Tested Period (out-of-sample): from 07/01/96 to 04/17/98
Number of Days Tested: 454

The shown results are compared with a Buy & Hold strategy.

1. Traditionally Computed MACD Indicator

1.1 Graph

1.2 Statistics

2. Five-Day Predicted MACD Indicator Using KnowledgeMiner's Analog Complexing Method

These results were obtained implementing the following daily procedure:

(Blue colored steps are user-driven. Red colored tasks are completely solved by KnowledgeMiner using a special simulation module inside KM).

I. Data Base

i. Updating the data base with the equity's actual close price and share volume

II. Modeling/Prediction using KnowledgeMiner's Analog Complexing Method

i. Searching for several most similar market patterns on the updated data base

ii. Generating several 5-day out-of-sample predictions and combining them to a most likely prediction

 

III. Applying Predictions to the MACD Indicator as a Decision Model

i. Calculating the indicator on historical and predicted price data

ii. Generating a signal on the predicted indicator

IV. Decision

i. Trading if a buy/sell signal was suggested using the next day's close price

 

2.1 Graph

2.2 Statistics

Other tests:

INTEL
NOVELL

Back to:  Report     Specials