I am looking for some parameters to filter intraday signals. Anyone can suggest me some that work? Thank you. (E.g.: moving averages, RSI, range...).
Hey Travis. I was asking a similar question in this thread---> http://elitetrader.com/vb/showthread.php?s=&threadid=168540 Nitro made one suggestion and it greatly improved the #s. I have tons of other ideas for intraday strats but not enough time to test them all. I would love to get a discussion started as well. Gotta run for now. Jim
Something that needs to be understood is that "filters" work against you just as much as they work for you. All too often people believe that they have found a "good" trading model, but they think they need additional "filters" to take out some of the bad trades. This is flawed logic, as there really is not such a thing as a "filter." There are some ideas which make sense logically, but I would hesitate to call them filters. A simple logical idea is taking long signals when price is above VWAP and short signals only when price is trading below VWAP. I would not call this a filter. Any form of technical information such as MACD, Stochastics, RSI and things of that nature should not be considered "filters." These are technical indicators which work until they don't. They will not help you reduce noise with in your model. They may even take out more good trades than bad trades. Short answer is be careful of "filters." You either have a model that works or not. No amount of "filters" will change that.
Some good advice above. "Filters" on my systems are around Certain days, certain news announcements (announcement effect), certain hours. Either, risk profile adjusted or systems shut-off. Group those intraday trades - win %, expectancy - into excel sorted by each hour block of the day. How many trades placed or pressed during each hour - 7 trades during 11:00 block with poor expectancy?
For many types of intraday systems you can improved your average trade by only taking longs when the price > 50-dayMA. Converse for shorts Alternatively, size differently depending on relation to the 50 day MA.
Good call - time zones. That's something I use as well. Moving averages. Yes, I am going to try that. Thank you both.
Walking a very fine line between filtering and over optimizing, but seasonality, day of the week and COT reports are others.
Travis. It's a visual thing. If it ain't going up. Don't buy it. If it ain't going down. Don't sell it. That's your starting conditions. You can build from there. Keep it damn simple.
So here's what my idea was. I am going to tell you, because someone might be interested. Roughly speaking, a good and consistent system I found is to simply get an average and buy against it (if you're below the average go LONG and viceversa), for example on the EUR. It works year after year, on most markets. It's the principle of the zig-zag. Markets don't go somewhere straight but by zigzagging, wherever they go. Now. This works great on overnight (with great drawdown), daily trades I mean. The problem is that I want to make it work at an intraday level. It works even better and reduces the drawdown, and that's the advantage. The problem with the intraday is that while it consistently makes 55% of wins, all profits are spent in commissions and especially in spread (of 1 tick on the EUR). If I could find out a method to filter some of the bad trades out, I would get closer to what Heypa recommends - only make trades that are profitable. I think my next step will be to look for a method to identify phases of range (multi-day range), which might help this system.