It varies from model to model. Testing typically shows 1-4k per contract max drawdown for each model. The portfolios do better than the individual models, of course. The whole thing is built for robustness and reliability, not to produce flashy numbers.Quote from LMeyers:
MM,
With something like this, what sort of expected MaxDD would we be looking at?
Yes, one unit is the 6/4. The expected average return of this unit is in the area of $5k/month. This lease fee works out to about 10% of the expected average profits, which is half of the incentive fee one would pay to the typical CTA. I settled on half because the client accepts the risk that the systems will not perform in the future as they have in testing (whereas the CTA is paid only when profitable), and the client must be compensated fairly for accepting that risk. Do you think this is unreasonable?Quote from electron:
What do you mean by the unit? The whole thing of 6/4? And what is the expected return per this unit per month?
Quote from Mathemagician:
Yes, one unit is the 6/4. The expected average return of this unit is in the area of $5k/month. This lease fee works out to about 10% of the expected average profits, which is half of the incentive fee one would pay to the typical CTA. I settled on half because the client accepts the risk that the systems will not perform in the future as they have in testing (whereas the CTA is paid only when profitable), and the client must be compensated fairly for accepting that risk. Do you think this is unreasonable?
M
I've been running the present portfolio for several months now. The individual models were completed and have been traded off and on for 1-3 years (different for each model). The actual results have been consistent with the hypotheticals over that time.Quote from electron:
No, I would not call it unreasonable. How long have you been running this deal and how does the actual performance compare to the backtested one?
Quote from Mathemagician:
I've been running the present portfolio for several months now. The individual models were completed and have been traded off and on for 1-3 years (different for each model). The actual results have been consistent with the hypotheticals over that time.
M