Errata: Known Errors in "Foundations for Scientific Investing" and in the "Q&A Book"
13th Edition FFSI (February 2024) with ISBN 9781991155474
Errors/Corrections in February 2024 Edition
Section 2.5.1 (Dividend Timelines): It is not a mistake, as such, but a revision caused by changes in the marketplace. The U.S.
will move securities market settlement from T+2 to T+1 on May 28, 2024 (but there is no change for
U.S. government bonds or U.S. futures and options, which already trade T+1; most FX already settles T+1).
Canada moves to T+1 on May 27, 2024. India already made the shift to T+1 in 2023.
To the best of my knowledge, the U.K., Europe, Australia and N.Z. are remaining at T+2,
but are all under pressure to make the change and are consulting with stakeholders.
This affects the dividend timeline. After May 28, under T+1, the U.S./Canada ex-date and
record date will be the same. So, if you buy a stock the day before the ex-date, there will
be only one business day to get your name in the book of record.
Standard settlement is referred to as "regular way" settlement; after May 28, 2024,
"regular way" will mean T+1 in the U.S.
12th Edition FFSI (August 2022) with ISBN 9781991155450
Errors/Corrections in August 2022 Edition
p. 12. A simple typo. "34 bps" in the second-to-last paragraph should be "35.7 bps".
p. 175. "PERP's jump in share price will reflect only this retention-driven expansion of its capital base" should
read "PERP's jump in share price will reflect only this expansion of its capital base" (b/c there is no
earnings retention in this case).
p. 528. Word missing. Should be "or in companies that have dealings [with] countries with human rights violations"
7th Edition of Q&A Book (January 2021) with ISBN 9780995117358
Corrections in January 2021 Q&A Book Edition
I failed to mention that Q63 and Q113 (correlation for Stock S and Stock F) have a spreadsheet that you can download at the main Web page.
I failed to mention that Q285 and Q348 (Bachelier [1900] option pricing model) have a spreadsheet that you can download at the main Web page.
11th Edition FFSI (January 2022) with ISBN 9781991155429
Errors/Corrections in January 2022 Edition
p. 195: "is a equal-weight special case" should be "is an equal-weight special case".
p. 244: In the middle of the page, "82%" should be "72%".
p. 366: seven lines from the bottom, active share is described as "the sum of the absolute values"
but it should be "half the sum of the absolute values".
p. 528: Three lines from the bottom, in the footnote, "occurs" should be "occur".
pp. 74, 187, 285, 502, 698: the characters "textit" should be struck out. This is a software glitch.
10th Edition FFSI (December 2020) with ISBN 978-0-9951173-6-5
Errors/Corrections in December 2020 Edition
Some of these errors are the result of material being moved around during the latest revision.
So, some items have accidentally been mentioned too soon relative to where they are defined for
the first time. These errors will be fixed in either a revised 10th edition in late 2021 or in a new
11th edition in 2022.
p. 7 basis points are mentioned in passing (in the box) before being formally defined on p. 10 (in the footnote).
p. 18 mean blur is mentioned in passing before being defined. It is a common theme in the book, but is not defined until the middle of p. 68.
p. 20 the SLLN is mentioned but not defined until p. 53.
p. 29 says here are two more job interview questions. In fact, these are the first to appear in the book.
p. 147 says that SNR R_i is lognormally distributed. In fact, it is SGR=(1+R_i) that is lognormally distributed.
p. 157 "time line" should be "timeline".
p. 261 seven lines from the bottom says that vec H and vec G can be used to span the frontier.
This is incorrect. (vec H + vec G) and vec G can be used to span the frontier. vec H is not a set of portfolio weights, because they add to zero.
p. 275 says that X is referred to on p. 321, but it got pushed just over the page to the top of p. 322.
p. 341 "purchases" (9 lines into the second bullet point) should be purchase.
p. 343 10 lines from the bottom, half a sentence that was supposed to be deleted is still showing. Strike it out.
p. 419 "news watchers" should be "newswatchers" in two places.
p. 531 line 13: "a couple of pages from now" should be "at the end of this section"
p. 533 "a much a lesser" should be "a much lesser"
p. 535 "more profitable that the call option" should be "more profitable than the call option"
p. 536 "the Quant Quiz at the end of this section" should read "the Quant Quiz on p. 534"
p. 557 It is perhaps not clear enough that the GBM shown is the discrete-time version of the continuous-time version used
by Merton-Black-Scholes (this is, however, discussed on p. 379).
p. 635 citation to Schwed appears out of order, a half-dozen entries before it should.
9th Edition of FFSI of FFSI (November 2019; Revised December 2019) with ISBN 978-0-9951173-2-7
Errors/Corrections in December 2019 Edition
p. 29, bottom 10 lines contain repetition of words concerning continuously compounded returns and APRs.
p. 57, "An Kendall-Stuart..." should be "A Kendall-Stuart..."
p. 73. Top 10 lines contain repetition of words concerning 60/40 and bonds, including "bonds bonds" etc.
Scalar is misspelt as "scaler" on p. 246 and p. 514.
p. 142. "Second, calculated..." should be "Second, I calculated..."
p. 186. "...to predict future stock prices" should be "...to predict future stock returns",
or "...to predict future changes in stock prices." I made the case earlier that predicting near-term prices
is not where money is to be made.
p. 197. "In some case," should be "In some cases,"
P. 230. "...under the old T+3 settlement system, it was most likely
three business days before the record date: Wednesday June 25, 2003."
should read "...under the old T+3 settlement system, it was most likely
two business days before the record date: Thursday June 26, 2003." [Thank you to Henry Sullivan.]
p. 247. "lenging" should be "lending". Lenging is actually an old English word that means lengthening, but it is wrong here.
p. 256. Just after "lighting efficiency work", the word "work" is repeated.
p. 274. There is an error in the middle of the Sharpe ratio box. "The Sharpe ratio can be ex-ante
(i.e., \bar R_d is a historical average) or ex-post..." should read "The Sharpe ratio can be ex-post
(i.e., \bar R_d is a historical average) or ex-ante..." The rest of the box is correct.
p. 301 "ha dbeen" should be "had been".
p. 401, Fama-MacBeth section, "as dependent variables on the right-hand side of the second-pass regression"
should read "independent variables".
Errors/Corrections in November 2019 Edition of FFSI [All revised and corrected December 2019]
On p.xiv, "Bhabra, Gurmeet" should be "Gurmeet Bhabra".
On p.135, 10 lines from the bottom, estimated sigma sub l approximately 0.01
should be estimated sigma sub l approximately 0.163. That is, a daily sigma was accidentally used
where an annualized sigma was intended. So, R_G+(1/2)sigma_l^2 approximately equal to 9.831% is incorrect.
It should be 11.15%, which is not a small adjustment.
On p.175, "prof-forma" should be "pro-forma".
On p.411, the private equity slice is $5.0t, but that is 1.9% of the pie, not 1.0%. That's why the
pie is not quite complete. The missing slender white slice should be in the private equity section.
On p.417, "footnote 3.7.7" should be "Section 3.7.7".
On p.423 footnotemark 7 appears as an 8, giving two footnote 8s.
8th Edition of FFSI (January 2019) with ISBN 978-0-9951173-0-3
Errors/Corrections
In the box on p.262, it should say "Let \bar R_d be average R_d." and "Then \frac{\bar R_d}{\sigma_d} is a Sharpe ratio."
The distinction between R_d and average R_d is not clear as it stands.
7th Edition (September 2017) with ISBN 978-0-9941386-6-8
Errors/Corrections
On p. 56, the footnoted challenge should have power 3/2, not 2/3, in the denominator.
On p.156, footnote 22, the N.Z. dividend payment percentages have improved. As of May 2018, 71 out of 112 NZX All Share
index stocks pay dividends, 33 out of 112 do not pay dividends, and the other 8 stocks have missing data (using the Bloomberg
Professional Service). So at least 71/112=63.4% pay dividends, but no more than 79/112=70.5% pay dividends. My guess is that
the true proportion is roughly 65%, because the stocks with missing data are unlikely to be dividend payers.
Either 47 or 48 out of 50 (i.e., 94% or 96%) S&P/NZX50 index stocks pay dividends. There is some ambiguity because
one stock, CBL, has been suspended from the index, and although it paid dividends recently, it is not obviously going
to remain in the index or pay dividends again.
On p.210 , "...reinvested dividends accounted for one-half to two-thirds of the ending wealth of an S&P 500
investor over the 50-year 1963–2013 time period..." should read "...reinvested dividends accounted for 79% of the ending wealth of
a passive buy-and-hold S&P 500 investor over the 50-year 1963–2013 time period..."
Similarly, on p.414, "Remember from Section~1.1.2 that something like half to two-thirds of 50-year ending
wealth for an investor in S&P500 stocks was due to reinvested dividends" should read
"Remember from Section~1.1.2, reinvested dividends accounted for 79% of the ending wealth of
a passive buy-and-hold S&P 500 investor over the 50-year 1963–2013 time period..."
On p.234, "...the closer are A+/- sqrt{D}/C (the local..." should be "...the closer are (A+/- sqrt{D})/C (the local..."
On p.321, the box that says "Private Equity (Small Business)" should say "Small Business and Private Equity."
I meant private equity in the "equity in unlisted companies" sense, not in the "private equity fund" sense.
4th Edition Q&A Book (February 2018) with ISBN 978-0-9941386-7-5
Errors/Corrections
On p.145, the answer to Q374 should be $5,000 not $4,000. Your $1,000 margin allows you to bet
on $500,000 worth of assets. If they go up in value by 1%, your profit is 1% of $500,000 = $5,000.
On p.156, Q410 refers to a "growth alpha", but it should refer to a "glamour alpha" in order
to avoid the "bad value" definition of growth investing.
On p.164, Q437, Answer (a) should say "$8t and $4.1t, respectively"
On p.209, the answer to Q76 should be (e), not (d).
6th Edition (November/December 2016) with ISBN 978-0-9941386-2-0
Errors/Corrections
An error in the data download for the 50-year perfect foresight example means that some non-trading weekdays
appear in the data as trading days but with zero change in the index level. This causes an over-estimate of days
upon which the index return is zero, excluding dividends. There are flow-on effects that change the peakedness of
the plot in Figure 1.13, the counts of runs in Table 1.1, the counts of stock market moves in Table 1.7,
and the numbers in the extended kurtosis section (Section 1.3.12). These tables/figures/numbers have all been revised
and corrected for the 7th edition (out September 2017). The revised and corrected numbers tall the same qualitative story
in all cases.
On p. 75, "...in the second half of the year, the product of moments will almost always be negative." should instead say "positive."
On p. 78, 380.3 should be 382.3 in the Quant Quiz.
On p. 87, X bar on line 5 should be R bar, as in the first line of the equation following.
5th Edition (January 2016) with ISBN 978-0-9941182-9-5
Errors/Corrections
On p.53, in footnote 19, "numerator" should be "denominator."
On p.54, two-thirds of the way down the page, m_n should be m_k to match the power in the summation.
On p.56, the last sentence in the box is incorrect.
The t does display peakedness relative to a normal distribution with the same mean and variance. There are similar
mis-statments on 49, 313, 314.
On p.117, the middle paragraph starting "Is there a bias..." needs to be struck out because \bar R_G is not a continuously compounded return. The surrounding text is correct.
On p.120 "randomness in u" should be "randomness in z_n"
On p.198 I say that stock trades settle T+3 in NZ and the US. On March 7, 2016, however, the NZX and the ASX moved to T+2 settlement.
The US aims to have T+2 settlement in place by the third quarter of 2017. Note that this affects the dividend timeline in
Figure 2.13: the ex-date is now only one business day prior to the record date.
On p.234 the square root sign on Eqn 2.68 should not be there.
On p.325 on the third line "r_k>r_j>r_i" should be "k>j>i".
On p.368, second paragraph, "more than four times" should now be "about four times" because KiwiSaver AUM
grew to about NZD30b at the end of 2015, and the exchange rate at that time was roughly 1.5NZD per USD.
On p.420 in the second line of the bullet point strike out "market capitalization,"
as this is not correct.
On p.421 the two references to Fama and French 2015 are meant to be 2015a, though it makes little difference.
4th Edition (September 2015) with ISBN 978-0-9941182-2-6
Errors/Corrections
On p. 425, 488 Benjamin Graham's The Intelligent Investor is reported as
being first published in 1973, but it should be 1949.
On p. 177 it says that as of 2015 the NZX has no designated market makers. In fact, during 2015
the NZX introduced designated market makers for their SmartShares ETFs (Craigs Investment Partners as DMM),
and for the NXT market (which has a single stock, GGL, trading so far) (First NZ Capital as DMM).
On p. 394 the aside about Dow's 1986 index is incorrect. My description refers to the transportation
average, not the industrials.
3rd Edition (December 2014) with ISBN 978-0-9941182-0-2/978-0-9941182-1-9
Errors/Corrections
Page 6 Seven lines up from the bottom. "per capital" should be "per capita".
Page 22 Second line. "chariman" should be "chairman".
Page 28 Footnote, "2014" should be "2014a".
Page 99 8 lines down from top, differnt should be different.
Page 105. $336.675 should be $336.8425.
Page 150 middle of page "exceeds their ask price" should be "exceeds the mid-spread price".
Page 200 \delta_P appears twice in the middle of the page and it should be \delta_B in both cases.
Page 208 alpha=IT.omega should be alpha=IR.omega.
Page 228 "The upside potential remains, it is just at a high level of risk." should be
"The upside potential remains, at the same level of risk, but at a higher level of expected return."
Page 287 "As stock prices fall, so does the leverage in a market-value
balance sheet..." should obviously be "As stock prices fall, the leverage rises in a market-value
balance sheet..."
Page 323 seven lines up from the bottom "the fund not hold" should be "the fund does not hold".
Page 408 footnote 6 "does go through the vertex" should be "does not go through the vertex".
Page 55. Footnote 19 has been accidentally suppressed. It reads: "Ruppert (1987) provides an excellent discussion of kurtosis and the relative importance
of peakedness and fat tails individually and in concert. Kurtosis is not just about peakedness and kurtosis. For example, Banerjee, Dai, Lesmond, and Noe (2014) discuss a positive association
between kurtosis and liquidity costs."
Page 446. "Peter Rothchild" should be "John Rothchild"
The discussion in Section 2.3.6 (DDM III) is not clear. Yes, if a stock does not pay dividends, then a plain vanilla DDM that begins
with, say D0, and then grows it through time, will fail to model firm value. A multiple-stage DDM can, however, easily accommodate a stock that
does not currently pay dividends. There can be a first stage with no dividends, a second stage of transition, and a third perpetual stage
with average dividend payout. This model will be a model of growth in EPS, with different levels of dividend payout over time.
1st Edition of Q&A Book (December 2014) with ISBN 978-0-9941038-7-1
Errors/Corrections
Page 8 Question 16: the word "purchase" is missing from the end of answer (d), and as it stands answers (a), (b), (c), and (d)
all appear to be correct. So, answer (e) should read "None of the above is false."
Page 54 Question 140: as it stands answer (d) is the only one that is false. So, either answer (d) reading "6 pennies"
should be changed to read "5 pennies", or the solution should be changed from (e) (none of the above is false) to
(d) (which is false as it stands).
Page 98 Question 236: This question is internally inconsistent. Both (a) and (c) are correct answers. The argument is that
long-term investors recognize that they must face exposure to the broad equity market to build wealth. So, if we take that as given,
then we end up building a model where we attempt to actively step away from the benchmark only if we think the return justifies
the risk and the T-costs. That model (i.e., our active alpha optimization) does not include the benchmark return and risk as a
function of our choice variables because we take these actions regardless of, and not needing a forecast of, the direction
of the market.
2nd Edition (August 2014) with ISBN 978-0-9941038-4-0
Errors/Corrections
Typographical errors:
Page 1. ...would ask" should be ...would ask:
Page 104. middle of page. "a significant coefficients" should be "all significant coefficients"
Page 134. Four lines from the bottom, in Footnote 22, "TTM" is repeated.
Page 143. "ask if $10.05" should be "ask is $10.05";
Page 196. In Footnote 55 the word "average" is missing just before the first comma in the last sentence;
Page 370. "stocks returns" should be "stock returns" seven lines into Section 4.2.10
Page 438. Lee 1996 is out of order in the references;
Page 440. "Peter Rothchild" should be "John Rothchild"
Page 47: For improved clarity, six lines up from the bottom of the page, "Of course, the ratio can be quite large..." should
instead be "Of course, the ratio (chisquared_nu/nu)/(chisquared_eta/eta) can be quite large..."
The discussion in Section 2.3.6 (DDM III) is not clear. Yes, if a stock does not pay dividends, then a plain vanilla DDM that begins
with, say D0, and then grows it through time, will fail to model firm value. A multiple-stage DDM can, however, easily accommodate a stock that
does not currently pay dividends. There can be a first stage with no dividends, a second stage of transition, and a third perpetual stage
with average dividend payout. This model will be a model of growth in EPS, with different levels of dividend payout over time.
Page 226: Footnote 84 has disappeared in a software error. It should say "A collar is where you already own the stock, and you sell a high-strike call option,
thereby giving up the upside potential, and you use the proceeds to buy a low-strike put option, thereby protecting yourself from the downside risk.
Properly matched, it can be a zero-cost collar."
Page 259: a word is missing nine lines from the top of the page "...undervalued securities." should be "...undervalued (overvalued) securities."
1st Edition (January 2014) with ISBN 978-0-9941038-0-2 (Soft Cover) or ISBN 978-0-9941038-3-3 (Hard Cover)
Errors/Corrections
Page 50. "reflected about its mean" should be "reflected about its median" because the mean does not exist.
Page 55. "magnitude greater than about 2% occur more frequently" should be "magnitude greater than about 2.5% occur much more frequently"
and "magnitude less than about 2% occur less frequently" should be "less than about 1.5% occur more frequently" That is, there is extra
probability mass near the peak of the distribution (relative to a normal), and extra probability mass in the tails
of the distribution (relative to a normal).
Page 99. V(alpha) should be V(alpha hat), the variance of the estimator of the mean.
Page 143. "lift their quotes" and "lower their quotes" need to be transposed.
Page 176. "MIN-UTIL-OBJ.XLS" should be "MAX-UTIL-OBJ.XLS," and the reference to it in the
second to last paragraph on the page should say "...if you maximize utility by varying..."
The discussion in Section 2.3.6 (DDM III) is not clear. Yes, if a stock does not pay dividends, then a plain vanilla DDM that begins
with, say D0, and then grows it through time, will fail to model firm value. A multiple-stage DDM can, however, easily accommodate a stock that
does not currently pay dividends. There can be a first stage with no dividends, a second stage of transition, and a third perpetual stage
with average dividend payout. This model will be a model of growth in EPS, with different levels of dividend payout over time.
Typographical errors:
"140" should be "185" in the Job Interview Question box on p8;
"peak" should be "peek" two lines up from the bottom of p22;
"Simlated" should be "Simulated" in the graph title on p57;
the second "then one run of one tail" should be "then one run of one head" on p91;
"confident" should be "confidence" on 8th line of text, p93;
"$2,801,309.95" should be "$2,801,302.95" as carried over from Equation 2.12 on the previous page;
"ask if $10.05" should be "ask is $10.05" on p137;
"held in pension assets" should be "held in managed funds" on p297;
"though" and "exaple" should be "thought" and "example" five lines up from the bottom of p224;
WisomTree.com should be WisdomTree.com on p339;
"until about 2005" should be "until about 1995" on page 304;
Note that Equation 2.54, on page 193, is correct as stated in reference to Question B.2.2, but
more generally you need to introduce a scaling term in front of the stated T-costs term. Question B.2.2 assumes
that you rebalance only once per annum, but if we were rebalancing B times per annum, we would multiply the stated T-costs term
in Equation 2.54 by B (e.g., multiply the T-costs term by 12 if rebalancing monthly). That is because
the alphas and the VCV have been scaled to be in annual terms, and the T-costs need to be similarly
scaled so that they do not swamp, or get swamped by, the other terms in the objective function (Tajaddini, Crack, and Roberts, 2013).
Another way to think about this is as follows. Suppose you rebalance daily, but that the first term in Equation 2.54
uses annualized alpha, and the second term uses an annualized VCV. As you step through time, you are always looking at annualized
active return and annualized active risk, but if you keep the T-costs term as stated in Equation 2.54, then this is only the
T-costs for that day, the T-costs will be very small, and that T-cost penalty may as well not be there, because it
will be totally swamped by the active return and risk terms. Conversely, if we used alphas on a daily
horizon, and a daily VCV, then you could leave the T-costs term as it is in Equation 2.54. Either way,
the active returns, active risk, and T-costs need to be on the same time horizon scale.