Most of my systems tend to be market-timing strategies. That is, they enter the market and usually exit within a couple of days. This has the benefit of having limited exposure in the market and during non-trading periods the capital can be used for alternative strategies/investments. However, I have noticed that even when running several strategies, I am in cash a significant moment of the time.
My main worry is that during those periods, I do not generate a return on my capital. Yes, theoretically one could be getting some interest on their savings account, however as I need my cash to be ready-to-go-in at all times, I need to keep it at my IB account. And IB does not give interest on this account. So I started to play around with the thought of moving into a low risk asset class as an alternative to moving money into a savings account (which in reality I am not able to do – it takes two days to transfer money from IB to my Dutch savings account and another 2 days to get it back).
Having a preference for trading ETF’s, I ended up with the idea to go long in a fixed income ETF during the times that my strategy is in cash. The ETF’s that I have selected are SHY (iShares Lehman 1-3 Year Treas.Bond), IEF (iShares Lehman 7-10 Yr Treas. Bond) and TLT (iShares Barclays 20+ Yr Treas.Bond).
So I fired up Amibroker and simulated these alternatives with the following approach:
- Use the Monthly EOM strategy as baseline
- Add some extra logic to the custom backtest module to simulate “go-long-in-Bond-ETF” as an alternative for going in cash.
- The “go-long-in-ETF” position will be exited when a new trade in the EOM strategy is taken.
- The custom logic basically takes the cash-position at any given time and adds the daily return of the day of the selected ETF to the cash – as if one was invested in the ETF. The logic also takes into account IB tx-fees (no slippage).
The logic described above is part of my standard Custom Backtest module. In this module I have implemented several functionalities, including the one described above. Recently I have made some great enhancements to it and I want to thank Frank from Engineering Returns. who has shared his CBT module. Any reader who is interested in my file, send me an email.
Here are the results in a summary table for the period 1-1-2003/21-10-2003. It can be seen that being invested in a bond ETF during the periods that the system has no active position, drives additional returns and improves the annualized sharp ratio.
During the investigations of this alternatives, I noticed that during first days of the month the ETF’s have a poor performance. I therefore made a first modification to go-long-in-ETF when day()>2 and exit the go-long-in-ETF when a trade in EOM strategy is taken or when the day()> condition is not met. The results are in the table below. Here we can see that this mod1 further improves the annualised sharp ratio of the total returns (baseline strategy+go-long-in-ETF) for the longer term oriented ETF’s.
Finally I made one last modification. Mod2 uses the RSI of the VIX as an additional filter for moving into the “go-long-in-ETF”. If RSIa(VIX,3)>70, move into real cash and do not “go-long-in-ETF”. The results below show a further improvement of Ann Sharpe ratio of the longer term oriented ETFs.
In summary, it seems that investing my free capital in a US treas bond ETF is more attractive then keeping it in my IB account where is does not generate any interest. I tend to prefer the mid-term ERF (IEF). I have created an equity chart for the MEOM with mod2 “go-long-in-IEF“.
I would be happy to hear your comments / suggestions!
– QD –