More Large vs Small Cap Seasonality: Day of the Week
I’ve been talking about seasonal biases in large caps relative to small caps recently. So far I’ve touched briefly on monthly seasonality and day of the month seasonality.
In this post I’ll add another to the list: day of the week seasonality. Large caps tend to be strong relative to small caps early in the week, and weak late in the week.
[logarithmically-scaled, growth of $1]
To illustrate, in the graph above I’ve assumed a simple pairs trade, long large caps (S&P 500) and short small caps (Russell 2000), investing only on either Mondays (green) or Fridays (red).
Geek notes: the two legs of the trade are not balanced 50/50% because small caps are inherently more volatile than large caps. Allocation is based on volatility over the previous 21 days (more volatility = less allocation, and vice-versa). Trades are from close-to-close using 2X leverage, and results are frictionless.
The graph shows that over the last 20+ years, large caps have consistently opened the week strong and small caps have (a bit less) consistently ended the week strong.
Why does this seasonality exist?
I wonder if it’s related to portfolio positioning ahead of the weekend or to seasonality in option pricing (read more about calendar reversion from VIX & More).
In any case, this one is especially interesting because, unlike the other seasonality plays we’ve been talking about, day of the week seasonality does not exist for straight long/short trading; it appears specific to this large vs small cap trade.
Why is this seasonality important?
All of the large vs small cap biases I’ve been talking about lately are about relative strength after adjusting for differences in volatility.
That means that they’re only useful for either (a) pairs trading, or (b) if a trader already had an opinion on the direction of the market as a whole and was deciding how to best express that opinion (either with large or small caps).
In our case, I’ll be using these observations for our new PWB Pairs Trading strategy to bias the direction of the pair (like I currently do for our directional strategies using the Monthly Seasonality Map).
Note that all of these recent studies will be included as an addendum to the Monthly Seasonality Map (read more) and we’ll be tracking them monthly moving forward.
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: Time-based | 5 Comments
To clarify, you buy SPY and sell IWM every Friday at close, holding until Monday close?
And you short SPY/buy IWM on every Thu close, holding until Friday close?
Could you post a table of avg. trade results?
Thanks.
RE to Evo34: you have the close-to-close dates correct. I’m assuming 2X leverage (frictionless).
I’m not providing trade stats b/c I don’t want to imply that this is a trade-worthy strategy in and of itself. Like all seasonality plays, it’s a bias that should be considered but never by itself justifies a trade.
michael
any input on how you adjust for volatility, i am trying to understand and see if i can code it in amibroker..
thanks
Bill
Hello Bill – for the purposes of these posts I’m keeping it simple.
Calculate the 21-day SD of the daily % log changes for both LC and SC. Then use some algebraic shenanigans to calculate what % you’d need to invest in each to make those 21-day SDs equal. This isn’t the best way to go about it necessarily, but keeps things simple for these tests.
Not sure if you could do that in amibroker?
michael
thanks a lot..you are amazing