Got Your Attention?
Good because I’m going to move fast as this is a very straight forward spreadsheet based mechanical system that uses the Canadian Dollar to trigger buy and sell signals for the S&P 500. The headline is not some bait and switch headline that has a catch. No, this is a 30% return on an initial investment made with a beginning trade in 1999 tracked through up to March 2, 2009.
For once it’s not a mean-reversion trading system either. No moving averages, no Chi-Scores. Nothing. This is as about as simple as it gets when it comes to defending your 401k against the ravages of the financial world: the professional market maker.
Introducing…The Loonatic
First things first. For those of you that might not be aware of any significance to the name, please note a few things at the outset.
First: The national bird of Canada is The Common Loon
.jpg)
Yep, that's a loon
Secondly: the nickname for the $1 coin here is called a Loonie:
Now you know why it's called that
Thirdly: many of the financial instruments in play correlate in some what to the US stock market. For instance, one of the spreadsheet based mechanical systems I developed tracks the relationship between the UUP ETF (US Peso Bullish) and the DOW (using DOG).
Now, while that system was mean revision related, albeit very slow – yet accurate, it didn’t really strike me as being a system that there could be a great deal of validation for a while due to it’s lack of rapidity.
In fact, the more I thought about it – the more I realised that if this relationship was true, than the inverse would also be true – that there would be a signal or signals that would relate to the value of the US Market and could be used as a timing mechanism for long entries.
No shorts, the system only entering when the market had reached a ‘low’ and was headed higher – then exiting before a top was reached.
Furthermore, I know from prior analysis of the informational entropy of the US and Canadian markets I knew that a simple timing mechanism could be created using nothing more than the price of the US Peso as valued in Loonies. For my Canadian readers – this means that you can track this system using nothing more than the Bank of Canada website’s currency exchange information.
Loonatic System Details Here:
0. Column A is the Date
1. Column B is the Adjusted Close of the S&P 500 (I prefer Yahoo! for their data breakdown. Google is still in ‘beta’)
2. Column C is the Value of the US Peso in Canadian Dollars
3. Column D is the 10 Day Correlation of the US Peso (in CAD) and the S&P 500
4. Column E is the 30 Day Correlation of the US Peso (in CAD) and the S&P 500
5. Column F is the Signal. Here’s the juicy bits:
Buy Signal == When the 10 Day Correlation is more than the 30 Day Correlation
Sell Signal == Anytime the Buy Signal has not triggered and you have S&P 500 shares
The Results
Since Feb 16, 1999 when the system took it’s first trade (Buy signal at 104.69), the system took 139 trades. Of those 139 trades, 80 were profitable giving the sytem a 58% accuracy. With a Two-Tailed P value of .0749 it is considered almost signficant.
The Average trade was a positive $.23/share with a Standard deviation of $3.23.
The Largest profitable trade was $17.76/share. (aside from that outlier the largest was $13.91)
Average profitable run was $4.64/share
The Largest losing trade was $12.45/share.
Largest DrawDown was -$13.87/share.
Average DrawDown was $-3.83/share
Assuming the system was used to buy 1 share of the S&P 500, this system would have generated a 31.46% return on investment of $104.69.
Expectancy = $1.09/share
UPDATE
Average positive trade was $2.16
Average loss trade was -$2.4
Expectancy = (.58*2.16)-(.42*2.4) = $.25/share
(HT to Wood who asked about the individual trades vs. the expectancy I had created on trade runs vs drawdowns)
The Takeaway
This system is about as unrefined as can be since it’s the product of about six hours work, most of it papertrading through the spreadsheet. I feel strongly that this system has some promise but it was a refreshing change of pace to look at the market data without the ‘mean revision’ blinders on.
Theme Song



(8 votes, average: 4.5 out of 5)

Cuervo, that is really cool.
So the average losing trade was $2.5825 ?
Again, good work!
No, it was
Average DrawDown was $-3.83/share
Post corrected.
Thanks Wood.
For some reason WordPress won’t allow me to use the strikethrough on the original expectancy.
Very nice! A couple questions – are you using pearson correlation / is that 30% total or annual?
@manatrader
Yes, it’s the Pearson correlation.
Yes, it’s 30% over 10 years.
Not so much to brag about but the system does have a positive expectancy.
I’m working on a revision to the system to increase performance with a post to come shortly there after.