1
/
of
1
store-traders-com
V.9:5 (187-188): Testing Indicators by Arthur A. Merrill, C.M.T.
V.9:5 (187-188): Testing Indicators by Arthur A. Merrill, C.M.T.
Regular price
$2.95 USD
Regular price
Sale price
$2.95 USD
Quantity
Couldn't load pickup availability
Testing Indicators
by Arthur A. Merrill, C.M.T.
Analysts consult scores of indicators to get clues for the future. At any given instant, however, some of the indicators are bullish and some are bearish. They are never unanimous. Which should we believe? Which should be given the most weight?
Too often, I suspect, analysts tend to give the most weight to the indicators that happen to agree with the analysts' own convictions. If the analyst is bullish, he tends to believe the bullish indicators; if he is bearish, he likes the bearish indicators. This bias may be totally unconscious.
To avoid this bias, some analysts use coefficients of correlation. The method I devised seems to give me simple, understandable and useful measures of performance. The method gives me batting averages. Here's the method:
First, we accumulate a data file for each indicator, logging in the indicator's forecasts. I ask an indicator, each week in the past 10 years, whether it is bullish, bearish or on the fence. Each week I log in the digit "2" if the forecast in that week was bullish; "1" if bearish and "0" if on the fence.
Analysts consult scores of indicators to get clues for the future. At any given instant, however, some of the indicators are bullish and some are bearish. They are never unanimous. Which should we believe? Which should be given the most weight?
Too often, I suspect, analysts tend to give the most weight to the indicators that happen to agree with the analysts' own convictions. If the analyst is bullish, he tends to believe the bullish indicators; if he is bearish, he likes the bearish indicators. This bias may be totally unconscious.
To avoid this bias, some analysts use coefficients of correlation. The method I devised seems to give me simple, understandable and useful measures of performance. The method gives me batting averages. Here's the method:
First, we accumulate a data file for each indicator, logging in the indicator's forecasts. I ask an indicator, each week in the past 10 years, whether it is bullish, bearish or on the fence. Each week I log in the digit "2" if the forecast in that week was bullish; "1" if bearish and "0" if on the fence.
Share
