Just the thing to read before bedtime (not!)

[monabri]

I signed on free 14 day trial on "Stockopedia" and I naturally wanted to review favourite HYP stocks. One of the measures on Stockopedia was the following "Altman Z-score" described below:-

" The Bankruptcy Risk Meter gives a visual representation of how likely a company is to head into serious financial difficulty within the next two years.
The meter is derived from the Altman Z-Score, a statistical bankruptcy indicator generated from a set of balance sheet ratios. If the meter is in the safe zone (*indicating a Z-Score greater than 3*) the financial health of the company is good, whereas a company is in serious trouble if the meter is in the distress zone (*indicating a Z-Score less than 1.8*). Tests have shown that the Distress Zone is 80-90% accurate in predicting bankruptcy."

Results for some of our favourites
NG =1.64
GFRD = 3.99
SSE = 1.37
TW = 8.62
BMS = 3.4 (ok, this is one of my non HYP shares - divvi hunting!)
ITV = 5.4
VOD = -1.64 (negative)
GNK = 1.79
RMG = 3.13
MKS = 4.37
MARS=0.56
BVIC =1.85
EZJ = 3.16
BP = 1.11
PSON = 1.32 (just thought I'd throw that one in!)
ADM= ?
GSK = 1.22
CLLN = 1.45
AV and SL = not calculated

Sleep tight.....



(Hopefully you guys will have seen this all before and not be too concerned....I hope!)

[monabri]

sorry I signed up!

[flint]

Monabri

Virtually all "screeners" produce results which at best are only a starting point for further investigation.

The description of the Altman Z score and the claim that it is 80-90% accurate should be taken with a pinch of salt.

For example, the inclusion of NG. and SSE in a list of companies likely to head into financial difficulty within 2 years is at the least highly contentious.

Their balance sheets look stretched ( always have, always will ) as their business involves involves borrowing vast sums to finance vast capex. The profits are a small % of both revenue and capital employed. Whilst the profit is limited by regulation, it is also effectively guaranteed by that same regulation.

SSE happens to be my largest holding, first acquired when Scottish Hydro was privatised, added to over the last 40 years. I am sure it will have had a sub 1.8 ALt. Z score throughout.

I will indeed sleep tight tonight.

Jon

[Gengulphus]

The description of the Altman Z score and the claim that it is 80-90% accurate should be taken with a pinch of salt.

Any description of a test as having a percentage accuracy, with no further description of what percentage is being measured, needs to be taken with shovelfuls of salt!

For example, suppose that among a group of 1000 companies, 100 have Z scores below 1.8 and 900 above 1.8. And that 9 of the companies with scores below 1.8 and 1 of the companies with scores above 1.8 go bust in the following two years.

Person A points out that 9 out of the 10 companies that went bust had Z scores below 1.8, and so claims that the 'Z score below 1.8' test is 90% accurate at predicting bankruptcy.

Person B points out that 9 out of the 100 companies with Z scores below 1.8 went bust, and so claims that the 'Z score below 1.8' test is 9% accurate at predicting bankruptcy.

Which is correct? Without a proper definition of a percentage accuracy, an unanswerable question - and even with such a definition, there are lots of opportunities for misunderstandings unless you, the person who produced the figure in the first place, and every intermediary in the chain of communication between you and the original producer understand and agree on the definition. There probably is a technical statistical definition, by the way - but the chances of everyone involved understanding and agreeing on it seem rather low to me!

Gengulphus

[Wizard]

The description of the Altman Z score and the claim that it is 80-90% accurate should be taken with a pinch of salt.

Any description of a test as having a percentage accuracy, with no further description of what percentage is being measured, needs to be taken with shovelfuls of salt!

For example, suppose that among a group of 1000 companies, 100 have Z scores below 1.8 and 900 above 1.8. And that 9 of the companies with scores below 1.8 and 1 of the companies with scores above 1.8 go bust in the following two years.

Person A points out that 9 out of the 10 companies that went bust had Z scores below 1.8, and so claims that the 'Z score below 1.8' test is 90% accurate at predicting bankruptcy.

Person B points out that 9 out of the 100 companies with Z scores below 1.8 went bust, and so claims that the 'Z score below 1.8' test is 9% accurate at predicting bankruptcy.

Which is correct? Without a proper definition of a percentage accuracy, an unanswerable question - and even with such a definition, there are lots of opportunities for misunderstandings unless you, the person who produced the figure in the first place, and every intermediary in the chain of communication between you and the original producer understand and agree on the definition. There probably is a technical statistical definition, by the way - but the chances of everyone involved understanding and agreeing on it seem rather low to me!

Gengulphus

"In its initial test, the Altman Z-Score was found to be 72% accurate in predicting bankruptcy two years before the event, with a Type II error (false negatives) of 6% (Altman, 1968). In a series of subsequent tests covering three periods over the next 31 years (up until 1999), the model was found to be approximately 80%–90% accurate in predicting bankruptcy one year before the event, with a Type II error (classifying the firm as bankrupt when it does not go bankrupt) of approximately 15%–20% (Altman, 2000)."
Source: Wikipedia

Based on this it appears to be Person B's approach to calculation that Stockopedia are referencing.

Terry.

[Gengulphus]

"In its initial test, the Altman Z-Score was found to be 72% accurate in predicting bankruptcy two years before the event, with a Type II error (false negatives) of 6% (Altman, 1968). In a series of subsequent tests covering three periods over the next 31 years (up until 1999), the model was found to be approximately 80%–90% accurate in predicting bankruptcy one year before the event, with a Type II error (classifying the firm as bankrupt when it does not go bankrupt) of approximately 15%–20% (Altman, 2000)."
Source: Wikipedia

Based on this it appears to be Person B's approach to calculation that Stockopedia are referencing.

I'd have said that it appears from it that Wikipedia is confused! It first indicates that a Type II error is a false negative, i.e. the test producing a negative result when it should have produced a positive result. In the context of predicting bankruptcy, that means a case where the test indicated that the company was not going bankrupt but it did in fact go bankrupt.

And it then indicates that in the same context of predicting bankruptcy, a Type II error is the test classifying the company as bankrupt - i.e. predicting that it will go bankrupt - when it does in fact not go bankrupt. I.e. exactly the opposite type of error to the type it first indicates!

Gengulphus

[monabri]

So, as I understand it....
The test is "will the company go bust within x years"?
If the company goes bust - the hypothesis is correct.
However, the false negative occurs when the test predicts failure but the company survives - the opposite (negative) happens.
Yes, there are assumptions and there are statistically inaccuracies ( degree of confidence used in the calc).

But assuming that the guys who crunch the numbers are better statisticians than me ( not difficult!) Then the Altman Z (or whatever flavour as it is dependent on the business each with its own limits) - it will be interesting (!) To look back in a couple of years from now at some of the companies. I guess VOD is the one that sticks out for me.

I have started to review high yield companies which have "good ratings" and good yields - I'll report back tomorrow and add to the initial list above.

[Wizard]

I also think it would be interesting to use the data from a couple of years back to see if many of the HYP usual suspects you list were expected to be bankrupt already.

Terry.

[monabri]

OK, Here's an update which includes some shares which we would consider HYPable (!) and others which you definitely would not (eg Just Eat - JE.).Where I've entered the double asterix, is where I know there to be current "issues".

Ticker	z	F'Cst Yld (%)
JE 	32.00	~0
3IN	24.70	4.1
VCT	14.50	5.7
PSN	11.50	5.7
TW	8.62	8.2
BWNG	7.60	6.4
ITV	5.40	4.0
ELM	4.92	4.9
MKS 	4.37	5.4
HFD	4.34	4.6
GFRD	3.99	7.0
MNDI	3.77	2.9
BMS	3.40	8.9
BLND	3.40	5.1
EZJ	3.16	4.1
RMG	3.13	5.8
TATE	2.96	4.2
AAL	2.89	1.8
WMH	2.72	4.7
G4S 	2.61	4.0
BATS	2.58	3.8
CRH	2.24	2.5
CCL 	2.17	2.1
PFC 	1.84	5.8
GNK	1.79	4.9
IMB	1.66	4.8
NG	1.64	4.9
SIV	1.64	**
UU	1.54	4.5
COB	1.50	**
CLLN	1.45	8.1
SSE	1.37	6.2
PSON	1.32	**
IRV	1.19	7.2
CNA	1.15	5.6
CPI	0.90	6.3
IAG	0.77	4.5
SBRY	0.76	3.9
SGC	0.60	5.8
DEB 	0.60	6.4
MARS	0.56	5.6
TSCO	-0.06	**
CNCT	-0.09	6.4
VOD	-1.64	6.0

[Wizard]

Not sure if Stockopedia retains historic Altman Z scores, but I just had a quick rough and ready go at calculating it. Not perfect as I am not 100% happy in my dissection of the numbers for all parts of the calculation, but I calculate as of 31/3/12 Vodafone had an Altman Z of -1.6. So basically no different to now.

One big negative driver of the Altman Z for Vodafone as far as I can see is parts B, Retained Earnings divided by Total Assets. I guess there are few companies with negative £84bn of retained earnings. But the biggest driver is Market Value divided by Total Liabilities. This part of the calculation confuses me a bit. In short few liabilities and a very high valuation will drive a very big negative component for a companies Altman Z score, my first thought is that surely a smaller level of liabilities is a good thing.

Terry.

Moderator Message:
Should this be part of strategy board as we are deviating from HYP here. Raptor.

[monabri]

Well, that's blown that theory with BMS with a Z of 3.40 !!

http://www.londonstockexchange.com/exch ... 02909.html

That hurts!

[Wizard]

Not sure if Stockopedia retains historic Altman Z scores, but I just had a quick rough and ready go at calculating it. Not perfect as I am not 100% happy in my dissection of the numbers for all parts of the calculation, but I calculate as of 31/3/12 Vodafone had an Altman Z of -1.6. So basically no different to now.

One big negative driver of the Altman Z for Vodafone as far as I can see is parts B, Retained Earnings divided by Total Assets. I guess there are few companies with negative £84bn of retained earnings. But the biggest driver is Market Value divided by Total Liabilities. This part of the calculation confuses me a bit. In short few liabilities and a very high valuation will drive a very big negative component for a companies Altman Z score, my first thought is that surely a smaller level of liabilities is a good thing.

Terry.

Moderator Message:
Should this be part of strategy board as we are deviating from HYP here. Raptor.

Raptor, not sure why this is off topic if the rest of the thread is OK. All I am trying to do is make a comment that whilst VOD may seem a risky HYP choice given its negative Altman Z score, it had a negative Altman Z in the past so maybe that should not be as concerning as on the face of it it seems.

If the whole thread is off topic fine, but unclear to me why my post is being singled out.

Terry.

[Arborbridge]

This my be a bit dry and technical, but considering whether your company is likely to go bankrupt is hardly off-toic, in my view. It's another safety factor akin to discussing cash flow to equity or net gearing.

Depressing to see more than a sprinkling of HYP shares below the supposed safety area. I believe that's only to be expected as HYP shares are those under the cosh for some reason.


Arb.

[Raptor]

Arb,
Not disagreeing with the argument just that it isn't restricted to HYP shares and the lists that are appearing contain shares that have not and do not meet HYP values, IMO? But am open to leave alone.

Terry, the mod box was just a heads-up about the way the topic was going not about the particular topic, sorry for confusion.

Raptor.

[Arborbridge]

Arb,
Not disagreeing with the argument just that it isn't restricted to HYP shares and the lists that are appearing contain shares that have not and do not meet HYP values, IMO? But am open to leave alone.

Terry, the mod box was just a heads-up about the way the topic was going not about the particular topic, sorry for confusion.

Raptor.

Agreed, I'm with you there - a little judicious editing before posting (to weed out irrelevant shares) would have solved the problem and not attracted criticsm

[monabri]

I posted this with the hope of giving another practical tool in our armoury. We are all familiar with looking at P/E and div cover etc but I introduced a risk term (for discussion).

I deliberately included several stocks that one would consider as "non HYP" just to give a broader picture as "risk" is an element for consideration in all companies.

I did include many (50% perhaps) that often find their way into HYP portfolios (TW/SSE/NG/GFRD/VOD RMG/BATS/PSON/COB etc).

After today's events on BMS (I hold as non HYP share) I have come to the conclusion that maybe I'm using the measure incorrectly or interpreting it wrongly!

It may be the way to view the measure is that although "BMS is sailing in stormy waters (pun intended) at the moment it is expected to be robust enough to weather out the storm"?


I mention BMS as it still stings (yes, I know it isn't considered as HYP type) But it could equally have been RMG/PSON/COB...

[Wizard]

Terry, the mod box was just a heads-up about the way the topic was going not about the particular topic, sorry for confusion.

Raptor.

No problem Raptor, understood. As ever your work as a moderator is appreciated.

Terry.