Last night I had some serious talk to a systems programmer who works with neural networks for >15years. He did research on the markets for the first time in his life over the last 6 months and I have to admit he came up with great ideas & results. He presented me this phenomenon: The trading behaviour of the S&P 500, (when analyzed with his algorithms) from 1960 - today shows some significant change in 1991 - 1993. Trading before these years was different from today's. The system seems very revolutionary and solid to me. Has anyone some idea, what could have triggered this change during this time ? It appears it is NOT related to the large -scale introduction of information technology into trading. A more likely possibility is, it seems to have something to do with options & derivatives. Anyone remembering what happened then ?
I have found this to be correct also from the backtesting I have done with different mechanical systems I employ. It seems to me there was a change in the operation and style of program trading and dynamic hedging employed. You need to have a tape reading tool with historical data plugged in to correlate the changes. It seems that whatever was employed in this period has only been magnified since, which has helped some of the trading I do.
Possible. Do you have any information how program trading changed the IBanks methods during this time ? However, I still have doubts and think it might be correlated to something completely different. Another thing I noted when tampering with the algo: It seems the phenomenon occurred as early as 1975. But it was so miniscule then, we didn't notice first. It became more powerful after the 87 crash and then developed fully 90-93. 100% certain: It did not at all exist before 1975. Also, we "found out" that the '99&'00 bubble was by far the most significant anomaly (both in size and duration) in the S&P over the last 60 years. What a surprise, lol
There's an obvious conclusion. In the first period of your study (1960-1991) the market did little. And the little it did do (the rally from 83-87) was largely filled in by the crash of '87. From 1991-2000 SPX went higher by what? 6x? And with not much volatility either, except for a few 10% corrections that were quickly checked. If you can get the data you need (it'll be easier to find Dow Jones data) I'd look at 1923-1937 and you'll see a similarity to the profile of the past decades SPX trade. It's even more startling if you compare the 20's-30's DJIA to NDX 91-06. Index futures were introduced in 1982 and OEX options around the same time (1983?). Both product lines were doing mega volume by 1986 so I doubt the "answer" lies there. The "daytrade" is pretty much the same. Back in the 80's when the S&P was trading on a 200 handle, ranges were generally a couple of points or around 1% of index value. Today ranges are still about 1%.
Hello Pabst, I attached a log chart SPX from 1950-today, actually the price action was pretty constant over all the period. Compared to the 70's bear market, the 87 and 89 crashes are completely a non-event, especially because both were followed by a large up-move soon over the following years. I may not make any further statements regarding the algorithm.
Another chart SPX 1950-today, the gray line is normalized, removing the S&P "yield" of about 8%/year from the chart.
Bitstream, what do you mean. All charts show the SPX 1950-today (sorry for removed scales), the blue line is an abolute candle (useless on that time scale, only for comparison purposes), the white lines are log(price) and the gray line on the second chart is a normalized log(price)