I am an experienced programmer that has been a long time fan of the markets. For a long time now, I have been working on a short term market trend identification tool (using esignal) and i am finally pretty satisfied with the results. I feel that the tool can be very useful for any trading but recently I have been looking at using it to trade the ES. But, I am not quite sure how to best utilize the tool (being a novice trader, I am severely lacking in strategy experience). If I always a fixed profit target (too simple?), it looks like 3 ticks has the best success percentage and 6 ticks has the best payback. I can't seem to run a backtest on it since it uses a "dagwood sandwich" of multiple instruments and timeframes to compute "trend". I havent seen a backtesting tool that would handle this many variables simultaneously. If anyone has any suggestions, I'd love to hear it. I have enclosed a screen shot from fridays trading with the trend alerts timestamped on the screen. If there is interest, I will post some more trades (even crappy days). Yes, this friday the tool worked extraordinarily well but even on a bad day, it's still pretty good (IMHO). I also have created and tested tools for auto trade execution, monitoring and exiting the trade so I'm hoping to sidestep some of trading's emotional issues. No, this is not a cut and paste job to stir up controversy. I was just hoping to get some feedback, good or bad.
Don't know if this is helpful, an open source trading simulator: http://www.moneytec.com/forums/showthread.php?t=18109
Wallace, thanks for the suggestion. I looked at that link but it does not have the ability to do what I need. I thought I'd try and post multiple images for the day, that last one was a resolution disaster.
I can't understand why you are getting short/long at those points, a lot of lines on the graph. Also when you are short and then get a long signal do you buy twice as many cotracts? (get out of short and establish long) Where is your stop?
5yrtrader, sorry for the confusing horizontal lines. They are a 1/2 baked thing I'm looking at concurrently (real time fib type level calculation) and I couldnt remove them from the image. I also realized I have created additional confusion by talking about "strategy" while trying to demonstrate a "tool". When I mentioned success percentages and paybacks I was using sort of a basic strategy: one contract, no stop loss, exiting when hitting the 3-6 tick fixed target or reverse positions (yes, 2 contracts to reverse) on an opposing signal. I normally would display "up" and "down" on the chart rather than "short" and "long" but I was getting in the spirit of using the tool to trade the ES. Also, if trying to see the "why" of a signal, any one chart might look a little off because the tool is calculating in "thin air" using input from multiple charts to decide up or down trend. The tool barks out a signal and I have a process that grabs the barked signal and incorporates it into the current chart (ES) noting the time and the chart's current bid for down signals and ask for up signals. The tip of the vertical lines are trying to point out exactly where the signal occured on the chart (I have offset the lines 1 bar in an attempt to give a clear visual).
How are you planning to execute trades? Mkt? Lmt? This has been the biggest issue that I have seen with fast trading systems. They test/backtest great on paper, but one tick of slippage wreaks havoc on systems with such a small avg. trade profit.....Just my opinion. There are also free simulators that test execution of such systems. www.strategyrunner.com www.efloortrade.com I don't know if SR links with eSignal. Pretty sure EFT does but you'd have to check with them.
One option is to program your strategy so that that it writes market orders to the paper trading function of esignal and forward test using it that. If the result's match your backtests then you should have a pretty goof guide as to whether your strategy is profitable. Runningbear
ed_berk, Yes, if I would slip two ticks per round trip, the whole thing takes a serious hit. That is why I'm counting on limit orders and my automatic execution setup to minimize this kind of event. I have been testing with limit orders at the bid for shorts and the ask for longs. This is because I use dynaorder's papertrader. As I remember, if you placed a market order with dynaorder, it always seems to slip one tick, even if the market never actually goes there. I feel pretty good about this setup for two reasons: 1) the lag between signal and order placement on my system is usually under a second, 2) Even in a quickly moving market, most times there is a 2 tick retrace in the ES (though I would never count on this to improve my entry). But I guess anything (bad) can happen. Runningbear, I have tried esignals paper trader in the past and found it unreliable (it seemed to improve my entries too often, usually filling my order at the closing price rather than the bid or ask) and not fully functional (could not do OCA orders). Since you mention it, I will revisit it since I haven't tried it once I switched to Dynaorder 2 years ago. Thanks to everyone for feedback.