From what I’ve read, ThinkAI keeps a database of stock-chart data, and as the day progresses it keeps searching the DB for charts that match-up best with the current day’s. Then it assumes that the rest of the day will look similar. So, it can’t actually predict the future, but only try to catch the puny humans blindly repeating history.
The approach sounds unsophisticated, but when you consider that most stocks are traded by the same group of traders, it might be just the right approach. For example, George is there trading BBT every day using the same methods. He’s a big part of why BBT behaves the way it does. And that is supposedly true for many stocks: each has its own dedicated group of traders.
ThinkAI got the close totally wrong today, but during the day, it made excellent calls and helped me scalp a few bucks out of the chop.
Here is a picture of ThinkAI in action (click to enlarge):
ThinkAI was predicting that SPY would move up a bit over the next few minutes, and that’s exactly what SPY did.
GDP came in worse than expected at -5.7 versus the -5.5 estimate and the futures dropped a couple of points, but are still green
The chart below is NYSE breadth plotted hourly. (That’s advancing stocks minus declining stocks.) So, each black dot is the breadth reading at that hour. The red line is a 6.5 period moving average, which gives a daily average, and the blue line is a weekly average. (Click chart to enlarge):
Notice that, for today’s rally, the black dots barley made it above the zero line (light blue), let alone up to the red upper bound. So, breadth was weak today, which is why the rally went flat in the last couple hours of trading.
The blue weekly average peaked on May 4th, so that is a negative also. But now look at the recent lows formed by the red moving average starting with the big puke on May 13th. The next two lows on the 21st and 28th were much less-bad. The two purple lines connect those points and show that there is an uptrend to the down days.
The market is struggling here, but it is not deteriorating yet. The next down day will tell us a lot since we will be able to see if the market is able to sustain the less-bad uptrend. The rising purple trend lines show that things are coming to a boil.
(Note: the NASDAQ breadth chart is substantially the same.)
Could that have been Pequot liquidating Wednesday afternoon? Maybe they were waiting for the market to return to the top of its range before dumping their holdings.
“The funds, with about $2 billion in assets, will return a ‘significant amount’ of cash to investors by June.”
June is only a couple of trading days away. If it was Pequot doing all of the selling, the market could be expected to float back up on Thursday, given a neutral news-flow. So, the bulls have something to pray for.
I’m thinking that today should be a “doji day” like May 5th or 19th where the market makes a doji candlestick on the daily chart. SPY’s slow stochastic is overbought on the 60-minute chart and it will be hard for the market to hold onto gains made today.
This picture shows a TradeStation RadarScreen that is set up for George Method. For any of the symbols in the list, it will generate alerts whenever the MACD crosses zero, when the slow stochastic lines cross, or when the 9ma crosses the 36ma (click to enlarge):
The interval column makes the indicator columns calculate for the 15-minute time frame. After adding the columns, turn on the alerts for the three blue indicator columns, and set the moving averages to 9 and 36.
Over the weekend, the futures spent a good deal of time in the 877 area without being able to sustain a bounce from there. I regard that as bearish, and it looks like the support at that level is being eroded.
George has found a free site for sharing charts: Freestockcharts.com. The site runs on SilverLight, which is Microsoft’s Flash competitor, so you will probably need to install the plugin.
UPDATE: See my new trading book! The General Theory of Day-Trading.
If the SPX can hold above the 875-880 area, it may be able to consolidate the rally gains in a rectangle pattern trading between 880-930 for a while.
However, if the market dives through the 875-880 level on heavy volume, then a double-top pattern comes into play. Subtracting the 930 peak from the 880ish neckline, we get 50 points. So, the target would be 880-50=830.
Perhaps the market won’t do anything so drastic the day before a holiday. But don’t forget that the February dive began on the Tuesday after a three-day weekend with a giant gap down on February 17th.
If the market closes weak like it did on February 13th, then a scary gap down becomes a possibility for Tuesday morning.