Iâve worked with the notion that every system has periods of good performance and bad performance, and that finding consistently profitable, net-positive edges is a very hard problem. So I took a completely different approach... paper trading and monitoring a population of 6 to 50 unique daytrading subsystems, each simple in design, but with somewhat unstable profitability, and all net-negative over the long haul, and find a way to forecast near-term profitability for each subsystem. This seems very robust to market changes, because each subsystem plays certain opportunities, days, or particular patterns. When none of the subsystems achieve a high enough forecast, the top-level system stays neutral. If not enough subsystems are profitable or the system is neutral for too long, I just eliminate the worst subsystems and search for new ones. Yes the search is in hindsight. But done over a population of systems it becomes robust when walking forward. I agree completely that systems are the way of the future. I think systems and traders today are having to anticipate what other quantitative systems will do in the market. Itâs becoming one big quantitative game.
Thanks for the reference, will read that thread. However, I don't think anyone has really weighed in on my earlier thoughts regarding t&s and large lot purchases at the market. Does this make a difference to any other "tape readers?"
I looked at an esignal chart from Oct 8 and can't locate your trades on the chart. IB graphical execution report you posted shows local time. Even after going hour by hour I can't find any time that makes it all fit. When did these trades take place? I'm having trouble matching them up with anything. Even if you were in Australia and the chart was from the trading day of Oct 7, I still can't make the trades fit.
Hi Grob109, I compared your fills to a one minute chart of ES for 10-8-03 and I could not match your fills at all. What time zone are you trading in? Thanx, Backwardation
This website helps alot with times zone and dates. http://www.timeanddate.com/ Depending on where grob is, he was either trading the trading day of Oct 8 or Oct 7. Neither one fits.
The AZ location is what I was guessing and Oct 7 at the end of day was a close fit, but some of the bars don't match. For example at 15:08 (central daylight savings time) shows a fill at 103650. Not even close. The :08 bar for Oct 7 had a low of 103775. Since you said it was a demo, I'm supposing the data was messed up.