Some well-known models under-perform AAII's published performance

Research
1

Some of the well-known ranking systems (Piotroski, Zweig, etc.) have published annual returns (CAGR) of 80-130% - see http://aaii.com/stockscreens/performance.cfm

However, the models in port123 yield much lower returns of 12-17% over 5-yr simulations 2005-2010 and 2001-2006. I tried using all stock universe >$5

Has anyone tried simulating the AAII models on port123? (requires an AAII membership)
Are these modelsdifferent from the ones in port123? What’s driving the significant difference in returns

thanks
Kiran

2

Hi Kiran,

A few points regarding the AAII screens:

The screen results are a guide. AAII states that they then review news items and decide whether or not a purchase is made. Stocks with negative news/events will not be purchased. So their exact picks and results will be difficult to reproduce.

The AAII database suffers from survivor bias - I use to subscribe to their screener CDs and can attest to this state. The historical results of many screens (especially the newer ones) are in sample. So this survivor bias may be driving earlier reported returns, especially in the value screens.

The cost of the AAII subscription is pretty low (about $25 as I recall). They provide the screen rules for each screen on the subscription. It’s worth the $25 for the info. The screener CD subscription is a bit more, about $175 a year with CDs mailed to you. I tried this years ago, but found that the better P123 ports outperformed the AAII screens (historical results) and also found I could not reproduce similar results real time.

3

The reason you get different results between the AAII pre-defined screens and the P123 or SS123 pre-defined screens is that they use different selection criteria.

For example:

– The AAII Piotroski 9 screen ONLY picks stocks that meet all 9 of the Piotroski criteria, while the P123/SS123 picks the top 15 stocks. Just recently, the AAII screen only had 1 stock for the month – HAST.

– The AAII Piotroski screens also add a price-to-book criteria and omit OTC stocks.

Right now, on the front page of SS123, it says the 5-year return of the predefined Piotroski is 108.6%. When I do a 5-year backtest on it, keeping 15 stocks, No OTC, rebalance weekly, I get 110.4%.

For the last 60 months reported on AAII, the Piotroski 9 shows 551% and the Piotroski 8 shows 122%.

If I do a 5-year Piotroski 9 screen on SS123 with the added Price/Book criteria, I get a return of 417% – in the ballpark of AAII’s numbers. However, instead of buying 15 stocks each period, it holds an average of only 2.2 stocks. Because it’s using an absolute criteria instead of a relative ranking.

Here are some items I posted on the AAII discussion forums comparing Piotroski screens:[quote]
When I compare StockScreen123 vs AAII from the end of 2005 to today, here’s what I see:

285.9% = AAII return
91.9% = SS123 return, predefined, 4-week rebalance, limit 15 per period
297.1% = SS123 return, w/1-week rebalance, adding no OTC and P/B, no limit
[/quote][quote]
I just tried the 4 differently monthly starting points on SS123 with 9-rated stocks only. From 2001 to present, $100 grew to:

$1551 on Weekly rebalance
$444 on Monthly cycle 1
$444 on Monthly cycle 2
$1504 on Monthly cycle 3
$529 on Monthly cycle 4
[/quote] [quote]
Here’s something scary – if I just use this criteria on SS123:

Universe(NoOTC)
Pr2BookQ>0.01
FRank(“Pr2BookQ”,#All,#Desc,#Previous)<=20

…and then backtest from 3/31/2001 - 4/16/2010 with weekly rebalance, then $100 grows to:

$973 when using all selected stocks – average of 1065 per week
$3728 when taking the top 100 of those using their Piotroski ranking
$5635 when taking the top 50…
$6586 when taking the top 20…
$14,969 when taking the top 10…
$17,542 when taking the top 5…
$25,556 when taking the top 3…
$930 when taking the top 1…

The 3-stock portfolio had a 63% drawdown over 2007 and 2008. But was up 539% in 2009. Year by year:

2010 = +78%
2009 = +539%
2008 = -42%
2007 = -36%
2006 = +27%
2005 = +62%
2004 = +136%
2003 = +243%
2002 = +134%
[/quote]

4

AAII buys ALL stocks that fall through their screen. If there are 250 stocks then they buy 250. If 3, then they buy 3. There is also a timing issue. They buy on the close and you can only buy on the open. There is no way you can replicate their performance. They confirm this is you talk to them. As said above, their performance is only meant to be a suggestion of what you might be able to do.