Research

Financial Neural Networks


Global Tactical Asset Allocation


General Analysis




Neural Networks and Stock Market Timing
  • The most frequent criticism of technical analysis is that is just doesn’t work. This is  easy to disprove. Let’s look at two major market moves this decade, using the basic  technical analysis tool of simple moving averages (SMA)...These chart findings gave clear evidence of a substantial market decline, and offered time to escape the long side of the market.
  • Another example uses a system developed over 30 years ago, the 4% Swing investment strategy, developed by Ned Davis of Ned Davis Research. The system is easy to explain and quickly disproves the theory that market timing doesn’t work...The system was profitable duringa 19-year back-test period from 1966 to 1985, prior to its publication, and has remainedprofitable since its publication.
  • Timing the market increases profit potential by avoiding the majority of price action opposite in direction to positions held. Mechanical trading systems offer a number of advantages over discretionary systems, including but not limited to, the exclusion of emotion from trading decisions and the ability to extensively back-test potential trading systems. Neural networks with genetic optimization offer a tool that can produce mechanical trading systems that are adaptable to changing market conditions.

Using Neural Networks to Forecast Stock Market Prices
  • The JSE-system demonstrated superior performance and was able to predict market direction, so it refuted the Efficient Market Hypothesis.  The JSE-system was able to predict market movement correctly 92% of the time,
  • Overall, neural networks are able to partially predict share prices, thus refuting the EMH.  
  • Neural networks also consistently outperform statistical and regression techniques.
  • The Tokyo stock trading system outperformed the buy-and-hold strategy and the Tokyo index. The self-organizing model constructed model portfolios with higher returns and less volatility (risk) than the S&P 500 index.
  • Although neural networks are not perfect in their prediction, they outperform all other methods and provide hope that one day we can more fully understand dynamic, chaotic systems such as the stock market.

Neural Network Predictions of Stock Price Fluctuations
  • Overall, it is clear that neural networks applied to individual stocks can yield statistically significant predictions. The networks explored here are some of the most commonly used within the world of neural networks, but appeared to have good results within the stocks explored.

Computational Neural Network for Global Stock Indexes Prediction
  • The results show that a neural network model with improved learning technique is a promising methodology for stock market indexes prediction. Results in this study show that neural network model achieves close prediction in terms of RMSE and MAE metrics.

Why You Simply Must Time The Market
  • There are two types of investors. The first type claims that the financial markets can not be timed; that their movements are random and unpredictable, and that market investment is worthwhile only because the long term trend is up. Their advice is to buy assets and hold them for the long term. With few exceptions, this type of investor has done poorly over the past decade. The second type of investor knows better. They use a variety of tools and techniques to determine buy and sell points, and move from long to cash to short positions.
  • Timing systems can outperform, match, or underperform buy and hold approaches. If one is willing to invest the time and effort to develop or evaluate profitable trading systems, the rewards can be great. In the realm of financial markets, the random walk theory and efficient market hypothesis survive in spite of considerable evidence of their fallacy (Murphy 2004) 
  • If the markets of the past ten years have told us anything, it is that buy­and­hold is a losing proposition 
  • All things being the same, an artificial neural network based trading system using a smaller number of inputs is superior to a similarly performing system using more inputs. 
  • Artificial neural networks can discover relationships between indicators (inputs) and (profitable) trading signals which may be nonlinear and not be obvious to inspection (Fishbein 2005). Genetic algorithms can help optimize systems and do so faster than exhaustive searches. 
  • Contrary to popular opinion, a record of successful live trading is not in of itself sufficient proof of the validity of a system. Such success may be a statistical anomaly, and not representative of the future long term performance of a system. 
  • The combination of artificial neural networks and genetic algorithms offers a unique way to develop powerful trading systems. 
  • Timing the stock market using a mechanical trading system based on artificial neural networks and genetic algorithms provides a statistically significant increase in trading returns over a buy and hold strategy.

Neural Networks, Financial Trading and the Efficient Markets  Hypothesis
  • Our results indicate that on some price series the return achieved is significantly greater than that which can be achieved on the bootstrapped samples. This lends support to the claim that some financial time series are not entirely random, and that—contrary to the predictions of the efficient markets hypothes is a trading strategy based solely on historical price data can be used to achieve returns better than those achieved using a buy-and-hold strategy.
  • We found that networks trained using a large number  of input variables can usually be reduced to much smaller networks that only use two variables

Improving returns on stock investment through neural network selection
  • The generalization ability of the ANN is also extended to commodity trading (copper). Robles and Naylor (1996) are able to show that the ANN outperforms the traditional Weighted Moving Average rule and a “buy and hold” strategy.
  • In equity, Gencay and Stengos (1996) show that the ANN outperforms the garch (linear model) in the identical research methodology. The ratios of the average MSPE of the testing model and benchmark are 0.961 and 1 for the ANN and the garch, respectively.
  • Burgess and Refenes (1996) prove that, the ANN has better generalization ability than Ordinary Least Square in predicting daily return of the ftse index.
  • Testing results show that the ANN is able to “beat” the market overtime, as shown by positive compounded excess returns achieved consistently throughout all architectures and training cycles. This implies that the ANN can generalize relationships overtime. Even at individual quarters’ level, the relationships between the inputs and the output established by the training process is proven successful by “beating” the market in 6 out of 8 possible quarters which is a reasonable 75%..
  • The generalization ability of the ANN is again evident in the Moving Window Stock Selection System as it outperformed the Testing portfolio in 9 out of 13 testing quarters (69.23%).

Financial time series forecasting with machine learning techniques: A survey
  • In regards to the employed machine learning technique, there seems to be a trend to use existing artificial neural network models which are enhanced with new training algorithms or combined with emerging technologies into hybrid systems. This finding indicates that neural network based technologies are accepted and suitable in the domain of stock index forecasting.
  • Lagged index data and derived technical indicators have been identified as the most popular input parameters in the literature.
  • In summary, there seems to be a consensus between researchers stressing the importance of stock index forecasting and that the reported results are predominantly positive. Artificial Neural Networks (ANNs) have been identified as the dominant machine learning technique in this area.

Stock Market Index Forecasting by Neural Networks Models and Nonlinear Multiple Regression Modeling: Study of Iran's Capital Market
  • This research examines and analyzes the use of neural networks as a forecasting tool.Specifically a neural network's ability to predict future trends of Stock Market Indexs is tested....While only briefly discussing neural network theory, this research determines the feasibility and practicality of using neural networks as a forecasting tool for the individual investor.
  •  This research validates the work of perceptron and describes the development of a neural network that achieved a 95.8 percent probability of predicting a market Index rise, and an 83.5 percent probability of predicting a market drop in the TEPIX.
  • It was concluded that neural networks do have the capability to forecast financial markets and, if properly trained, the individual investor could benefit from the use of this forecasting tool
  • It is clear from the model accuracy probabilities that neural networks can accurately predict financial markets if given the proper data upon which to train. When compared to regression analysis,neural networks are a better tool

Global Tactical Cross-Asset Allocation: Applying Value and Momentum Across Asset Classes
  • Momentum and value strategies applied to GTAA across twelve asset classes deliver statistically and economically significant abnormal returns.
  • Performance is stable over time, also present in an out-of-sample period and sufficiently high to overcome transaction costs in practice.
  • We argue that financial markets may be macro inefficient due to insufficient ‘smart money’ being available to arbitrage mispricing effects away.
  • Value and momentum effects are not only present within specific asset classes, but also transcend across entire asset classes.

Gaining The Benefits Of Global Tactical Asset Allocation In a Hostile Environment
  • Global tactical asset allocation (GTAA) is an investment approach that seeks to exploit short-term market inefficiencies to generate uncorrelated absolute returns by taking positions in various asset classes, regions, styles and currencies. Due to its low correlation with traditional asset classes, GTAA can diversify an existing portfolio without making material changes to the portfolio’s overall risk characteristics.
  • GTAA allows for tactical shifts in the effective asset class exposures of the total portfolio without causing large capital movements. This limits any market impact of the tactical moves and also reduces transaction costs.
  • GTAA enables a true separation between GTAA the asset allocation decision and the security selection decision.
  • GTAA requires only a limited cash allocation to affect significant exposures for the portfolio.
  • GTAA utilizes liquid exchange-traded futures and similar instruments, it permits a further reduction in transaction costs compared to buying and selling securities in the physical or cash markets.
  • GTAA overcomes the weakness of traditional asset allocation by greatly expanding the investment universe:
  • GTAA overcomes the weakness of traditional asset allocation by reducing trading costs
  • GTAA differs from many traditional investment approaches in that it seeks to derive outperformance from macro or top-down decisions. GTAA managers are not looking for inefficiencies between individual securities within a given market, but rather inefficiencies between whole markets or regions.
  • Most academic studies of market efficiency focus on the internal dynamics of a single, integrated market. These studies often conclude that large liquid markets, such as U.S. equities or bonds, are relatively internally efficient.  As a result, in such markets it can be difficult for active security selection to produce sustainable risk-adjusted out performance. By contrast, many studies have shown that relatively easily observed variables have the ability to predict broad market returns and market anomalies (like momentum or seasonality of returns) are significant and persistent. In other words, inefficiencies do exist in the macro or cross-market environment. Thus, perhaps counter-intuitively, asset classes are often more inefficiently priced than the individual equities and bonds within them.

Diversification: It's All About (Asset) Class
  • Most investors, including investment professionals and industry leaders, do not beat the index of the asset class in which they invest, according to two studies by Brinson, Beebower et al entitled "Determinants of Portfolio Performance" (1986) and "Determinants of Portfolio Performance II: An Update" (1991). This conclusion is also backed up in a third study by Ibbotson and Kaplan entitled "Does Asset Allocation Policy Explain 40%, 90% or 100% of Performance?" (2000).
  • Furthermore, the studies show a high correlation between the returns investors achieve and the underlying asset class performance; for example, a U.S. bond fund or portfolio will generally perform much like the Lehman Aggregate Bond Index, increasing and decreasing in tandem. This shows that, as returns can be expected to mimic their asset class, asset class selection is far more important than both market timing and individual asset selection. Brinson and Beebower concluded that market timing and individual asset selection accounted for only 6% of the variation in returns, with strategy or asset class making up the balance.
  • Many investors do not truly understand effective diversification, often believing they are fully diversified after spreading their investment across large caps, mid or small caps, energy, financial, healthcare, or technology stocks or even investing in emerging markets. In reality, however, they have merely invested in multiple sectors of the equities asset class and are prone to rise and fall with that market.
  • All too often, private investors become bogged down with stock picking and trading - activities that are not only time-consuming, but can be overwhelming. It could be more beneficial - and significantly less resource-intensive - to take a broader view and concentrate on the asset classes. With this macro view, the investor's individual investment decisions are simplified, and they may even be more profitable.

COP Strategy (Class OutPerformance Strategy) - An Investment Policy/ Strategy
  • Studies show that about 90% of a portfolio’s return is due to class selection
  • Since you cannot successfully time the market or select individual stocks, asset allocation should be the major focus of your investment strategy, because it is the only factor... of your risk and return that you can control.
  • Very few, if any, individuals have the skills or resources to select individual stocks, especially, over a broad range of asset classes. Funds are much the better investment choice.
  • Research on [investing] performance suggests that 100% of returns are derived from asset allocation. [as opposed to timing or individual security selection]
  • Almost all independent researchers and investment authors (all of the authors above) will tell you that the average investor (and in fact, other than experts) cannot competently pick stocks or time the market. Their major focus should be on picking an asset allocation and diversification strategy and a strategy to implement the asset allocation in a manner that meets their individual comfort zone with regard to volatility or risk
  • Results of two hypothetical investors--- one who had perfect market timing and one with perfect asset-class selection (using industry sectors) over the periods from 1942 through 1991. Perfect timing outperformed the S&P500 by over 20 times -- wow. Perfect asset-class selection outperformed perfect timing by a factor of 1000 times. 
  • A newer study by Ibbotson and Kaplan, published in the January/February 2000 issue of Financial Analysts Journal, again found that 90% of the return (good or bad) of investment portfolios and funds is determined by the asset class selection.  Only 10% is attributed to selection of investments within an asset class.  This was an up date of a 1999 study by Ibbotson and Kaplan and of famous 1986 and 1991 studies by Brimson, Hood and Beebower.  With this level of acclaimed academic research it is difficult to understand why so many people “play” at investing in a random selection of stocks.  As Bernstein says, “asset allocation should be your major focus”.  
  • In December 1995, Fisher Investments published a research paper entitled “Beating the Market with Style (read “asset classes” for style) Rotation Strategy.  From the Fisher paper: “historically the market shows defined periods of outperformance and under performance for given styles...” [these periods offer] “semi-predictable style rotation.” “..there is a large spread between [the performances] of winning and losing styles.”  Fisher maintains that if one could only stay in the top four and avoid the worst two, of 6 styles studied; one could beat the market by an average of 3% points per year. 
  • A 2002 paper entitled “Multi-style Rotation Strategies” by Ahmed, Penn State University; Lockwood, TCU; and Nanda, T. Rowe Price, came to similar conclusions to mine and Fisher’s as evidenced by their title and sub title of  “The Benefits are Considerable.”
  •  Looking at the Investing in Top 15 and Top 10 lines, one can see that even in the bad markets of 2000, 2001, and 2002, there were always asset classes that could produce positive results. 

Value and Momentum Everywhere
  • Value and momentum ubiquitously generate abnormal returns for individual stocks within several countries, across country equity indices, government bonds, currencies, and commodities.

Value and Momentum Investing, Together at Last
  • The combination portfolio outperformed the momentum version by an average of 1.1 percentage points a year from February 1973 through February 2008, and beat the value version by an average of 6.4 points a year — all while incurring less risk.

 Do Industries Explain Momentum?
  • Industry portfolios exhibit significant momentum, even after controlling for size, book-to-market equity (BE/ME), individual stock momentum, the cross-sectional dispersion in mean returns, and potential microstructure influences
  • Once returns are adjusted for industry effects, momentum profits from individual equities are significantly weaker and, for the most part, are statistically insignificant.
  • Industry momentum strategies are more profitable than individual stock momentum strategies. Industry momentum strategies are robust to various specifications and methodologies, and they appear to be profitable even among the largest, most liquid stocks.
  • Industry momentum appears to be contributing substantially to the profitability of individual stock momentum strategies, and, except for 12-month individual stock momentum, captures these profits almost entirely
  • Unlike individual stock momentum, industry momentum profits remain strong among the largest, most liquid stocks

Stock Market Prices Do Not Follow Random Walks: Evidence from a Simple Specification Test
  • We have rejected the random walk hypothesis for weekly stock market returns by using a simple volatility-based specification test . These rejections cannot be explained completely by infrequent trading or time-varying volat~lities. The patterns of rejections indicate that the stationary mean-reverting models of Shiller and Perron (1985), Summers (1986), Poterba and Summers (1988), and Fama and French (1988) cannot account for the departures of weekly returns from the random walk.
  • Our results do, however, impose restrictions upon the set of plausible economic models for asset pricing; any structural paradigm of rational price formation must now be able to explain this pattern of serial correlation present in weekly data
Subscribe to: Posts (Atom)