The nanos had a fun weekend... did lots of mating and mutating. <ggg>
They are still predicting the market at 6650 for Monday. I did spend quite a bit of time over the weekend looking at their backcasts vs. actuals and a few things became evident. In particular, I've been trying to put together a "trading system" based on the projections which would make sense/profit. The following are some of the conclusions I've drawn to date.
1) They project during a trend (up or down) quite closely... which is what we've seen the past week or so.
2) At reversal points, they overshoot for 1 day. Then they recover almost immediately and predict the trend quite accurately again. To be specific, the variance on a significant reversal day will be on the order of 150 to 200 points (bottom) or -150 to -200 (top).
3) The more they overshoot a reversal point the bigger the overall reversal in terms of magnitude... so overshooting by 200 points ultimately signifies more of a reversal than if they only overshoot 150 points.
4) The difference between where we closed on Friday vs. the prediction became positive, indicating a cross-over between the forecast and actual lines. This is a mandatory sign for a reversal... (ie. a potential overshoot) but not strong confirmation. A big variance on Monday would be a strong signal based on the backcasts. In most instances therefore, the nanos miss significant bottoms and tops by a day but they also tend to weed out false reversals.
5) There appears to be a correlation between the length of a reversal and how quickly the nanos get back onto the trend line. If there is a reversal signal on Monday, and by Tuesday the variance between the forecast and the actual drops low (ie. the nanos get very close to the trend), this would indicate that the reversal will be a few days in duration. In essence, they hyper-react during short term reversals. If there is a reversal and the prediction remains slightly low for an extended period of time (ie. it takes them a few days to catch up to the trend), this would indicate that the reversal is longer lived in terms of duration. (Naturally this is all easier to see in graph form!!)
6) When a MAJOR/extended bottom or top occurs (ie. that last a month or more), the forecasts take on a very distinctive, almost flat appearance which cuts off the tops or bottoms. This is not the case now, which indicates we are simply oscilating within a price channel.
So.. summary... the nanos are forecasting the market down 110 points... will it happen?? Well either it will and the nanos are just accurately predicting on trend.... or it won't in which case the magnitude of the variance will indicate the size of the reversal. (Note - this implies particular importance for stops on Monday). The fact that the forecast and actual lines crossed on Friday sets the market up for a reversal but does not really confirm it. If there is a reversal and they're right back on trend on Tuesday then it will probably top out late this week or early next week. If they stay below the actual and gradually move towards the trend over the course of the week, then we may see an upward trending market for a few weeks (with minor oscillations of course. When they project a day which is 150 to 200 point above the actual market, that would be a sell signal (although they would have missed the top by a day).
Without getting into any numbers, the nanos are quite bullish on a number of individual stocks which would provide secondary confirmation that a reversal is in order.
On a couple of the stocks I'm running them on, there is no lag between actuals and forecasts on reversal points. In one case they actually appear to be able to predict a move upto 4 days in advance on occasion. I'm hoping that as I continue to run the market level nanos that they also develop this capability, which would obviously optimize profits. I am running the market level nanos with a significant amount of back history... which hopefully makes them smarter.. but also takes much longer to go through a generation. So... the market level nanos are only 1/4 as old as a few of the stock level populations which accurately predict the reversals.... The expectation that the market level nanos can predict reversals without a lag seems reasonable once they have run long enough to learn how to do this... I guess time will tell!! In the interim you try and work with what you have. |
