There is one of my problems. Strategy authors want to charge based on strategy objectives. I want them charging based on strategy credibility.
Predicting the stock market is one of the most difficult things in the world to do. Nobel Prizes have been given to economists whose work suggests it can’t be done, as well as to economists who think it can.
Most back-tested systems do much, much better in backtesting than they do going forward.
If there are 100 models that each can do 100% a year in backtesting, I would consider it extraordinary if even one of them would do 75% a year over the next decade in real time. I would be very surprised if even one large hedge fund has done 75% a year for the last decade.
I guess what seems obvious to me is not what seems obvious to you.
Of course, I completly agree.
This strategy was published only 5 weeks ago, but I have trading this systems for more time. Also, 25k E-mini S&P500 portfolio is near to a year life in C2, and it shares 5 system with 30k Futures Portfolio.
Nice marekj, I think is a very good criterion.
I took a glance at your 3 systems and they are all doing well. Congratulations. Certainly, the approach to diversify strategies should reduce volatility and DDs.
But getting 100% a year seems quite optimistic.
Are you using more leverage in your new system? If not, why do you expect the long-term returns for your new system to be so much better than your existing systems, which are doing about 30% annualized according to the Grid? Diversification should reduce volatility, but not usually increase expected returns.
If, like you, I were running a system on C2 and doing more than 100% annualized return (on less than a year’s record)–and I might just be doing that in a private system–I would still not expect long-term future returns of over 100% a year.
I think this is the error: try to predict market behaviour. It would be nice…
But ¿who said you need predict market behaviour to ear money?
You don´t need to know what market will do for earn money, you need statistical advantage, very different concept.
I´m statistician PhD with more than 8 years working with futures. I know perfectly what is overfit and how to employ statistical techniques to avoid these examples than you commented. But, if the goal is to predict market behaviour, be sure than this model will fail.
Statistically, real account results are not significantly different from backtesting to date. When this will happend, if happends, we will be able to speak about obviousness. So far there aren´t statistical evidences about any problem with the systems, and it is all we can say since a statistical point of view.
I have doing trading for more than 8 years and also manage an investment club for 3 years ago. This strategy is new in C2, but i´m not new in markets :).
Really, getting 100% year could sound very optimistic, but we must have in account than we are employing a portfolio with 11 trading systems. This is the secret. Results, in general (it depends of correlations), aren´t additive. If you have 2 uncorrelated trading systems:
System 1: Anual 10% Max DD 15%
System 2: Anual 12% Max DD 17%
The anual result will be additive, 10 + 22 = 22%, but not necessarily the drawdown, probably will be lower than 15+17 = 32%, depending of correlation.
This is the reason why we can expect important profitability.
You can try it very easy: take the portfolio builder from C2 and mix uncorrelated trading systems. Also, if we sum the monthly fee of each one of them I think this portfolio fee is not expensive.
No. I employ an unique contract for trading system.
When portfolio increases equity over the double, 60k, I will consider to double position size. Probably we will begin to employ a money management algorithm to improve the result.
Have in account the previous point. This strategy employs 11 system with 30k, the other portfolio than I have employs 5 with 25k. There are 6 more strategies with only 5k! The uncorrelated strategies let us to make this. That´s the reason to ratios are higher still.
Almost trading system are working well since 2006, and others since before, 2000 or 1996. I employ very simple trading system, with few or none parameters, to have robust strategies.
Anyway, all strategies wear away with time (ones more than others) so we employ 11 different strategies. It give us an aditional diversification effect and let than, If one or two strategies begins to perform bad, the others will compensate this effect.
I´m reasonably optimistic with future results.
I also have a PhD (in quantitative social science methods from the Univ. of Chicago). Whether fit properly or overfit, most models underperform because the model is optimally fit to an existing dataset, not to future data. That’s just elementary statistics.
Logically, this is because the trading signal generated from past data is fit not just to whatever is real, lasting, and important, but also to some things that are random. The real part continues to work in the future but the random part does not. Also, the future is not the same as the past (as your comments recognize): times change.
I’m not criticizing your models or your success with them; you have a good concept and good success so far. You probably have real trading and modeling skills. And if your models work as well as you project, your fees will turn out to be warranted.
From info above it is clearly visible, that your business model is based more on marketing expectations that credibility of your systems. That brings the question about credibility of you as a person and money manager. Smart trader knows when to cut his/her losses or reverse position. My suggestion: lower price of your system to $100, offer for current subscribers one month free. From business point of view, you should see results very fast. Also you know what I call by intelligence. Not education. When you have two choices on the table, ability to chose the right one.
I’m resting my case. This was the first time I did type on C2 public forum, and probably it will not happen again by some time. Good luck!
If you can deliver just half of the performance you expect in several consecutive years your fee is just fine but we don´t know yet of course. But basing on the information you published it´s absolutely crucial that the correlation of your strategies don´t change significantly in order to not suffer large losses/account wipe.
I´m employing portfolio techniques in my own strategy and in my opinion one has to make sure that there is enough cash right from the start to survive even strong temporary changes in correlation. Put differently, to be somewhat “safe” there has to be enough money to survive additive maxDD of all strategies and still continue trading like normal. At least that´s how I handle it. But this way prospected annual returns are automatically limited because you can not “leverage” that much.
To me it seems your approach is to maximize returns whereas I´m always watching the worst case risk. Your published screenshots already state that minimum account size required for the portfolio is 47.400$ so you are definitely in the risk zone with 30k starting balance even if correlations stay as good as in the backtest.
I don´t say your strategy won´t work. But it´s neither “safe” which is important for investors to understand. (You already mentioned that you apply aggressive capital management.)
Lower price for my ‘business model based on marketing expectation’?
Ok, ok… Thanks for your lessons marekj, but let me manage my strategy.
thanks for your critique, I think is very correct.
I absolutly agree, correlation is the primary pillar.
I didn´t desing this portfolio thinking about maximize returns, thinking in Return/Max. DD. The risk is the main metric for me in trading. Other combinations have higher returns but this is the best for me.
I was checking the value than you commented and really is strange, because almost of systems are intrady. If you remain open positions for days logically you will need more capital for margins…
I found how Multicharts calculates the value:
Account. size required=MinCloseToCloseDrawDown + Margin * MaxNumContractsHeld.
The MaxNumContractsHeld is calculated along all portfolio live, so the value is not exactly the maximun margin requeriments than portfolio has needed each moment.
Any which way, I´m going to check the historical simultaneous positions to determine exactly the minimun capital each moment and the effect of can´t take new positions.
I dont agree with the ‘neither safe’ adjetive, I think than drawdowns are acceptable comparing with returns for intraday strategies (a lot of trades for sample) in large period.
Portfolio now is TOS
Reducted price to 149 / month
Backtesting results on your website go back to 2010.
It would be really interesting to backtest the model from 2007, to see how well it performs during the 2008 crisis.
Would that be possible?
Thanks in advance,
Note than backtesting in website are from 01/06/2006 until 2017, not from 2010:
Is´nt possible to have previous backtest because a lot of commodities only have data since 2006 and we need some months to calculate indicators.
Soon we will have a new portfolio configuration with new systems developed along this year. I will update this post and website when configuration will be finish.