Alexander, A very nice strategy! I like your idea how you rotate or dynamically allocate the four asset groups Stocks, Bonds, Gold and Cash

Anton, thanks. The idea of adaptive allocation is not new, but we are employing it now jointly with volatility balancing with good success. Further examples are http://goo.gl/4dgppt and http://goo.gl/ZsDZCk. It goes a bit in line with your iM-Best(SSO-TLT) http://goo.gl/qG1xAQ system, which I like very much! Happy to exchange further..

Alex, can you expand a bit on your new C# backtest software. How easy is it to plug in other models? Do you plug in a trade table or does the backtester actually run the model?

It's a nice Fund-of-ETFs strategy. I'm also interested in your idea.

Besides pure ETFs, would you also include some relevant futures contract to offset your downside risk in some bad months?

Or, during some particular periods that performance were flat, would you also consider to use other non-linear derivatives (e.g. exchange-listed options) to overcome such situation?

Many thanks for your sharing your strategy website.

@Anton, the 'C# backtest software' is rather complementing the backtesting package which we normally use (Quantshare). We plugged our main routines into this native and slim C# app, so our "production strategies" can run there faster. It actually runs the models, maybe at a later stage we might think about offering it, but that's actually not our main business.

@Wesley, yes, indeed we do much of our own trading in futures instead of ETF. Good part of our recent work is on Hedged Volatility and the above mentioned adaptive strategies which also include a "soft" hedging element. See recent blog articles http://www.logical-invest.com/blog/

As most of our readers are retail investors we give several options of how to trade the strategies depending on account type and size, e.g. plain ETF, shorting 3X leveraged ETF, futures and sometimes OTM/ATM options. Latter is rather a small element for hedging, not for driving performance. Example here http://www.logical-invest.com/universal-investment-strategy/

as far as i understand, you take a strategy and you start fiting parameters, or going for something like this:

".... we will identify the worst performer. We will decrease funds invested in that asset and distribute those funds to the rest. So if gold underperforms all other assets, we will sell some gold, divide the proceeds in three and buy equal amounts of the SP550 index, Treasuries and Cash. "

....and, yeah, the performance goes up...am i missing something?

Hi Tasos, share your general critique on overfitted parameters, worst one can do.

This is why we always share heatmaps or paramater grids like the one here over a 20 years backtest http://tiny.cc/tdw1qx

Allocating systematically capital to TopX performers equally or by a ratio, or like in this case reducing x% from worst performer seems a rather well-accepted process in quite some literature to me. But maybe I did not catch your point?

Hi Alexander,

My point is: did you have a training set and a testing set?... or in the same data you reduced the exposure to the bad ones?

Yes, as part of our process we run both in-sample/out-of-sample tests, when reasonable also walk-forward or full bootstrap tests. We normally do not publish these as our articles are for the retail investor, and even we publish 80% of the construction we keep a bit of "secret sauce" when it comes to proprietary processes.