DVI system (2/2)

In the second post on the DVI system, I will explore some effects on portfolio trading this strategy. In the table below I have summarised the performance results of the various stages of the system as well as the top 3 performers of the DVI system as I left it in the earlier post.

I like this strategy as it has been constructed based on simple rules, it works across a broad set of different trading instruments and it has shown a decent performance in the period 1/1/2003 – 1/8 2010 without optimizing the parameters. So what would be my way to trade this strategy? Would I invest in the best performing instruments or in the top 3 or …? In trying to answer these questions, my view is that we are stepping into the domain of strategical asset allocation and portfolio management. In an earlier post I emptied my head on these subjects and decided to start trying out some stuff. So here goes my first experiment.

My first thought is what would happen if I would have had unlimited funds and would be able to trade all 26 ETF’s in my universe of instruments? So I tweaked my AB code to have 26*10.000$ initital capital and trade each instrument with 10.000$ in each position it takes. Commission fees are set by IB schedule, slippage costs set to 0$ and risk free rate Rf=4%. The results are shown in the table at the bottom of this post. The results show a significant reduction of the drawdown from 14.8% (average of 26 ETF’s) to 9.3% when trading a portflio. All other performance indicators are more of less the same.

It is good to see that trading the full portfolio reduces the drawdown, however I am limited in my trading capital. I cannot spend more than say 25k$ on this strategy allowing me to take a max of 5 positions of 5.000k$ each (avoiding tx costs inefficiencies). So I further tweaked by strategy to simulate what would happen if I would trade  the top 5 performing ETF’s according to the previous results. Again results below. Results improved drastically to 17.7%  profit/year,  drawdown of 9.7% and DVR of 2.28. This is great but….I do not like the top 5 ETF’s!. The tickers are EEM, VNQ, EHW, FXI, IYR. These provide limited diversification of markets (2x asian, 1x em markets, 2x real estate) and most have experienced strong bull markets in the past. Based on this insight I will probably not trade this. It’s too good to be true.

So what whould happen if I were to take positions in the top 5 of a dynamically adjusted performance ranking? I tried out 4 different rankings and traded the top 1,2,3,4 or 5 ETF’s in those rankings when a signal was given. Each ETF’s had its own equity pool of 5000$ and each position would be 5000$ as well. The 4 ranking mechanism’s tried out were: DVI score (mid-term trend), DV2 score (Mean reversal) , short term returns*long term returns and ROC(30). In the table below the results are summarised.  It can be concluded that trading the top x ETF’s based on DV2 score on average provides similar results as trading the full portfolio. I have added the equity curve & performance table for trading the top 5 DV2 ranked ETF’s as eye-candy below.

So in short, it looks like trading the top 4 or 5 DV2-ranked ETF’s brings in similar results as trading the DVI strategy at full portfolio level and this could be a good way for trading this strategy for the people who have smaller trading accounts (like myself :-)).

Happy Trading! QD

sDD_DVI amibroker code

Table with results:

4 thoughts on “DVI system (2/2)

  1. I think we need to see the average profit/loss per trade to see if the system is tradeable/how much of the profits will be consumed by commissions/slippage/eod estimating.

  2. Hi Michel,

    glad you have the pragmatic and straight-forward approach that many of my favourite quant blogs show.

    btw your results remind me some findings by Larry Connors who is using DV2’s cousin – the RSI2. It may then be the case to test some pyramiding approach as he does: basically you would be adding more as the DV2 becomes overstretched using something like 20% of full position for 1st entry, 30% for the 2nd and 50% for the 3rd).

    Will do some testing and let you know if time permits – going on vacation for 2weeks shortly 🙂

    Paolo

Leave a Comment