But i gave it a try, and here is the model with a new variable: I inserted a PreWar Variable, very much like the GulfWar, and the time frames for this variable is SET PREWAR = %iF(T>=1990:08:10.AND.T<=1991:01:25, 1, 0) SET PREWAR = %iF(T>=2002:05:03.AND.T<=2003:03:14, 1, 0) The variable is a psychological variable, and really only affects P/E, and after the war starts is turned off. The estimate of the impact is estimated at 3 P/E points, or 138 SP points undervalued. . . again, this is linear math, non linear is a bit more difficult. . . and one can alternately use the price of oil as the proxy anyway. . . . . . this is how the forecast looks with the new variable, and the data is cut off as of February 14th, as the model is based upon earnings based upon that point. . . however, i added the prewar vaiable as a known event cutoff. . . . good luck. . . and remember, forecasting a point will always be wrong, but the fair value estimate is what the model tries to estimate, and not the fair value in the typical sense, but the fair value as to the economic and technical factors that effect the underlying earnings and the estimated growth rate as reflected in the P/E ratio. The model is linear in the fact that the P/E ratio stay within a given range, approximately 10 - 30, but what causes the larger moves today versus 10 years ago is that the underlying earnings is 3 times higher, $45 versus $15, so one P/E move is a much larger move today than before, and the analysis to say that this will be permanent is based upon the earnings not collapsing to the lows of the past. . . just like japan's volatility is now so much more muted do to such low earnings. . . . sportguy