Putting the Monthly Seasonality Map to Good Use
I’ve been on a seasonality kick the last couple of months fleshing out how different seasonality events (Fed days, options expiration, turn of the month, etc.) impact today’s market.
The culmination of that effort is the Monthly Seasonality Map where I quantify bullish/bearish biases for each day in the coming month.
It’s too early to get too excited about, but since launching in April, the Seasonality Map has called the closing direction of the S&P 500 correct 60.0% of the time with winning predictions 1.03x losing ones.
I’m pleased with that and I think it’s time to put the Map to use in my own actual trading.
Beginning with Tuesday’s trade (06/08), both our YK and Scotty strategies began adjusting trading decisions based on what the map says about that day.
As I’ve repeated ad infinitum, I don’t think seasonality alone justifies making a trade; however, I do think it justifies NOT making a potential trade.
So for example, if our strategy calls for being long 100% on a given day but the Seasonality Map has tagged that day as having a -50% bearish bias, I’ll reduce the trade to say long 75% (note: this is a fictional number and not quite how I’m actually adjusting for seasonality, but it’s similar in spirit).
Again, the idea isn’t to take a position based on seasonality, but use it to possibly reduce the aggressiveness of a position.
That’s all for this post (a rare non-geeky one). I just wanted to share my own thoughts about how I intend to apply seasonality to my actual trading.
The Monthly Seasonality Map (like everything I do) is a perpetual work in progress, and reader input in always invited.
Happy Trading,
ms
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Random Stuff | 7 Comments
Good post, short but very apt. I have just started to do the same in my trading – actually 2 days before your post (great minds think alike:)). The change in my trading was necessitated by the May crash, which came out of nowhere (not a slow, gradual rollover like 07/08).
Michael.
You have discussed many promising strategy ideas on your blog over time, and we’ve assumed that the reason you were choosing not to incorporate any of them (other than the volatility filter) into your existing strategies is that it would necessitate the commencement of a new track record (e.g., when you added the volatility filter to YK in late 2008).
We are very pleased to hear that you have chosen to add the seasonality factor to your YK and Scotty strategies, but — in light of the importance you place on transparent track records (a practice we greatly respect) — are curious to learn if you are planning to start a new track record for the new “YK(C)” and “Scotty(B)” strategies. If you are not planning to do so, then please share your rationale.
RE to George: good questions…
If I don’t include something discussed on the blog in our proprietary strategies it’s either because (a) the idea isn’t “good enough”, or (b) it doesn’t match the spirit of the strategy (ex. YK can’t suddenly become something other than an aggressive short-term strategy…just doesn’t make sense). However, I’m not at all above improving an existing strategy and that’s something I’m constantly working on doing.
The track record isn’t a factor because the track record is real-time actuals. Any modifications to the strategy would only be reflected from the day they went into effect (again, I never issue backtested results, only real-time actuals).
The AMF was a radical redesign and I think justified a scratch and replace. I don’t cherry pick returns, so I’ve continued to carry both YK versions in all of our stats and performance charts. This Seasonality modification doesn’t fall under that header. The strategy is getting a trim, not a whole new hairdo.
michael
Hi Michael,
I’m a little confused by your statement:
“As I’ve repeated ad infinitum, I don’t think seasonality alone justifies making a trade; however, I do think it justifies NOT making a potential trade.”
It seems that NOT making a potential trade is equivalent to making both the original trade and the new one.
To clarify, are you saying you will take less than the full recommended position if the Seasonality Map dissents with your strategy, but not more than the full recommended position if the Seasonality Map and your strategy agree?
Also, if a strategy makes money both in backtesting and real-time (as I believe you are implying about the Seasonality Map), why wouldn’t it be tradable? Perhaps I missed that in an infinite number of prior posts… haha.
Thanks.
RE to Ryan: I put the “ad infinitum” bit in there as a subtle apology for sounding like a broken record about certain things (like that statement =)
This paragraph is correct: “To clarify, are you saying you will take less than the full recommended position if the Seasonality Map dissents with your strategy, but not more than the full recommended position if the Seasonality Map and your strategy agree?”
Two reasons I can’t see trading seasonality alone:
1. It needs to be coupled with a price-based indicator because (by it’s nature) it’s not getting a full view of the market. Saying the first day of the month has historically been bullish is fine, but what if the last day of the previous month is up or down hard? How does that change the seasonality bias?
2. Seasonality plays just haven’t been as strong or consistent (alone) as price-based plays, be they something very short-term (ex. RSI(2)) or something long-term (ex. long-term MA crossovers). Good enough to fine tune an existing signal, but not good enough to stand on their own.
Just my $0.02.
michael
I’d love to see a post from you about lunar effects. As you probably know, CXO just did an interesting post on the topic:
http://www.cxoadvisory.com/calendar-effects/lunar-cycle-and-stock-returns/
The original research shows persistence across multiple indexes, including indexes for other countries, which I imagine would appeal to you.
It seems to me that there is a lot more to explore on lunar effects, as the granular daily breakdown CXO did only looked at the S&P 500 over the last few years. I imagine that a granular breakdown of a larger data set (for instance, including foreign markets) would give a much better picture of how the moon phase affects the markets.
RE to FM: I did a similar (semi-tongue-in-check) study way back when and came to similar results as CXO. I’m wondering how much of the lunar cycle observation is acually the Monthly W I’ve written about?
http://marketsci.wordpress.com/2010/05/04/the-%e2%80%9cmonthly-w%e2%80%9d-in-video/
On my todo list to take a look. Thanks for the comment!
michael