Profitability of alternative methods of combining the signals from technical trading systems

• Date: 01 January 2019
• Published date: 01 January 2019
• DOI: http://doi.org/10.1002/isaf.1442
• Author: B. Wade Brorsen,Jasdeep S. Banga

RESEARCH ARTICLE

Profitability of alternative methods of combining the signals

from technical trading systems

Jasdeep S. Banga

1

|B. Wade Brorsen

Department of Agricultural Economics,

Oklahoma State University, Stillwater, OK,

USA

Correspondence

B. Wade Brorsen, Department of Agricultural

Economics, Oklahoma State University,

Stillwater, OK 74078, USA.

Email: wade.brorsen@okstate.edu

Present Address

1

Jasdeep S. Banga, KeyBank, 127 Public

Square, Cleveland, OH 44114, USA.

Funding information

National Institute for Agriculture (NIFA) Hatch

project, Grant/Award Number: OKL0293;

Oklahoma Agricultural Experiment Station

Summary

Past efforts determining the profitability of technical analysis reached varied conclu-

sions. We test the profitability of a composite prediction that uses buy and sell signals

from technical indicators as inputs. Both machine learning methods, like neural net-

works, and statistical methods, like logistic regression, are used to get predictions.

Inputs are signals from trend‐following and mean‐reversal technical indicators in addi-

tion to the variance of prices. Four representative commodities from agricultural, live-

stock, financial, and foreign exchange futures markets are selected to determine

profitability. Special care is taken to avoid data snooping error. Both neural networks

and statistical methods did not show consistent profitability.

KEYWORDS

data science, futures trading, neural networks, technical analysis

1|INTRODUCTION

Technical analysis involves predicting asset price movements from an

analysis of historical price movements. Beja and Goldman, (1980)

argued that the trends exploited by technical analysis are due to mar-

ket frictions that cause markets to adjust slowly in the absence of

technical trading. The trend‐following systems piggyback upon the

actions of informed traders and work best when the market is unsta-

ble. Reversal systems, like oscillators, should work well when the mar-

ket is stable. This study seeks to use machine learning and statistical

methods as a way of optimally weighting among trend‐following and

reversal systems.

Brorsen and Irwin, (1987) report that among a survey of 32 large

commodity fund managers only two were not using objective techni-

cal analysis. Oberlechner, (2001) surveyed foreign exchange traders

and found that a majority of the foreign exchange traders used some

sort of technical analysis. Taylor and Allen, (1992) found that 90% of

traders in London used technical analysis as a primary or secondary

source of information. Park and Irwin, (2007) found trading strategies

based on technical analysis were profitable in the futures markets until

at least the early 1990s. As more money was devoted to trading based

on technical analysis, its profitability dropped. A large number of

trend‐following technical traders could create market bubbles (Beja

& Goldman, 1980; Etienne, Irwin, & Garcia, 2014). Improved technical

trading systems that could optimally switch back and forth between

trend‐following and reversal systems could increase traders' profits

as well as potentially reduce instability created by trend followers.

The efficient markets hypothesis says that the current price

reflects all available information about the commodity (Malkiel,

1989). In the absence of technical traders, markets have proven to

be slow to adjust due to market frictions such as risk‐averse traders

and behavioural anomalies such as loss aversion. Technical analysts

recognize the trends arising from slow adjustments and exploit them.

Sometimes, even if the trend is random but many investors follow it,

then the subsequent prediction becomes self‐fulfilling, and sometimes

creates a bubble. Boyd and Brorsen, (1991) found a strong relationship

between market volatility and technical trading profits. This relation-

ship could be useful to traders in determining whether to use a

trend‐following or a reversal system.

Various technical trading rules have been used in past research.

Lukac, Brorsen, and Irwin, (1988a and 1988b) used 12 trading systems

approximating the full “universe”of trading systems. They found that

technical trading systems produce statistically significant net returns,

compared with the buy‐and‐hold benchmark strategy from 1978 to

1985. Park and Irwin, (2005, 2010) used over 9000 trading rules from

12 trading systems and found that technical trading strategies were

Received: 13 October 2018 Revised: 28 January 2019 Accepted: 4 February 2019

DOI: 10.1002/isaf.1442

32 © 2019 John Wiley & Sons, Ltd. Intell Sys Acc Fin Mgmt. 2019;26:32–45.wileyonlinelibrary.com/journal/isaf