January 14, 2015

Using 52-week highs in a S&P500 monthly rotation strategy

One area of recent interest for me is trading rotational strategies on a monthly timeframe using S&P500 stocks and ETFs. Areas of exploration include Momentum and Dual Momentum. Recently I came across The Secret to Momentum is the 52-Week High??? on Alpha Architect, a blog I highly recommend on reading along with the quant mashup Quantocracy.. The article is a synopsis of research done comparing momentum vs. 52-week highs as ranking filters for a rotation strategy. A new idea I had not tried. What a great way to start the year, testing a new idea. Even though often they do not work out, one needs to be exploring all the time.

In the paper, they used 6 month holds which is much longer than I want to hold any position. I dropped the holds to one month and see if that would give me interesting numbers or not.

As a side note, my delay on posting this year is because I have been busy with my son’s FTC robotics team. I wish I could have done this when I was in high school. If you have high school aged children into math, science, building or programming, I recommend you check out FIRST. We have been doing this for seven years. It is a great experience.

Test Notes

Test dates are from 1/1/2005 to 12/31/2014 using the S&P500 stock universe as they existed.

The Rules

First we want to compare the top 25 vs the bottom 25 and see if there is something here.

Test 1: Closest To 52-week high

  • Stock is in the S&P500
  • On the last trading of each month, rank stocks from closest to 52-week high of closes to furthest.
  • Buy the top 25 on the next open
  • Sell all at the beginning of the next month at the open

Test 2: Furthest From 52-week high

 

  • Stock is in the S&P500
  • On the last trading of each month, rank stocks from furthest from 52-week high of closes to closest.
  • Buy the top 25 on the next open
  • Sell all at the beginning of the next month at the open

 

Results

150114a

A very promising start. As always we need to look deeper and compare to Buy and Hold.

Buy & Hold

150114b

Comparing to buy and hold on the SPY, now the results don’t look so great.

What about buckets?

Next we check to see if looking at buckets of 25 stocks each shows a pattern. This means the first bucket contains stocks ranked 1-25. The next bucket is from 26-50. And so, while the last bucket is stocks 476-500.

150114c

What I like to see is a pattern with the best results on top and they get worse as you move down. Of course it is never perfectly that way. Unfortunately we do not see that. What I take from this is avoid the bottom bucket, but then again that is an area most people are not trading.

Testing on the Russell 3000 stocks gave similar results.

Final Thoughts

Even though the 52-week high idea did not work for me, it does not mean that it does not work in the context as presented in the paper. This part of doing research, taking ideas that sound interesting and trying them for the context you want. The lesson to learn from this is always be looking for new ideas and test them thoroughly. Imagine had the first bucket returned 11%. We would have thought that the idea would work. It was by looking at all the buckets that we understand that it is noise.

Backtesting platform used: AmiBroker. Data provider:Norgate Data (referral link)

Good quant trading,

Fill in the form for the spreadsheet

For a spreadsheet with yearly returns and testing against Russell3000, fill in the form below.

 
 
Visited 27 times, 1 visit(s) today

Click Here to Leave a Comment Below

Ilya Kipnis - January 14, 2015 Reply

Yeah, I’m not a fan of these long-term momentum strategies. In fact, Alpha Architect recently released a paper about the long term reversion effect of momentum, which is why the anomaly may not have been picked up for all these years, what with all the academics using god knows how many years of monthly data. Trading on the 52-week high seems to be asking for huge drawdowns, and indeed, they do appear.

The Whole Street’s Daily Wrap for 1/14/2015 | The Whole Street - January 15, 2015 Reply

[…] Using 52-week highs in a S&P500 monthly rotation strategy [Alvarez Quant Trading] One area of recent interest for me is trading rotational strategies on a monthly timeframe using S&P500 stocks and ETFs. Areas of exploration include Momentum and Dual Momentum. Recently I came across The S… […]

Using 52-week highs in a S&P500 monthly rotation strategy | XLNtrading - January 17, 2015 Reply

[…] Using 52-week highs in a S&P500 monthly rotation strategy » Alvarez Quant Trading. […]

Abel Duarte - June 8, 2017 Reply

Hey Cesar,

not sure if you wrote another article following up on this piece but there are a few things that can be done to improve the results.

Most stocks near the 52 week high tend to pullback as those who don’t think the price can go higher sell out. Your 1 month time frame might be too short as you are literally buying the stocks making new 52 week highs that will soon pullback.

I would recommend

1) increasing the holding period slightly
or
2) making use of the 52 week high momentum effect using diagonal spreads using options (the closest month to expiration will decay as the security pulls back, then you can enjoy the continuation of the move with the remaining back option. You ideally want the back option to be slightly in the money and the sold option to be ATM or OTM so when it does pull back you get to keep most of the premium.

    Cesar Alvarez - June 9, 2017 Reply

    (1) Some people will say that you are at a breakout and should buy then. I like your suggestion. Should not be that hard to test.
    (2) No options data.

Aaron - April 19, 2018 Reply

I don’t personally believe a long only momentum strategy with no filters can work in a bear market. I think you need to ensure the stock and/or the market is going ‘up.’

    Cesar Alvarez - April 19, 2018 Reply

    I agree with you that having a market timing filter. Most of the strategies I trade have one. It would be interesting to see how these results would change with one. Maybe a future post.

Leave a Reply: