His "crash alert"went off the day before that monster rally. But to his credit, he did say wait for confirmation. To his possible discredit, he did not jump (or at least publish) on that mini rally. A crash alert is like a home smoke detector. It gives many false positives, but can be useful when it gives a true positive. However, if the alarm goes off, and there is a hot meal on the table and you don't eat, you will remain hungry.

Of course are you novice ? Monster rally always occurs on panic . Now I said that it must wait the break of 8200 on my fondamental model which is more deterministic (on the contrary the crashindex is a stochastic model it is fuzzy of course). We didn't break 8200 at that time so what's the problem with you . But after we broke 8200 you saw the result So learn the abc of what a support and resistance is before crying

This is the archive of yahoo message http://groups.yahoo.com/group/harrytrader2000/message/23 "From: harrytrader2000 Date: Tue Dec 31, 2002 9:50 am Subject: Crash alert At yesterday's close we have an alert crash it is the first time : http://perso.wanadoo.fr/harrytrader/crashindex_301202.gif As we put it this must be confirmed by the break of the very important support of 8200 (on hourly scale yesterday's projection was 8193)." So (As you can read I even repeated the precaution twice)

It is at -5.8 at Friday's close (crash alert is normally under -6 ). I will post the chart here later because it is not practical to do so with excel. I gave the break zone on the other post but that comes from my fundamental ("deterministic" but I don't like this ambiguous term so I prefer the term fundamental) model not from the crashindex which is just a stochastic model. Nothing genius in stochastic . Deterministic is much harder to find ... since officially there is none

Oh its fuzzy, harry. As fuzzy as you. In a bear market, any "crash " indicator after a 6 week rally will eventually prove correct. Even a novice could tell you that, if you seek further education.

Yoiu said I'm ridiculous ? Well just look at you in a mirror : BY DEFINITION a STOCHASTIC indicator is FUZZY So if possible you must use a stochastic model with something else. In fact for trading I only use my fundamental model, but people like "simplicity" (in fact they like simplisTICITY which is not the same thing) so I give them the crashindex. That's what I used before I got the other model. The backtest of crashindex was giving 200% when I was using it. Nevertheless I never trust backtest on any STOCHASTIC system with stock market where law of preobability is not stationnary globally where I must trust backtest on my fundamental model but it is harder to backtest since it more ressembles the decoding of genomes rules manipulating many matrix components.

I'd probably pay closer attention to harrytrader if I had the slightest idea what he was talking about --Db

The usual definition of crash is a certain amount of downside drop. This is not my definition. And I have a typology of crashes that is to say there is different kind of crashes. The most classical one is the crash occuring with compression of the time: on my model it means we reached the exact target of the model but within a smaller time period than usual. This is not generally a dangerous crash it's just to frighten people The "real" crash (I mean the dangerous one) is when there is a forcing of a projection. That happens at the super high or at the super low. And that happens because there has also been a forcing on the upside before. When there is a forcing we usually immediatly retest the zone and come back again after a very very long time so that statisticians have difficulty to detect something else that pure phenomenon of persistency of probability that is to say they think that the apparent "trend" is just an appearance and not a true trend but my model can affirm that there is true trend. In practice persistency or not persistency that's the period where public think that this will last forever.