This is a multi-part series looking at how some of the simple indicators we’ve talked about in the past fared through this very interesting trading year.
In this post we look at a short-term indicator that’s new to the blogosphere, DV(2). We introduced DV(2) earlier this year and it has since been taken in all sorts of different directions by its original creator CSS Analytics, as well as Market Rewind.
I’m not as up to speed on those folks’ work as I should be, so I’ll just be testing my original super simple rules: go long the S&P 500 index at today’s close if unbounded DV(2) will close below zero, and short if it will close above zero. Note that I’m using the ETF SPY to calculate DV(2), but assuming trades were placed on the index itself.
[growth of $10,000, logarithmically-scaled]
Like our previous tests in this series, these results are frictionless (i.e. do not account for transaction costs/slippage), but could be very closely reproduced using actively-traded mutual funds (my weapon of choice). Unlike our previous test, I have not included return on cash. Buy and hold is shown in grey and the strategy in red.
YTD Performance
The strategy did extremely well not just sidestepping the 2007-09 drawdown, but also pulling down strong gains. Note however that, like our previous test of RSI(2), the strategy ran cold the second half of this year, mostly as a result of shorting a market that has consistently managed to push higher when short-term overbought.
My conclusion here is the same as for RSI(2) (and daily mean-reversion for that matter): this breakdown in short-term mean-reversion is a temporary byproduct of this very bullish leg up in the markets and NOT an indication of things to come. And I still expect short-term MR to be the play du jour again once this market returns to some normalcy.
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: Trading Strategies | 4 Comments
Confucius Says
The superior man understands what is right; the inferior man understands what will sell.
And in that one sentence, the 2500-year old philosopher sums up what ails this industry: no sense of doing the right thing for the sake of doing the right thing.
h/t Barry Ritholtz’s Bailout Nation
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 | Leave a Comment
This is a multi-part series looking at how some of the simple indicators we’ve talked about in the past fared through this very interesting trading year.
In this post we look at blogosphere darling RSI(2), trading the S&P 500. Unfamiliar with this short-term indicator? Read our previous geekery here, here, and here.
Over the last decade, RSI(2) has become more predictive the deeper it’s travelled into overbought (high) or oversold (low) territory (read more), so I’ll assume a scaling approach: go 100% long the S&P 500 index at today’s close if RSI(2) closes below 5, 75% long on a close below 10, 50% below 15, and 25% below 20. Go 100% short on a close above 95, 75% above 90, 50% above 85, and 25% above 80.
[growth of $10,000, logarithmically-scaled]
Like our previous tests in this series, these results are frictionless (i.e. do not account for transaction costs/slippage), but could be very closely reproduced using actively-traded mutual funds (my weapon of choice). Unlike our previous test, I have not included return on cash.
Note that I have also not included a “long view” (prior to 2000) because, as we’ve talked about ad infinitum, these type of very short-term indicators worked in precisely the opposite way prior to the late 1990’s (read more).
YTD Performance
After a decade of very consistent calls (see middle view) the RSI(2) indicator fell down badly in 2009, particularly when shorting against high readings. Put another way, when the indicator expected the market to pull back (“revert to the mean”) after making a strong short-term move up, the market has often continued higher.
We’re covering this breakdown in short-term mean-reversion in our monthly State of Short-term MR report. In a nutshell, I think (but do not know) that this is a temporary byproduct of this very, very bullish leg up in the markets and NOT an indication of things to come. And I still expect short-term MR to be the play du jour again once this market returns to some normalcy.
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: Trading Strategies | Leave a Comment
This is a multi-part series looking at how some of the simple indicators we’ve talked about in the past fared through this very interesting trading year.
Up to this point all the indicators we’ve been testing were from the (now defunct?) State of the Market report. But if you recall, the intermediate-term indicators we used in the report were proprietary (because they were a little too similar to concepts we use in our own MarketSci strategies).
It’d be silly to show you results for indicators you don’t know the rules for, so in their place I’m going to test another intermediate-term indicator: TradingMarket.com’s Buy New Lows Strategy. The rules tested are: go long the S&P 500 index at today’s close if the market will end the day at a 10-day low, and move to cash if it will close back above its 10-day moving average (SMA).
[growth of $10,000, logarithmically-scaled]
As mentioned, all tests are frictionless and account for a return on cash of half the nearest 13-week Treasury bill. Buy and hold is shown in grey and the strategy in red.
YTD Performance
TM’s rule did well sidestepping the worst of the 2007-09 drawdown (see middle view), but has done a poor job of keeping pace with the 2009 recovery.
I think these results are representative of most intermediate-term (IT) indicators. By most technical measures, the 2009 rally was IT overbought a good bit of the time, meaning most IT indicators would have kept the investor on the sideline too often.
For those interested, click the thumbnails below for the combined results of the SOTM report’s proprietary IT indicators. SOTM junkies know that the IT indicators actually measured both overbought/oversold AND momentum, but for simplicity’s sake I’ve assumed a trader just bought when the market was oversold and sold when it became overbought.
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: Trading Strategies | 4 Comments
Riddle me this…
Assume that at the close of trading each day we could know which way the 10-year Treasury yield would move the following day. We begin a portfolio with $1 and each day at the close we go long the S&P 500 index (frictionless) if yields will fall the next day and short if they will rise.
This is of course an impossible alternate universe where we have perfect foresight of next-day Treasury yields, but bear with me. The results:
[growth of $1, logarithmically-scaled]
From 1962 that $1 would have grown to almost $4 million by early 1997, because the S&P 500 was very much inversely tied to the 10-year yield (and the 5-year and the 30-year for that matter).
But note how in early 1997 this relationship suddenly and without warning flipped on its head. For the entire 10+ years that have followed since, Treasury yields and equities have either had no relationship (note flat spell in the middle of this decade) or a very positive relationship.
I was using Yahoo’s simple index ^TNX when I stumbled on this observation, so my first reaction was that this was a phantom; that there had been some change in how the index was calculated. But the same results play out in futures data (going back to 1983) and more recent ETF data.
Possible Connection to Daily Follow-Through?
Astute readers will note that the shift occurred right around the same time as the shift in daily follow-through from being momentum to mean-reversion driven.
For the uninitiated, for most of the stock market’s history, up days tended to beget further up days (and vice-versa) but around the late-1990’s, that relationship flipped contrarian; up days now tend to beget down days (and vice-versa). [read more]
Below is a chart showing the original ”strategy” (grey, left scale) with a second strategy (red, right scale) that goes long the S&P 500 index at the close if the S&P 500 closed up today, and short if it closed down (unlike the first, this one is reproducible). Note the similarity.
[growth of $1, logarithmically-scaled]
So What Does it All Mean?
I’m not sure.
I don’t think the answer is related to some temporary change in market fundamentals. The shift was just too abrupt and sharp for that. Instead, this appears to be a change in some basic characteristic of the market.
Is it related to the similar shift in daily follow-through? Possibly, but I’m not sure how you could demonstrate that conclusively.
Could this be a phantom based on the quality of data used? I won’t rule that out, but again, I did reproduce the observation using futures and ETF data.
Is this observation important? Assuming that it’s not just a phantom, then yes, all market relationships that were this consistent for this long that suddenly make this drastic a change are important…I’m just not sure why or how yet.
That’s all (for now). I’m putting this riddle out there for our very smart reader community to chew on. But as always, more to follow.
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: Evolving Markets, Follow-Through, Treasuries & Interest Rates | 9 Comments
-
Subscribe to RSS Feed
Subscribe via Email
My email: michael @ marketsci.com (remove spaces)
