Tag Archives: financial model

An intelligent member’s view on trading strategy optiziming while building financial models and algorithms

A comment by a member was posted at:


Yes, you probably are. I am not saying optimizing should not be done, but I must admit that I have not been able to succeed at it consistently. It must be done properly, and I still do not know, after years of trying, to be able to tell you I know how to do it properly. I can say that I know quite a lot about it, but I doubt that you would have the patience to let me try to explain to you, by trial and error, how I might be right and how I might be mistaken. One hint of where I am coming from. Most optimists (sic) think a robust parameter set optimization should be insensitive to variance near the optimal points of return.

I hold that the problem of finding the optimum point in parameter space is more akin to finding a watering hole in an African game reserve before the competing animals make the place a mud hole or poachers arrive with their guns and make the place even more dangerous. The longer you wait around to be sure that the watering hole you found is valid, and real (not a mirage) and safe to drink from, the more likely you will be very sorry.

This is easily most intelligent view on it!!

Learn more more by joining my FREE newsletter!

















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!

Youtube opinion on so many Matlab toolboxes but should focus on Simulink or trading strategies for financial models?


Youtube opinion on so many matlab toolboxes but should focus on simulink or trading strategies for financial models?


Please let me kknow via commenting below

Want to learn more as I do? Join my FREE newsletter.



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!

Brief overview of all financial models and quant strategies we use for our QuantLabs.net Premium Membership monthly service

Brief overview of all financial models and quant strategies we use for our QuantLabs.net Premium Membership monthly service

This is a VERY brief description of the models and strategies used for the QuantLabs.net Premium service. Do note that these are technically and mathematically advanced so some terminology may be advanced for some users. I am hoping to post detailed articles on each with some YouTube videos (www.Youtube.com/quantlabs) to accompany these definitions. This will be in the future before the service goes into a monetization phase.

Note each the following models/strategies are applied to each market symbol screened daily by our real time application which broadly sweeps various global market regions for best performing equities. This is usually the top three of each region. For now we track the regions of Denmark, India, Australia, China, USA, Canada, UK, France, Switzerland, Brazil, Hong Kong, and Germany. More may be added as demand requests it.

Equity Invariants

This does simple invariance assuming an independence and identically distribution (i.i.d). Two plots are generated. The first is a histogram to check various variables are independently distributed. A second scatter plot is generated to check the variables are independent under i.i.d. An ellipsoid circle will display to indicate local dispersion.


Compounded Return Estimation Interval

This projects a distribution of compounded returns from an estimation interval of the investment horizon. Distribution of prices at the investment horizon is then computed.


Pairs Trading Combination

This uses classic pair trading techniques against 5 picked (from the same day) equity market symbols. You would hopefully see a profit in the short and long positions by using a mean reversion behaviour. Total payoff and total transaction costs are displayed as well based on certain capital amounts for each position.


Bayesian Jointly Uniform Prior Correlations Implies


This shows how a jointly uniform distribution on correlation implies marginal distribution of each correlation when it peaks around zero. This also plots both univariate marginals and bivariate marginals.


Shrinkage Estimator
This computes the multivariate shrinkage estimator of location and scatter plot data under normal assumptions of any market asset supplied to the model.

Statistics Summary Projection

This projects a summary of various statistics of arbitrary horizons.  This displays of each supplied market symbol:

single-period standardized statistics, central moments,

single-period non-central moments,

single-period standardized statistics,

multi-period cumulants

multi-period non-central moments

multi-period standardized statistics

Equity Projection Pricing


This projects the distribution of market invariants (i.e. compounded returns of a stock) of estimation interval of an investment horizon. The estimation is normally assumed which also computes the distribution of prices at various investment horizon points which is done analytically using Monte Carlo, delta, and duration approximation. Investment horizon points include 1 day, 1 week, 1 month, 1 year, and 2 years.


Equity Quantile Value at Risk (VaR)


This computes a quantile using VaR. The objective uses a Student T distribution and does simulations using the generated sample quantile. This does extreme value theory (EVT) approximation as well. A plot is generated displaying these three methods of exact, simulation, and EVT.


Distribution of Market Invariants


This will project the distribution of market invariants which include stock market compounded returns. This will then estimate the distribution of prices and performs a two step mean variance optimization.  The risk aversion parameter is set at 10. Mean variance inputs are provided analytically and numerically (using mean and covariance). First step of mean variance optimization uses quadratic mean variance optimization which determines one parameter frontier of a quasi optimal solution. The second step evaluates satisfaction for all allocations of the frontier and will pick the best. Two plots are created where one is the mean variance pricing frontier. A second plot displays the satisfaction as a function of standard deviation on the frontier.


Note that more models will be added as demands allows. Also, the focus of the service is on popular FX currency pairs, best performing daily equities, some fixed income assets, and futures.







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!