think there is any significant inconsistency among your comments. I suspect the issue of momentum signals being more effective in EOD trading than intraday trading is the essence of it. My successful EOD systems depend heavily on momentum. You can see them at www.MyTechTrades.com I will modify the systems to depend less on momentum statistics for intraday versions that I will test. Thanks again.

 

 

 

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

The key factors in successful prediction for quant neural networks

The key factors in successful prediction for quant neural networks

This was a discussion found on Linked In:
We make a lot of researches to being able to do that.

The keys are:
-Transform the problem in classification problem
-Looking at the trend in the predicted probability to make good trading strategies
-Strong financial knowledge
-Don’t trade on unstable market
-Constructing/including good explanatory variables!
-Time intervals
-Volatility

Agree?
What kind of predictive tools do you use: NNs, RBF, KNN, etc?
It’s a simple way of objective evaluation and comparison. If I make 10 trades and have a W/L ratio of 80% and you make exactly the same 10 trades in the same market at the same instant and have the same 80% W/L we can have a very different ROI as I only trade with a capital held in reserve factor of X but you trade with one of 2 or 3 X.
When I am trading (whatever), one must decide WHAT to buy/sell, WHEN, and HOW MUCH? Each affects (obviously) overall profit earned.

Success to me is growing my capital. The amount traded (HOW MUCH) is obviously crucial to any SUCCESSFUL trading system. Thus, my reason for making the point that measuring success in dollars (without relating it to invested capital – ROI) seems to take one only half the way down the road to success.
Yes, but you can invest as much or little capital as you like and if your Win/Loss ratio on 1000 trades is only 51% your capital will not grow much at all as trading costs will deplete that 1 %.

So without a decent W/L% to start with the rest is largely a useless consideration.

We use all data mining/statistic/mathematic technologies including financial time series techniques.

Most of the time our final model include an ensemble of these techniques.
Have you tried your technique on INFORMS Data Mining Contest (http://www.kaggle.com/informs2010 )?
When selecting any investment, hopefully one does consistently pick “winners” (those with gains not losses). However, in addition to identifying and measuring investment selection criteria to pick “winners” one must also evaluate the risk factors (systemic, economic, political, validity of underlying assumptions, stability of underlying model).

Selection and risk factors (variables and relationships or functions) are used in tandem to do the return/risk analysis necessary to make the decisions: (1) to proceed or not, and (2) size of investment. I also personally include a liquidity factor into my calculations. I learned a long time ago that there is no point in making great paper profits IF you can’t sell your assets to TAKE the profit.

Profit, Risk, and Liquidity (for me) represent a triad of minimally necessary components for any investment decision. One without the other two is academically interesting but not practically useful. We KNOW we have necessary and sufficient (economic) success ONLY WHEN our ROI (not just generated cash) reaches our targeted goal.

Sound like you’ve written (or at least read) a number of economic text books, but I’m wondering how you would calculate, from your statement “evaluate the risk factors (systemic, economic, political, validity of underlying assumptions…”, the political risk factors in an automated trading for Forex,say the EUR/USD, that has a maximum trade duration of 180 seconds and might trade several hundred times in a 24 hour day?

I mean how would one categorize the political risk tomorrow at 10:07 GMT vs. 10:14 GMT to know if it has changed enough to make the trade? What is your approach to change in political risk at 3 minute intervals or even 10 minute?

I have a hard enough time with just the model and price action at that time granularity so adding an econometric and political component must require a very sophisticated global data gathering network. Hat’s off to you for your achievement there!

How would you utilize your trading infrastructure in HFT where trade duration is sub second? Trying to gage political changes at 10 ms intervals could require very large scale computing resources.
Interesting comments on methodology.

One method I’ve found very useful is to do separate model for long vs. short trades.
There seems to be a different topology in price structures that has a directional component.

The standard approach is to do a model on the future change in price and assuming the output of the dependant variable has a possible range of 0.0 to 1.0 the assumption would be that a value of 0.2 means a future decline and a value of .8 means the market is likely to go up.

A useful alternative is to separate the training set into sub sets of up, down and sideways future market movement to focus model learning to distinguish, for example, strong from mild or weak up movement, and then combine the up, sideways and down models in a parliamentary voting system. I think this works well because each model is constrained in the domain it is learning as opposed to trying to learn the entre range of future market behavior.

I think we are in the pretty same field..
I like classifiers more than trackers, and trackers more than predictors. In the case of classifiers one can mention the ratio of correct classification. There are two problems with that. One is the fact that correct classifications do not make money, you still need a ratio of money made/money lost by the overall system. Classifications are better understood by the confusion matrix (false positive, false negatives, etc.). Again I care a lot about false positives, but do I care so much about false negatives?

Ww use exactly this kind of methodology in our solution :).
I trade stocks and options to a lesser degree – not Forex. I would never claim to be even competent in the Forex forum.

I was trying to make a GENERAL point. MEASURING investment success should be based on ROI – not gross (or even net of expenses) revenue unrelated to invested capital (whatever the investment and trading activity).

Granted, one can ASSuME nation/state fiscal intervention and macroeconomic events NORMALLY wouldn’t impact a 180 second event. However, as regards your hypothetical regarding Forex Euro/USD trade durations of 180 seconds, one past significant political event might include 911 (as an extreme case). Bet that fouled up your models.

I appreciate your comment regarding my Econ tendencies although I would really characterize myself as econometric (applied discrete math with a profound appreciation for meta-mathematical concepts) with just a pinch of Economics (BS from Purdue).

What I took away from my Econ was an appreciation for its fundamental concepts such as optimal allocation of scarce resources, micro-economic theory, nature of markets (free and otherwise). Money is one of my scarce resources. what I mentioned were the results of trading a uniform random distribution with a 55% chance of correctness and creating a smooth equity curve.
I am trying to understand your metric: every dollar won/every dollar that flowed to the system. It looks reasonable, but I have never seen this before. Again we are not talking about prediction, which was my original point. From strictly a game metric viewpoint I use the percent of winning trades * the ratio of all dollars won over all dollars lost. This is easy to interpret as everyone presents to you the percent of correct trades, if you compute the second ratio you get how many dollars you make in a trade. Both results are presented for a great number of systems, and it is standard to compare. Now if you multiply them, you get an overall sense of profitability when comparing systems, which is very good for optimizations. Again this is a bit off the point because this is not measuring ANN capacity to predict. So I cannot comment on your more exotic measure as it is not pertinent to prediction and not easily comparable to other systems, but it looks great, so: congratulations.
ou are right that this is closer to an L2 metrics. I love L1 metrics, which replaces the mean with the median. In this way, you avoid outliers (or gross mistakes) in a first measure, which can pull the result into a skewed interpretation. You are free to play with error metrics. I do not like the original Sharpe ratio because it penalizes good and bad outcomes. A modified Sharpe only takes into account bad outcomes only. In the case of prediction, under predict is better that over predict. For example, to say that the market will hit +20 points, and it turns out that it hit +26 points, in terms of money, these are 20 points made, and 6 left on the table, but who cares. Much better that the market hitting 19 points, and you have to get out on a stop. So one issue, even in prediction, is: do you want mathematical correctness of the prediction or do you want something closer to making money? Be creative on your metrics for system design (do not be fooled by them). Once the system is designed, fall back to traditional metrics to see if there are any leaks.
we already have spoken privately. I had a chance of pointing out some weaknesses of your competition data. So I am not entering it. Again I was the chairman of the first international competition on non-linear finance in 94. The board of the competition was made up of top researchers on their fields of prediction, and some went to make a career out of the Gov. labs into predictions, then selling the companies for a mint to Swiss bankers. Call it the first X prize for finance prediction and trading. We spent countless hours on how to approach the data problem, how to make it fair, how to avoid cheating, etc. So I have concerns. However, since you are the only taker to my offer, I am going to look at your data and see if I can transform it into a simple up/down problem and submit it to you. You win by sheer exhaustion. ;.)
About the keys, you have listed out, I like them a lot. The only one I would completely ignore is: -Constructing/including good explanatory variables! , if I knew what to look for, adaptive methods are much weaker than sheer statistical decision support approach. Here people either do not know or do not care about the explanation. The metric is: can you make money.
I have reservations about the metrics too, as per above. It is not a real ROI because that would require all the capital that is in play, even if some are de-levering the future (STOCK/OPTION) contract. The system that I deal with is measured as all years above 40% (real number is proprietary) for 25 years with a maximum instantaneous drawdown of 5% over the entire period, de-leveraging 2:1 in the S&P500 for the whole investment. I think that such measure is more consistent with an ROI. However what is good for investment understanding is not so clear for system design. I have to be clear on this point: most of the people in this thread and in all similar trends and groups on linkedin, and I dare to say elsewhere too, are really talking about designing and measuring a part of the entry signal of a trading system. Sorry to disappoint but it is not where money is made and all the effort should go.
I am not sure that we are talking about indicator inflexion points as predictors. Your site does not have enough information for me to make any judgment on it. Even if you were to carefully fully disclose it, there is a big problem that I call Future Leak. This problem plagues the majority of the designs (we can discuss this is another thread). So it is hard to offer a counter point to my skepticism without scientific testing.
I really like what you are saying about profit/risk/liquidity. Many designers do not take trading cost and slippage into their design. However, the first paragraph about looking at all trade conditions and Louis’ comments on an aggregate of different models strike a wrong note with me. I guess that I have minimalist tendencies, that is why I like Apple’s designs. The more parameters, systems and structures you have in your final system, the more likely you are to be unstable. If you are dealing with non-linear systems, the problem is far worst. Do get me wrong, back in 1990 my students, and I started the application of committees of models to prediction (a thorough back to Academia). However, markets are about the toughest prediction problems I have ever encountered and stability is at a premium. Requiring that the computer penalize complexity in the design like the modified Minimum Description Length criterion or very large out of sample selection tests will keep you honest.
your comment is extremely interesting. You have trained your systems to perform in different market conditions. This is excellent. My first paper on ANN, I trained an ANN to read aloud English and the Spanish (we also tried Turkish). At various points in the training, we switched languages. We observed the same mistakes that children make different ages (generalization). We observed the clustering effect during training (specialization). And we observed the emergence of ‘accents’, the same way that people have. So by training on specialized market conditions first, you will cause them to become specialists. The question is that you need to have an arbiter to be trained on classifying the market conditions.
Another interesting point your technique evokes is the use of a hand crafted ‘true market.’ I really recommend this technique to be used right at the onset. Take a market, then define your objectives: for example, you are after trends; hand pick the starting points of the trends and end points, later add realistic entry and exit points. This not only will give you incredible familiarity with the market, but it will help describe to the algorithms what you consider a good trade – very powerful technique. Knowing the optimum result, you can then proceed to create an automatic (and sub-optimum) trading system. You now have an objective comparison stick and a way to detect where are the failures occurring and why.
The use of committees and of selectors smack right up against the stability and time dependence issue. Remember that markets change and change a lot. Fewer parameters are preferable. However, with lots of care and knowledge the complexity can be tamed and correctly switch systems.
IMHO markets are manipulated all the time. Sometimes macroeconomics drive the markets in sudden events, other times they are used to explain market manipulation. This happens in all time frames. The reason is that financial instruments are a commodity: some are sellers other are buyers. Whoever you are, if you are big, you have to have an inventory to sell/buy. So the markets cannot go without taking a breadth in one direction only because the big guys run out of inventory. It is that simple. I am going to refrain from posting my predictions on sure bets on the market in case someone wants to prove me wrong. But on a weekly basis it is obvious where the DJI is headed. You can also guess short term and long term turning points. You can be sure about certain times and certain companies. For example: my favorite company is obviously headed to above $300. So knowing this, people will let it cross to $290 and slam it. It will go back up if it does not coincide with our second dip at 9600 (oops.) if we don’t break that support and turn the market trend down (it is negative for the year), expect commodity prices, especially on the food stuff to skyrocket soon. Our new financial situation, for the students of ’29, has all the ingredients of a rhyming period, down to the environmental disasters. However, that is another topic for another thread.

HOW DO YOU START A PROFITABLE TRADING BUSINESS? Read more NOW >>>

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!