NQ futures day trading systems
- Thread starter DT-waw
- Start date
DT,
I would be very careful about trading based on the results of your tests. Unless I'm reading the reports wrong, it looks like some of the systems have under 50 trades for a test period of many years. That is an awfully small sample size. You are testing on multiple markets, which helps to make up for the small size of each individual market. But are you using the same parameters across all markets?
The reports you posted earlier were optimization results, picking the best results from optimizing 2 inputs. The problem is that you could take almost any system with 2 inputs, that works on a timeframe generating only 50 trades, and optimize the 2 inputs to find a good combination. Unfortunately, just like all the disclaimers, "past success is no guarantee of future results".
Anomoly's test avoids the problem of too small a sample, since he generates 1-3 trades per day. His challenge is to make sure the system performs consistently in a variety of market conditions. I have a big stack of systems I developed that worked great in the 6 months prior to when I developed them, but they suck nuts before that.
The reason for that is that you often see a pattern in the market, then go back a few weeks looking for it and see it more (talking about daytrading here). So the pattern looks good and you code it up, only to find out in more extensive testing that the pattern isn't there at all 9 months ago. Which means its likely to disappear 1,2,3,10 months from now. If you knew how long until the pattern would break, you'd be rich
-Jeff
I would be very careful about trading based on the results of your tests. Unless I'm reading the reports wrong, it looks like some of the systems have under 50 trades for a test period of many years. That is an awfully small sample size. You are testing on multiple markets, which helps to make up for the small size of each individual market. But are you using the same parameters across all markets?
The reports you posted earlier were optimization results, picking the best results from optimizing 2 inputs. The problem is that you could take almost any system with 2 inputs, that works on a timeframe generating only 50 trades, and optimize the 2 inputs to find a good combination. Unfortunately, just like all the disclaimers, "past success is no guarantee of future results".
Anomoly's test avoids the problem of too small a sample, since he generates 1-3 trades per day. His challenge is to make sure the system performs consistently in a variety of market conditions. I have a big stack of systems I developed that worked great in the 6 months prior to when I developed them, but they suck nuts before that.
The reason for that is that you often see a pattern in the market, then go back a few weeks looking for it and see it more (talking about daytrading here). So the pattern looks good and you code it up, only to find out in more extensive testing that the pattern isn't there at all 9 months ago. Which means its likely to disappear 1,2,3,10 months from now. If you knew how long until the pattern would break, you'd be rich
-Jeff
"YES, but what have you done for me lately?"
Screen Shot of equity curve from TS 6.0 Strategy Report
I present performance reports once again. Now they're based on data from 01/01/1985 to 22/04/2002 from Yahoo Finance.
System was optimized for each index.
jeffm: now the system is backtested on higher number of trades.
Conclusions:
1) Its hard to beat buy-and-hold strategy ( except Nikkei, currencies, interest rate futures ).
2) Returns depend on parameters combination. Example: The best parameter combination for DAX index was almost the worst for S&P index ( with -44% return )!!!
3) Each market had its own best parameter combination. Its impossible to predict which combination will be the best in the future. You should optimize system for each market, every market is different.
4) Returns depend on which market you choose to trade.
5) Unless you trade on many markets with several parameter combinations, your results will be greatly affected by luck factor.
DT-waw
System was optimized for each index.
jeffm: now the system is backtested on higher number of trades.
Conclusions:
1) Its hard to beat buy-and-hold strategy ( except Nikkei, currencies, interest rate futures ).
2) Returns depend on parameters combination. Example: The best parameter combination for DAX index was almost the worst for S&P index ( with -44% return )!!!
3) Each market had its own best parameter combination. Its impossible to predict which combination will be the best in the future. You should optimize system for each market, every market is different.
4) Returns depend on which market you choose to trade.
5) Unless you trade on many markets with several parameter combinations, your results will be greatly affected by luck factor.
DT-waw
And that was EOD trading.
AnomalyResearch: your Equity Curve is impressive. Can you calculate your system's annual return?
I prefer day trading systems. I know they will have lower Profit Factor ratio due to higher impact of slippage and commissions.
I guess day trading systems have more stable results and can produce higher acct return.
DT-waw
AnomalyResearch: your Equity Curve is impressive. Can you calculate your system's annual return?
I prefer day trading systems. I know they will have lower Profit Factor ratio due to higher impact of slippage and commissions.
I guess day trading systems have more stable results and can produce higher acct return.
DT-waw