Earnings Accruals

From what I have been reading it is more valuable to screen for earnings accruals based with earnings in the denominator. This gives you much bigger liquid companies to choose from after screening as dividing by assets and also yields bigger alpha.

Would this formula work for selecting companies where cash flows have been higher than earnings over the last twelve months? I’m looking for companies which would get an unexpected boost to upcoming earnings as they have been raking in real cash higher than their posted earnings.

(NetIncBXorTTM-CashFITTM)<0

It could also be divided by earnings and used just to screen out the biggest offenders where earnings are being overreported by management.

(NetIncBXorTTM-CashFITTM)/NetIncBXorTTM<0.5 <— for example

So far these screens are giving me good alpha, but with huge draw downs and hits to Sharpe ratio…

There are a lot of ways to detect accounting anomalies in the form of accruals. Basically, accruals are the differences between accounting and cash-based metrics. Accruals exist to match up revenues with expenses and let the tax man know how much he is due. Most of the time accruals are benign. While accountants are usually unintentionally bad at matching up accounting and “true” economic value, sometimes aggressive accounting will cause a firm to intentionally overstate its income. To the extent that a firm exercises conservative accounting metrics, accruals will tend to be small. Also, firms which understate earnings tend to pay less taxes and therefore free up more cash for capital investors. The bottom line is that cash flows tend to be a better measure of economic value than GAAP earnings. Why the earnings-centric culture in Wall Street? Who knows?

I don’t think your formula will detect most forms of accruals. Were you aware that your formula is the functional equivalent of the following:
(DivPaidTTM + DepAmortTTM)
since P123 defines CashFl() as net income plus dividends and D&A? Perhaps you meant to use OperCashFlTTM()?

In any case, there are lots of ways to do this. Below are a few ways I do it. Method 1 measures the differences in non-cash net-assets (equity) from one period to another. Method 2 compares net change in cash versus the change in cash before capital investors.

  1. ((AstTotQ - CashEquivQ - LiabTotQ + DbtSTQ + isNA(TxPayableQ,0)) - (AstTotPYQ - CashEquivPYQ - LiabTotPYQ + DbtSTPYQ + isNA(TxPayablePYQ,0)) )
    / SalesTTM

  2. (NetIncCFStmtTTM - (NetChgCashTTM - (DbtLTIssuedTTM - DbtLTReducedTTM) + (EqPurchTTM - EqIssuedTTM) + (DivPaidTTM-PfdDivTTM) ) ) / SalesTTM