Testing the CRISMA trading system: evidence from the UK market

Abstract

A number of recent studies on technical analysis using individual measures such as filter rules, moving averages and trading range break-out have provided a measure of support for their usefulness. The current study tests the multiple-component CRISMA trading system of Pruitt and White (Journal of Portfolio Management, 14, 1988) in a UK context. The system seeks to identify equity trades (and subsequently, call options written on these shares) by using jointly the three technical filters of relative strength, cumulative volume and the relationship between 50-day and 200- day moving averages. The results over the period January 1987 to June 1996 show that the CRISMA system would have been profitable, generating an annualized profit of 19.3%. However, when adjusted for market movements and risk it was unable to predict significant excess returns. Further, the results were not stable over time and trades on larger companies fared better than small. When the signals were used to trade options, CRISMA was able to predict high returns (mean return of 10.2% per trade even in the presence of maximum retail costs) but with only 55% of trades profitable (proportion not statistically significant). Overall, the results are consistent with weak form efficiency in the UK equity market.

Is the CRISMA technical trading system profitable?

“…As mentioned in Goodacre et al (1999), the mean, Ordinary Least Square, and Scholes-Williams adjustment techniques can be biased against the CRISMA system so we pay particular attention to the market-adjusted results. The marketadjusted excess return is positive, however this is not statistically significant so we conclude that there is little evidence of long CRISMA system being profitable for our entire sample.…”

“…More recently, doubt has been cast on the profitability of CRISMA. Goodacre, Bosher, and Dove (1999) find that CRISMA is not profitable on the UK equity market and Goodacre and Kohn-Speyer (2001) find that CRISMA is not profitable on a different sample of US stocks.…”

“…Goodacre et al (1999) apply CRISMA to 254 U.K. stocks from the FTSE 350 Index over the 1987-1996 period. They find that profits of between 6.9% and 19.3% p.a.…”

“…This fact that virtually all their stocks had traded options available is further evidence that these papers are based on large stocks. 2 Goodacre et al (1999) and Goodacre and Kohn-Speyer (2001) also study large stocks due to their focus on the constituents of major share indices. The fact that the size differences do not appear to explain the previous divergent results is further motivation to study CRISMA profitability across stocks of different liquidity to see if this might explain the varying results.…”

“…A binomial proportion test statistic is used to determine whether the CRISMA system predicts more profitable than unprofitable trades at the four levels of commission. Goodacre et al (1999) identify that the mean-adjusted, ordinary least squares market model, and the Scholes Williams (1977) market model are biased against momentum-based trading systems such as CRISMA. These trading systems seek to identify relatively good performance by a security in the expectation that such performance will continue.…”