Unless this is a 100% mech method, IMO it's best to have a program find the pattern, then recall a chart and a trader manually either place or not place a trade after evaluating market sentiment, for instance 10 min MACD can be still under 0 line, though rising after 3 dip, also daily is already & hourly has now turned bullish. So would you take a short off a 2 min MACD 0 line downwards cross? I think that's how Maestro works with his 100s of monitors, system is partially automated, it's still a human who makes the final decision in the process.
OK, here is my first back test. Data set: S&P 500 E-Mini Future (symbol ES), 5147 5-minute OHLC bars Data set dates: 09/07/2006 to 12/06/2006 (3 months when Dec 2006 contract was the front month contact) Start of trading session: 10:00 EST every business day (original poster specified not trading the first half hour) End of trading session: 15:55 EST every business day Overnight position carry: Not allowed (open positions are closed at the end of the day) Indicator used: MACD(12, 26, 9). That is, 12 5-minute bars for a fast EMA, 26 5-minute bars for a slow EMA, and 9 5-minute bars for an EMA of MACD itself (which is the so called "signal line"). Trading costs assumptions: 0.25 bid/ask spread (usual for ES), 0 slippage (very optimistic), $2.40 commission per side. Long Entry Rule: When MACD is above zero and MACD is above its signal line. Exit from Long Entry: When either MACD is below zero or MACD is below its signal line. Short Entry Rule: When MACD is below zero and MACD is below its signal line. Exit from Short Entry: When either MACD is above zero or MACD is above its signal line. Results: System made 572 trades, for a loss of $6500 per contract over the test period. Below is a chart for one of the better days (although even for that day the system was barely profitable). The circled letters on the charts indicate at which time and price a position was taken: "S" for short, "L" for long, and "F" for flat. Now, I am not sure if the rules that I identified are what the original poster had in mind. You'd also note that instead of the 2-min ER2 chart, I used the 5-minute ES chart. I'll retest the 2-min ER2 tomorrow. Let me know if I missed anything else.
If you trade with as quick a wit as you write, you'll never work another day in your life! I mean, I would have said "try learning how to trade the system before you try coding it" ... I mean d*mn, all these coders we have around here you don't have a clue about trading! But your reply is a lot funnier. JJ
Also I believe where a 'retail' ATS would fail is at profit taking, entry conditions have been suggested, what about exit? R:R? That's why I prefer partial automation, program finds entries and a trader makes final decisions. Makes life a lot easier.
Interesting discussion. Been following this system (MACD zero line cross is not exactly a new idea) and have a couple of thoughts on the idea of backtesting it. 1. Exits are everything in trading, it is possible to make a profit using random entries, if the exit strategy is a good one. 2. The problem with backtesting of course, is that it cannot factor in the traders discretion. Seems to me that an exit rule of a cross of the zero line in the opposite direction will give back way too much profits to make such a system profitable. At the very least, an exit rule which locks in profits on good trades is necessary (trailing stop or maybe cross of a 10 bar EMA ?). another strategy is to partially close the position after x no of points proft. 3. In my opinion, people pay too much attention to entry signal, and not enough to exits. The old question seems pertinent - if you had a trained chimp to choose either your entries or your exits, which would you choose ? The answer is of course, the entries. Then at least you could take a profit and more importantly, limit your losses. Excellent thread this though.
I see a few basic problems with that backtest.... Some of the best entries setup pre-market and trigger right around the open, filtering based on 10:00am is not a valid filter/rule for the system as originally presented. Signals from the 5 minute MACD are not a good representation of the signals from the 2 minute MACD. Your test did not include the 10 minute chart filter. A more accurate test of the original SPM, will use both the 10 minute and 2 minute time frames and include pre-market data for signals near the 9:30am cash market open.
How about putting up a few execution reports showing that you actually trade this method trader28lite? I am sure everyone would be interested in whether you really trade it profitably or not...
You know you've made it when you get your own personal troll... I think I'll keep him in the shed out back