Initially, Altman worked with a cutoff of 2.675; scores below that were presumed to signify significant bankruptcy risk. The probability of accuracy was very high in the percentage of bankruptcies it predicted, but at the cost of too many (in Altman?s judgment) Type II errors (flagging firms as distressed that do not wind up filing bankruptcy). Accordingly, Altman recommends use of a cutoff of 1.81.
While privately-owned companies are not included in Portfolio123, we understand that we must necessarily depart from literal application of the model if we are to make it useful to assess stock prospects. Accordingly, users may want to consider this variation in cases where they believe the current market value of equity may be temporarily distorted and not indicative of the value the equity is likely to have over a longer term. Altman recommends use of a score of 1.23 or greater
Variables X1 through X4 are the same as in the original model. X5 is omitted in order to minimize potential industry effect, which can be troublesome as comparisons from one to another become less apples-to-apples. Altman recommends use of a cutoff score of 1.10.
Please read the documentation for more details and caveats.
The Z Score is really cool. If you build a value system for distressed stocks (falling prices, bad recommendations, deep value) - the Z Score really adds something substantial. Used on its own - it seems to add very little. So for those who like to ‘bottom fish’ - this is a valuable tool. Thanks!
Here is a simple ranking system described as above with equal-weight factors without Zscore on the Russell 1000 and it holds 25 stocks:
CAGR 19.64% MAX DD -75.11% Sharpe 0.55
This is with the Zscore factor added in:
CAGR 23.98% MAX DD -64.75% Sharpe 0.74
There’s actually a fourth variation created to evaluate emerging-market debt. We didn’t add that because it seems to be really far afield of what we do.
Please do read and take seriously the caveats in the documentation. I’m completely serious when I say this was not designed for use by equity investors. It’s not so much that the factors are deficient (they’re fine). It’s the way Altman researched and calibrated the model. When dealing with credit default risk, you accomplish nothing if you succeed in predicting bankruptcy with 95% accuracy if you also wind up with a lot of false signals. You may even come across literature associating Z score with cutoffs in the high 2s. Altman initially contemplated a 2.675 cutoff but lowered it, even at the known cost of increasing his misses on default calls, to reduce the number of false positives from what had been an exorbitant number. We don’t think this way in the stock market. We’re not wired into binary outcomes with only a tiny portion of a sample being of the negative sort.
So if you want to use this, you may need to be creative. For example, if you’re looking to pursue cigar-butt value, or perhaps very high equity yields, where all the companies you’re looking at are known dogs with your task being to weed out the worst in the kennel, that might be the sort of area where Z score might help.