No. of Recommendations: 31
HIAR: Higher Inventory and Accounts Receivable

In his 2003 book, Practical Speculation, Victor Niederhoffer mentions that increases in inventory and accounts receivable are bad news. To test this, I created a simple short screen that looks for large increases (>40%) in both and then sorts by low volume ratio (on the assumption that low relative volume is usually bad news).

The screen:
Price > 5
Price * Average Monthly Volume 3m/20 > 500
ADR = False
Exchange <> OTC
Industry <> Misc. Financial Services

Inventory Q1 > Inventory Q5 * 1.4
Accounts receivable Q1 > Accounts receivable Q5 * 1.4

Sort low VR = 100 * 10-d Volume/3-mon Volume

The results for a 5-stock monthly can be found in Keelix job 315607. The screen is quite consistent for a short screen. Most short screens do well during bear markets (no surprise) but then blow up in your face in years such as 2003 and 2009. Not so with HIAR, which returned 4% in 2003 and only lost -7% in 2009. This feature makes it a much more useful short screen, as one is not always sure when a bear market has ended.

For the four bear market years of 2000, 2001, 2002 and 2008 the screen had an CAGR of 79%. For the other years since 1997 the CAGR was 7%.

CAGR 24
GSD 43
Ratio 0.56
.
1997 9% 4mon
1998 - 2
1999 76
2000 93
2001 98
2002 120
2003 4
2004 10
2005 - 4
2006 -13
2007 - 8
2008 22
2009 - 7
2010 13
2011 22
2012 8 YTD

DB2
Print the post Back To Top
No. of Recommendations: 0
Dr. Bob,

Great screen. It would be interesting to see if an inversion using decreases in inventories and accounts receivable would lead to a solid long screen as well. If nothing else, it probably would make a good filter to find which companies to avoid.
Print the post Back To Top
No. of Recommendations: 3
So I guess you oughta call it ShortHiar?

Not like a hair shirt.

Jim
Print the post Back To Top
No. of Recommendations: 5
It would be interesting to see if an inversion using decreases in inventories and accounts receivable would lead to a solid long screen as well.

I'll check it out.

Of course, people might not want to put money on a screen called LIAR. :-)

DB2
Print the post Back To Top
No. of Recommendations: 0
Interesting screen. I decided to see how inventory and accounts receivable worked as a overall performance predictor.

For SIP stocks from 1997 till 10/2012, price > 3, 3m daily dollar volume > $500K:

Sorting into 20 bins using the factor (bsq.inv_Q1/bsq.inv_Q5)*(bsq.AR_Q1/bsq.AR_Q5)

The bins:
2.8 6.2 6.7 8.8 11. 5.6 8.7 5.6 8.8 6.1 6.2 4.2 2.5 2.1 1.8 1.6 0.1 -3.1 -6. -15

This is indeed a much better predictor of failure at the extreme lower 5% but not very good at
predicting a winner.

RAM
Print the post Back To Top
No. of Recommendations: 0
This is indeed a much better predictor of failure at the extreme lower 5% but not very good at predicting a winner.

I looked the the inverse (long) LIAR screen with a sort on high VolumeRatio.

Decrease
in Inv & AR CAGR GSD
<30% 11 51
<20% 11 45
<10% 14 48
< 0 22 48

Better than SPY but nothing too exciting.

DB2
Print the post Back To Top
No. of Recommendations: 15
Here's a GTR1 backtest (treated as a long screen) from 19970902 through 20121004 (sorry, I've been too busy working on the user interface to update the database for the last few days):

http://backtest.org/gtr1/2011/?aprc%281%29gt5:product%28av63...
Translation:
step0: [Actual closing Price; share_lag=1 days; quote_lag=1 days] > 5
step1: [[Actual Average daily Vol 63d; lag=1 days]*[Actual closing Price; share_lag=1 days; quote_lag=1 days]] > 500000
step2: [Security Type] != 30,31,36
step3: [Fama/French Industry Code] != 48
step4: [SI Inventory Q5] > 0
step5: [1*[SI Inventory Q1] - 1.4*[SI Inventory Q5]] > 0
step6: [SI Accounts Receivable Q5] > 0
step7: [1*[SI Accounts Receivable Q1] - 1.4*[SI Accounts Receivable Q5]] > 0
step8: [[Actual Average daily Vol 10d; lag=1 days]/[Actual Average daily Vol 63d; lag=1 days]] Bottom param0; Long, Cash When Fewer
Holding period = 20 mkt days; Fully rebalance every 1 periods

Here are the results when the 5-stock variant is given a weight of -1 (short) in a blend with 2 parts cash (which I can only guess is what Keelix.com does):

                Avg     Min      Max      SD
CAGR: 9.86 1.11 18.40 4.97
TR: 403.96 17.98 1162.36 327.58
GSD(20): 51.14 45.61 55.74 2.52
DD(20): 32.60 27.64 36.75 2.28
MDD: -67.66 -75.10 -56.66 4.65
UI(20): 32.92 24.31 41.63 5.46
Sharpe(20): 0.37 0.18 0.56 0.11
Beta(20): -1.45 -1.62 -1.28 0.08
TI(20): -10.60 -15.91 -5.18 3.12
AT: 10.99 10.84 11.21 0.11

http://backtest.org/gtr1/2011/blend.cgi?-1:aprc%281%29gt5:pr...

An average CAGR/GSD(20) of 10/51 vs 24/43 for Keelix.com. There seems to be enough of a difference between our short screen backtests that it's worth a serious investigation. I'll soon be allowing VL and SI Pro subscribers to get current screen picks and field values one way or another (if no one offers to help with some of the more sophisticated options for validating subscribers, I'll just go with something simple such as the MD5 hash of the latest installation file as a password), thereby opening the backtester up to more scrutiny.

In the meantime, people with more time than I have can (1) compare the long version of this screen at Keelix.com with my long version and (2) get daily portfolio values for the long version from my blender and verify that the daily portfolio values of the blend make sense. My backtester applies daily margin interest to the value of shorted positions and cash interest on the entire cash portion of the blend. If "Margin Premium" is set to 0, then those two interest rates are equal, and so in the case of a -1:2 blend, the entire portfolio is effectively earning cash interest on roughly its net value. I have no idea what Keelix.com does as far as interest (margin or cash) is concerned.

Robbie Geary
Print the post Back To Top
No. of Recommendations: 3
I added the Inventory and Receivables growth to my Stockscreen 1-2-3 implementation of Altmanz_Z short screen and used the lowest Altman_Z score as the final sort. I backtested it using 4-wk holds over the periods Mar2000-Mar2003 and Oct2007-May2009 (two long bear periods). A 1-stock version returned $550 and $470 respectively (starting with $100 each period) while the S&P returned ($39) and ($44) during the same periods. Max draw-down was 48%. It had returns in excess of the S&P for 34 of 61 periods.

A 4-stock version returned $502 and $225 respectively with a 30% max DD, while exceeding the S&P return for 35 of 61 periods.

Essence of screen:
Assets >$1M
Price > $6
3-mo. avg daily vol > 100k
US stocks only
Not in Finance or Utilitity sectors
Not in EATING industry
Altman_Z score < 1.23
Current Inventory > 1.4 * Inventory 4 quarters past
Current Receivables > 1.4 * Receivables 4 quarters past
Sort by lowest Altman_Z score and select 1
Print the post Back To Top
Advertisement