# TMFJake (30.83)

## TMFJake's CAPS Blog

Recs

### 21

September 02, 2009 – Comments (35)

Somebody posted a blog post today asking this question, and I realize that I haven't posted a CAPS Ratings performance update for several months.

CAPS Ratings Quintiles

Statistics

Each rating quintile is treated as an index and measured over the time series. To use the 5-star stocks as an example, we are NOT looking at stocks that recently achieved 5-star status and looking at their TTM returns. We ARE measuring the performance of stocks that have already achieved 5-star status over a time series. That is to say, on 1/07, all stocks that had achieved 5-star status are included in the 5-star index. Throughout the time series, stocks would be added or removed from the 5-star index depending on whether they crossed into or below the 5-star rating threshold.  In this model, each quintile index is rebalanced every day so that positions are equal weighted.

#1) On September 02, 2009 at 4:51 PM, BravoBevo (99.96) wrote:

In running the math, did we perhaps forget to "carry the one" into the next column?  Interestingly, from the 2nd line of the statistics (in reverse order) titled "YTD %Return":

S&P = 9.32%
5 Star = 52.52%
4 Star = 40.35%
3 Star = 35.02%
2 Star = 43.22%
1 Star = 10.12%

In other words, every single tier of rated stocks in the entire securities universe - from the highest-regarded 5 stars to the most despised 1 Star - all performed better than the S&P.

This event, where all classes of securities surpass the S&P, can only happen in Lake Wobegon where "all the women are strong, all the men are good looking, and all the children are above average."

Report this comment
#2) On September 02, 2009 at 5:24 PM, TMFJake (30.83) wrote:

Certainly not a desired result with the 1-stars...

Report this comment
#3) On September 02, 2009 at 5:33 PM, portefeuille (98.47) wrote:

Could you please elaborate on those findings.

Report this comment
#4) On September 02, 2009 at 5:34 PM, portefeuille (98.47) wrote:

An update of the numbers presented here would be really nice.

Report this comment
#5) On September 02, 2009 at 5:38 PM, portefeuille (98.47) wrote:

The sharpe ratio for the "1 star stocks" is obviously wrong if those for the other subgroups and the benchmark are correct.

Report this comment
#6) On September 02, 2009 at 5:40 PM, portefeuille (98.47) wrote:

What time period is used for the calculation of the sharpe ratio?

Report this comment
#7) On September 02, 2009 at 5:45 PM, TMFJake (30.83) wrote:

I agree that's it's a little surprising that the S&P is underperforming all ratings quintiles, but it's not at all unfeasible.  And it's not a good result for our 1-star and 2-star performance objectives.  Fortunately, the spread on the high end with the 4 and 5 stars makes up for any bad behavior from the 1 and 2 star stocks.

Here are the returns from the the beginning of the time series:

Returns since Jan. 2007

1-Star= -53.18%
2-Star= -25.09%
3-Star = -13.75%
4-star = +3.96%
5-Star = +40.69%

SPY= -30.29%

@portefeuille, I'll publish updated Harvard-Yale analysis as soon as they send it to me.

Report this comment
#8) On September 02, 2009 at 6:31 PM, UltraContrarian (30.68) wrote:

TMFJake (96.14) wrote: Certainly not a desired result with the 1-stars...

1 and 2 star stocks beating the S&P is undesired in terms of CAPS model performance, but the structural forces of the speculative rally are so huge that it's almost impossible to avoid.  To me the 2 star performance inversion over 3 and 4 star stocks is a lot bigger problem than performance versus the S&P.  (Although both will probably be sorted out quickly.)

Report this comment
#9) On September 02, 2009 at 6:34 PM, UltraContrarian (30.68) wrote:

Also this data explains pretty well how I became an allstar green thumbing 1 and 2 star stocks and red thumbing 4 and 5 star stocks. (Which should be pretty difficult, but it wasn't.)  And the long run charts indicate how dificult it will be to keep up this strategy's performance.

Report this comment
#10) On September 02, 2009 at 6:39 PM, portefeuille (98.47) wrote:

I think the sharpe ratio may be okay. Sorry for that.

Report this comment
#11) On September 02, 2009 at 6:44 PM, portefeuille (98.47) wrote:

Also this data explains pretty well how I became an allstar green thumbing 1 and 2 star stocks and red thumbing 4 and 5 star stocks.

Yes, I also noticed that (on May 21).

------------------

something on this "contrarian" issue (see this post):

One thing that I found somewhat annoying and disturbing was that my "Average Pick Rating" (apr) has always been at 4 stars, at least since I first looked at it. The annoying/disturbing aspect of that has been that I would have liked being the "caps" game outsider who is catching all those "falling knifes" and really "unpopular" stocks right when they turn (some sort of master of bottom fishing ...).

I just noticed that my apr is now at 3 stars. Yeah, baby (you see that it does not take much to "put a smile on my face") ...

Of course (OCD and curiosity going astray and all that ...) I had to check what the apr of some other players was. I started from the top of the ranking system. And I may have actually found something new and maybe not entirely uninteresting (see comment #63 here)).

The system I found breaks down somewhat beginning with trimalerus who is currently ranked #26.

But all players in the top25 (not mentioning them because they are of any importance (see comment #29 here), but because I started at the top) either have an apr of four stars or ...

or they might be in the group of "contrarians within the 'caps' game" (see comment #3 of this post).

They are (apart from my players portefeuille, hdgf2 and hansschmidt): ultralong, fransgeraedts, bullmarketn09 and tigerpack. They all have an apr of 3 stars. And you might want to have a look at the point score graphs for those players: 1,2,3,4,5,6. I think they show a certain similarity.

Well, not that exciting all that, but now that I have written it I might as well "publish" it ...

------------------

(from here)

Report this comment
#12) On September 02, 2009 at 6:46 PM, portefeuille (98.47) wrote:

Those that think they are contrarians usually are not. The star rating system is not that great especially considering those \$100M and \$1.50 rules that can really distort things.

Report this comment
#13) On September 02, 2009 at 6:50 PM, portefeuille (98.47) wrote:

@portefeuille, I'll publish updated Harvard-Yale analysis as soon as they send it to me.

Can you maybe give us a hint? I was somewhat impressed by the results of the first paper and if they are similar for the period starting where their study period ended then fransgeraedts may be right about the potential of the "caps" game.

Report this comment
#14) On September 02, 2009 at 8:01 PM, TMFJake (30.83) wrote:

@portefeuille, we have attempted to replicate the same methodology as the Harvard study with data through early 2009 (when we ran the test) and found that the results were consistent with their first study.

Report this comment
#15) On September 02, 2009 at 8:01 PM, TMFJake (30.83) wrote:

@UltraContrarian, right on both counts. :)

Report this comment
#16) On September 02, 2009 at 8:09 PM, TMFJake (30.83) wrote:

@portefeuille: It won't surprise you to hear that I agree with you on the potential for CAPS.  Although we need to do something about that 2-star band as UltraContrarian notes...

Report this comment
#17) On September 02, 2009 at 8:09 PM, portefeuille (98.47) wrote:

tmfjake, with all the data you have I think the best way to improve the "caps" game star ranking system is to simply backtest different ones. I don't think that the star ranking has a large feedback on the calls that are made in the "caps" game, so there should be no major problem with that.

Report this comment
#18) On September 02, 2009 at 8:09 PM, TMFJake (30.83) wrote:

Of course, long term the 2-stars have exhibited behavior in line with expectations...

Report this comment
#19) On September 02, 2009 at 8:18 PM, portefeuille (98.47) wrote:

Although we need to do something about that 2-star band as UltraContrarian notes...

But please don't try to do any "ad hoc" things do "fix" it. The formula is still proprietary, I guess, so I can't help in finding the origin of this "anomaly" but I think it should not be all that difficult to spot where it comes from. My naïve suggestion would be do use smaller subintervals (so use say 50 instead of 5 "star intervals"). But not knowing that proprietary formula I could really just guess what is going on. Maybe there are some "discontinuities" in that "formula". It would be hard to imagine how a "smooth, continuous" formula could produce these performance jumps from one star period to an adjacent one. One explanation could of course be some sort of feedback (star ratings -> calls), but, as I said, I doubt that, especially since the 2 star rating is showing the anomaly and who really bases their call on a 2 star rating vs. say a 3 star rating. I am confident that you find the "culprit".

Report this comment
#20) On September 02, 2009 at 8:19 PM, portefeuille (98.47) wrote:

would be do

would be to

Report this comment
#21) On September 02, 2009 at 8:20 PM, portefeuille (98.47) wrote:

Of course, long term the 2-stars have exhibited behavior in line with expectations...

Okay. It would have been strange otherwise.

Report this comment
#22) On September 02, 2009 at 10:17 PM, ChrisGraley (29.26) wrote:

Sorry TMFJake, but I have my doubts about the math. While I do understand that it is possible that all stocks can outperform the S&P as a group, that would usually mean that the DOW would outperform the S&P as well. I understand that that there is a slight possibility that the average of all stocks could outperform both the DOW and the S&P, but I can't fathom this scenario in a down market.

Can anyone elaborate?

Report this comment
#23) On September 02, 2009 at 10:24 PM, portefeuille (98.47) wrote:

Can anyone elaborate?

The S&P 500 index is a market capitalisation weighted index of 500 large cap stocks "actively traded in the U.S.". Those 5 "caps" indices are neither confined to U.S. large cap stocks nor are they market capitalisation weighted.

The Dow Jones index is simply ridiculous.

Report this comment
#24) On September 02, 2009 at 10:26 PM, portefeuille (98.47) wrote:

Or even shorter. The S&P 500 and Dow Jones indices care about XOM, those 5 "caps" indices don't ...

Report this comment
#25) On September 02, 2009 at 11:24 PM, TMFJake (30.83) wrote:

Agreed.  We'll be running August numbers in a few days, which will give us a chance to double check the math as well as update returns over the last 30 days.  But I agree with portefeuille's answer to you Chris.

Report this comment
#26) On September 03, 2009 at 9:25 AM, MikeGi (25.30) wrote:

Hey Jake,

Would it be possible to see the data from a different starting point?  For example, how have the rankings done over the past 1 year, 2 years, or sincethe Lehman bankruptcy, etc.  A graph that users could interface with would be fantastic.  Not sure how much of this you do manually vs using some tools you have made.  Or even if you would be willing to make the data public.

Also, what is the std dev of the different categories?

Finally, would it be possible to view these graphs/stats on an individual sector basis?  i.e. how have 5 star financial stocks performed vs 1 star financial stocks.

Report this comment
#27) On September 03, 2009 at 9:52 AM, ozzfan1317 (74.28) wrote:

Good stuff thanks for the info.

Report this comment
#28) On September 03, 2009 at 10:51 AM, TMFJake (30.83) wrote:

@MikeGi: At this point, unfortunately, we don't have a slick tool that we could put on the web site.

But I'm happy to address specific questions.  In this blog, I've shown YTD and returns since inception.  From looking at the data, it looks to me like 5-star stocks are up 11% from 9/1/08, a nice recovery after a 50 percentage point decline by 11/20/08.  That was the trough, so 5-stars returned >60% from their lows.  So, 5-stars didn't fare any better in the Lehman collapase, actually did worse than the S&P, but they didn't collapse as much in early 2009, and have had a much stronger recovery. Here's the StD info:

Daily Standard Deviation
1-Stars=2.12%
2-Stars=2.25%
3-Stars=2.18%
4-Stars=2.18%
5-Stars=2.07%
SPY=1.99%

I'll run the sector data an post soonish.  Yell at me if you don't see it within the next week.

Fool On!

Report this comment
#29) On September 03, 2009 at 11:35 AM, BravoBevo (99.96) wrote:

TMFJake, thanks again. If the chart is cumulative and is not against the S&P, then next time why not show the chart with one additional line representing the S&P?

Report this comment
#30) On September 03, 2009 at 11:51 AM, BigFatBEAR (29.02) wrote:

This reminds me of 2004, when the Democrats somehow mysteriously lost the primaries...

:)

Report this comment
#31) On September 03, 2009 at 12:00 PM, JakilaTheHun (99.91) wrote:

In other words, every single tier of rated stocks in the entire securities universe - from the highest-regarded 5 stars to the most despised 1 Star - all performed better than the S&P.

This is not surprising at all to me.  During the downturn, small caps and micro caps were absolutely decimated, whereas, large cap companies didn't suffer quite as greatly.  The major indexes like the Dow and the S&P, of course, consist of a lot of larger, well-known companies.  Whereas, the CAPS universe probably includes more small caps than anything.

This was actually one of the reason I was very bullish on small caps from November to April.  I green thumbed most of the small cap double long ETFs and red thumbed most of the small cap double short ETFs.

It is, of course, unlikely that this result will happen again unless the market nosedives and punishes the small caps the most again.

Report this comment
#32) On September 03, 2009 at 2:06 PM, TMFJake (30.83) wrote:

@BravoBevo:  The S&P, actually the SPY, is in the chart.  It's the orange line.

@BigFatBEAR:  Ha!

@JakilaTheHun:  Agree.  Certainly wouldn't expect to see something like this over any long term period (and we haven't), but for any time series of less than 12 months, wouldn't surprise me at all.

Report this comment
#33) On September 15, 2009 at 9:57 PM, CVMp (91.68) wrote:

The YTD #'s are simple wrong.  It is simply not possible for them to all be that much above the S&P.

Report this comment
#34) On December 30, 2009 at 7:32 PM, Tastylunch (29.22) wrote:

Can we get quarterly updates on this?

Report this comment
#35) On December 30, 2009 at 8:48 PM, portefeuille (98.47) wrote:

or daily ones?

Report this comment