Dottom, Interesting system. I've done this a few times and it's fun. There are many ways to increase the success rate of your system (e.g., confirmation) and a similar number of ways to increase its profitability (e.g., better targets and stop-loss exits.) You may have tried them all already - or maybe not - are you an experienced system developer? There are those who say that the only way to increase your confidence and discipline is to go through a psychological workout, which may be useful but not the complete answer, imo. Others insist that the "only" way is to develop your own system or method. Oh well...
With all do respect, I disagree with you on the 50% hit-rate. This would assume an entirely random system -- and the market is not random. It trends, oscillates, staggers, etc -- noise may appear random, but at some time-frame you will find a trend or S/R level for stagnation periods. I believe any system falls under two main variables: 1) Win / Loss ratio 2) Profits per Winner / Losses per loser ratio You could easily come up with a system that is successful 90% of the time but with a very low ratio for #2. Likewise, you could also reverse this and have a large number of losers with very large winners. I think it is just a preference for each trader as to what category they fall under with their system. While using my simulated account, I have a very large win to loss ratio (trades) but I also have a +5 stop-loss. So eventually, when I get stopped out at that level, its going to drag down my #2 ratio. I haven't been trading the e-minis for along time, but I think over a three week period, a +35% gain (don't have the numbers right in front of me) is not a "market fluke" with my style of trading. Only time can be the true judge of that, though. However, I definately can say it is very easy to find a system with a very large percentage of winners. Now, the trick would be to design a system that had a large number of winners while knowing which winners to ride and which to close out quickly. aphie
I wouldn't, I was just making a point that aspiring e-mini scalpers need to realize the built-in disadvantage of bid/ask spread. The larger your profit larger/stop loss the less of an impact the bid/ask spread & commissions has on your trading. I also tried on 2m and 3m, but had to move up to 5m bars to where my average profit in backtesting was large enough to overcome inherent disadvantage in bid/ask. Yes, I use IB. I have a custom VB application that enters oco orders for me. I use Wealth-Lab to generate the trade signals one bar ahead of time. I've seen various VB apps that run on top of TWS, including one that integrates with Wealth-Lab. But I still like to put each order in myself. 5m is the shortest time frame I trade off currently. If I was able to push the Profit Factor on my scalping system, or % win rate I may consider an automated system, but it's darn hard to overcome the inherent bid/ask disadvantage + commissions from a pure scalping perspective. If I was going for > 2 pts profit target then I don't consider it scalping anymore.
I agree, %win has nothing to do with profitability of a system, but has a lot to do with the resulting tolerance of the user. Heck, you can get a >90% win system quite easily. But few can handle the big hit when you lose that 10% of the time? My 65% win-rate was based on fixed profit target of 1.50 in ES, fixed stop-loss of 1.00. No trailing or variable stops or profit targets b/c I built it as a test of a pure scalping system. I look at money-flow via various continuation patterns. My point in posting the #'s was to show that a scalping system with a Profit Factor of 2.78, win rate of 65%, and a very smooth equity-curve can be barely profitable in real-time trading when considering bid/ask spread + commissions. It was just a reality check for those serious looking at scalping e-minis. btw, I personally prefer to look at the smoothness of the equity curve. I'll give up net profits for smoothness anytime.
Dottom, So what? (I may be missing the point in your last statement.) Call it whatever you want, the question is, does it make money? Actually, in my experience, it is often much harder to push the profitable trades from 65% to 70% or 75% than to push the profit size from 2 pts to 2.5 or 3 pts. Especially if you have the flexibility to trade 2 contract or 3 contract lots.
Aphie, I hope you have great success when you begin to trade real money. Reading your posts sounds like I'm playing back my own thoughts when I first started trading the e-mini. You are doing all you really can do right now on paper, but just be sure to plan on having results much worse than you expect. When its real, suddenly the bid and ask start jumping all over the place and you learn the true psychological meaning of the term "head-fake". I've experiemented for a long time to find a way to scalp the minis. One day I was doing the math (as I'm sure many of you have) and got the "brilliant" idea that all I needed to do was generate one point per day and I could write my own check for any amount I wanted. 10 contracts for $100k, 20 for $200k, who knows, I could do 100 contracts and hit the magical 7 figures for the year...and still take a couple of months off for vacation. What a plan. One measly point. So I started looking at the trading action during the day. One thing I noticed was how rarely the market would open at either the high or low of the day and then trade in one direction without going at least 1 point in the opposite direction. So my idea for a system was to buy or sell (the choice really didn't matter) the open and immediately put in an order to sell for a point profit (actually 1.25 to cover costs), or, for the sake of the test, on the close if my point didn't get hit. Well, it wins about 93% of the time, both long and short. But as you can quickly see, the 7% of the time that it does open on the high or low kills you with huge losses. The system would range from large losses to barely break-even depending on the market environment during the test period. So then I started thinking, maybe I could filter it somehow, like only buy if we're up in premarket and vice versa. That seemed to have no affect on the results. Or use a stop loss before the losers get too big. So I analysed the maximum adverse excursions and all the other stuff you're taught when designing a system. I found that as I tightened the stop, it actually lost more money, because the accuracy came down and I was still only getting a point on the winners. The next step was to raise the target and "scalp" for more. By the time I reached a target of about 10 points I started to see profitable results in some conditions. But wait a minute...that's not scalping! Suddenly, the scalp system I was looking for turned into a trend following system and I already had one of those. My simple dream of "one measly point" greatness was shattered. Maybe there is a way to scalp for a living, but I certainly haven't found it. The allure is always there though, and I'm sure I will continue to test ideas as they come along. My only suggestion for those who think it can be done profitably over the long term is to try it. Maybe you'll be one of the chosen few. For now, I will stick with the longer term stuff.
My scalping method does very well backtesting, barely profitable in real-life. The profit target and stop-losses are too small relative to the impact of bid/ask difference & commissions. Also, if I was able to more accurate simulate bid/ask in back-testing I'm sure my back-testing result will better reflect RL performance (i.e. not that great). Conclusion- go for larger average profit per trade so as to minimize effect of bid/ask disadvantage & commissions. (i.e. scalping eminis is TOUGH) Agreed. I look at both profit-targets and trailing stops, obviously depending on the method being used. I've heard a lot of people say to trade multiple contracts. Perhaps take small profits on one, bring the second contract to break-even stop, and let the third ride via a trailing stop-loss. Or some other similar variation. I've always told people that they need to test each methodology separate. In the example I just described you are really trading 3 separate exit methods. Lots of traders do the usual 2 contract approach- take small profits on one and move the second to break-even and "let it ride". Again, this is two separate exit signals and you need to test each one. Each one has it's own separate risk profile. Having qualified that, I do agree with you. In fact, the ideal approach is to trade two systems with the same or similar entry but separate exits. Assuming they are both profitable, you will find that in many cases the resulting equity curve growth is smoother with trading one contract on each method, than trading two contracts using any one method. Another example is to trade one method that takes breakouts and another that fades them. Again, assuming that both methods are themselves independently profitable, you may find that the combined risk profile is more appealing. Not to mention the psychological benefits, but I don't want to get into that.
Dottom, I agree with both statements. However, before someone throws in the towel when confronted with #1 above, I would suggest they try to move the system to a different timeframe, even market, try to grow the profits without losing too much of the other good statistics. I often trade a system that barely makes money in real trading of a single contract, similar to yours, but which gets much better results when trading 3-contract lots, taking the second contract out at a higher target level and trailing the third. I agree with several other traders that smart entries are a good beginning, but smart exits are even better for overall profitability, and money management considerations (like looking at trading multiple contracts and how to phase them in or out) is the most powerful tool to make a good system better.
As we are talking about this, the thought emerges in my head that we may need to start a different thread focusing on system development and improvement techniques. It may be very interesting and useful to many of us. But, I am not the type to organize things...
If I could more accurately simulate the effect of bid/ask spread on backtesting my scalping method, the resulting performance statistics would not be anywhere near what I showed. That was the point of sharing the stats, but I do appreciate the advice. I mentioned the 'one tick rule' which is what I've bulit into all my backtesting. I charge myself one extra tick for every trade to simulate bid/ask. If the statistics look good, then I charge myself another tick for slippage (for eminis- other instruments maybe one tick is not enough). If it's still profitable after that then I proceed from there. I wanted others looking at paper trading or backtesting emini scalping systems to apply the 'one tick rule' to their results to see if it went from profitable to not-so-good.