Using strength of recent closes to time broad asset classes
Back in November I posted a strategy for timing SPY using the strength of recent closes. I looked at the 4 day average of a closing strength ratio reffered to by some as internal bar strength or IBS (IBS = (close-low)/(high-low)). The equity curves were pretty impressive.
I’ve recently tested a couple related strategies. One idea involved a long/short approach to timing country stock ETFs. I found some strategies that appeared to work great for a while but have become less and less effective over time.
Another idea I recently tested goes long one or more broad asset ETFs based on average IBS.
Here’s a description of one implementation.
- Basket of ETFs includes SPY, EEM, GLD, TLT, VNQ.
- Near each day’s close, calculate the 4 day moving average of IBS and rank each ETF by this value.
- At each day’s close, buy equal shares of the 2 ETFs with the lowest average IBS. Sell positions 3 days later.
- At any given time the strategy will be holding 2-5 ETFs purchased on 3 different days.
- $100K invested per trade (split equally among the 2 ETFs being purchased).
Below is an equity curve before slippage and commissions and any dividends. Average trade is .28% with 58% winners.
I’ve tested out IBS moving averages between 1-7 days and it seems pretty robust. I’m sure there are ways to improve the strategy with either some kind of equity curve timing, stops, etc. It would also be interesting to try out other combinations of ETFs. This is the first such combo I’ve tried.
Posted on January 4, 2013, in Other. Bookmark the permalink. 5 Comments.
Pretty cool, nice to see IBS working outside of equities, too. Would be interesting to see how much each ETF contributes to the returns.
Drawdown in 2008 looks rather painful, maybe some kind of volatility-based sizing approach could help reduce it.
Thanks, it works better than I expected. I should add charts showing how much each asset contributes over time. TLT had the smallest average trade profit, though it may just be a function of less volatility relative to the others.
I agree that 2008 would be very painful to endure. More work would need to be done to make something like this a tradeable strategy.
Hi,
I have one crucial question, You enter position on same day when signal is generated. But if you work with EOD data, signal is generated after the close. How do you solve this problem. You buy on other exchange systems in after market hours?
Thank You,
Peter
The fallacy in this backtest (and other EOD systems) is that it assumes you execute trades before really knowing the closing value. MOC orders are probably out of the question for entries. You could retest the strategy using values 15-20 min before the close to see if a MOC approach is nearly as profitable. Otherwise, you would need to execute trades very close to the close, or right after the close in after hours.
Pingback: Notes and Quotes from Jaffray Woodriffs's Hedge Fund Market Wizards Interview - ennlightenment