It seems that most robust trading systems have a larger maximum drawdown than average profit per year. Of course, there are systems with small drawdown and large profits but they are often they curve fitted. Has anyone an idea how to get a better relation between those numbers for robust systems?
Gain/Pain Relation
- Thread starter tbmuc
- Start date
-
- Tags
- gain gain/pain relation pain
- Joined
- 22 November 2010
- Posts
- 3,661
- Reactions
- 11
Learn coding/programming and add additional robustness to said trading systems.
Coding is not the problem. I have coded a lot of systems. Most of them ended with a poor Gain/Pain relation. Other ideas found in the net with better numbers seemed to be curve fitted and led also to poor Gain/Pain relation after reoptimization with seperate in-sample/out-of-sample data.
skc
Goldmember
- Joined
- 12 August 2008
- Posts
- 8,277
- Reactions
- 329
I don't know the statistics between max drawdown vs average profit per year, but I am surprised that you find max DD to be greater than avg profit.
I guess this can happen if you are doing, say long only directional trades and run into a bear market and didn't have a overall market filter or something like that. So trading both directions would probably be helpful on that front.
In systems that have a relatively high trading frequency (may be > 300-400 trades per year), I would be surprised if you find max DD > average profit per year. If so then I'd double check the actual edge of such a system.
Thank you very much for your useful hints. The aim of my question is to find characteristics of systems with a better Profit/DD relation.
What kind of systems do you mean. Do you mean systems in lower timeframes than EOD or mean reversal sytems.
Mean reversal systems seem to have a much smoother equity curve than trend following systems, but mostly the MAE is much higher than the profit, preventing the use of stopps. This may be dangerous in market anomalies.
What kind of systems do you mean. Do you mean systems in lower timeframes than EOD or mean reversal sytems.
Mean reversal systems seem to have a much smoother equity curve than trend following systems, but mostly the MAE is much higher than the profit, preventing the use of stopps. This may be dangerous in market anomalies.
skc
Goldmember
- Joined
- 12 August 2008
- Posts
- 8,277
- Reactions
- 329
My comment applies to both trend following or mean reversal systems. Assuming you are able to find enough entry trigers, to achieve a high trade frequency you can either trade lower timeframes or trade more uncorrelated markets. Say if you just trade long Australian equities your trade frequencies will be largely constrained by both your capital available and portfolio heat... and it's easy to see that there'd be large drawdown on a market correction. But if you trade multiple futures market both long and short, you can potentially hold more positions as you would be less concerned with portfolio heat.
Mean reversal systems don't necessarily prevent the use of stops. It may prevent the use of tight stops, and it demands even more attention in position sizing. But stops can be used under the same principles as any other trading systems... i.e. get you out of a position when you are wrong to prevent further loss. And like a trend following system, the stop needs to be wide enough to avoid being stopped out within normal fluctuations/MAEs. Yes occasionally a market abnormality might gap through your stop and cause a lot of pain.. but you should have some hint of how much risk is involved and position-size accordingly.
Thanks again for your hints. If I understand you right, combining uncorrelated futures are the way to get high trade frequencies and thereby lower drawdowns. How do do that, if you don't have a 6-digit account? I tried metatader-brokers, but it the plattform seems ideal for manipulation.
Related to mean reversal: I tried a few of Larry Conners systems, they often have a MAE of 30-50% with a profit of 2-5% per trade. More worse, they enter trades if an unlikely intraday limit is reached. You put up to 50 limits in the market and under normal circumsdances only 1-2 are reached. But that may be very dangerous under abnormal market conditions, when a lot of a entries may be triggered. Do you know more tradeable mean reversal systems?
Related to mean reversal: I tried a few of Larry Conners systems, they often have a MAE of 30-50% with a profit of 2-5% per trade. More worse, they enter trades if an unlikely intraday limit is reached. You put up to 50 limits in the market and under normal circumsdances only 1-2 are reached. But that may be very dangerous under abnormal market conditions, when a lot of a entries may be triggered. Do you know more tradeable mean reversal systems?
skc
Goldmember
- Joined
- 12 August 2008
- Posts
- 8,277
- Reactions
- 329
If you don't have the capital to trade more markets then the only other way to increase frequency is to reduce timeframe. However, you certainly don't need 6 figures to trade multiple futures, considering the required margin is usually a few $k.
Don't know about Larry Conners systems, but a MAE of 30-50% against a 2-5% profit sounds like a recipe for disaster. Mean reversal strategies typically involve various spreads and pairs... here's a thread on equity pairs for example. https://www.aussiestockforums.com/forums/showthread.php?t=14508
The market is a risk asset. By definition if you want the gain you have to put up with some pain sooner or later. Probably best to work out how you are going to deal with it and work on making your gains more than your drawdown's in the long term.
The question was related to the account size, if it is realtistic to run EOD future systems with a 1-2% risk per trade and an account less than 100'000$.
skc
Goldmember
- Joined
- 12 August 2008
- Posts
- 8,277
- Reactions
- 329
What makes you think that way?
Let's say you have $50k, risking $500 per trade. Let's say you trade 1 contract per market, with average margin of $5k (correct me if I am way off with overnight margin) and average holding times of 5 days. So the maximum you can hold is ~8 markets at a time, and you'd have ~400 trades a year. As long as the 8 markets you trade are not highly correlated, it "should" give you an equity curve without massive drawdown.
I am not an expert in future systems, but it appears doable on paper. Someone more versed in futures might be able to offer better advice.
Trembling Hand
Can be found on the bid
- Joined
- 10 June 2007
- Posts
- 8,852
- Reactions
- 205
Probably need to double capital due to the large contract sizes. And outside of US Index futs the ON margin is up around 10-15g.
But I agree. Positive expectancy + non-correlated markets + reasonable trade frequency = Smooooth equity curve
Yes, you could trade end of day on SPI, sp500, oil and many others with an account of that size with risk per trade limited to 2%.
