So it’s not really a buy and hold philosophy, but HOLD HOLD HOLD over two or three major market dips over two to three years, and then eval not one year, but at year two or year three?
Nothing is mentioned about adding to the position during those market dips. This is not “playing the market” but adding to your position during those times normal folk would panic-sell. “Buy when others are fearful,” says the Oracle
Again, as you wisely explain, a quarterly or even annual eval may have already sold, skipping a longer-term value analysis, and not even considering adding to the position.
That’s what it boils down to, yes. However, I somewhat disagree with it, since this is a somewhat outdated view of investing. I think you evaluate every annual report to see if the thesis is still intact.
My opinion that the notion of value itself is dynamic and since move to information age the conventional measures like p/e, ev/ebitda were unable to catch it, hence the underperformance if this basis strategy in the last decade
The edge here is not the 40%. It is the mismatch between valuation horizon and career horizon.
Greenblatt’s own evidence is brutal: professionally managed accounts returned 84.1% over the first two years, while self-managed clients turned the same system into 59.4%, even trailing the S&P 500 at 62.7%.
The reason the anomaly survives is structural. A trade can be obviously cheap and still be unholdable for institutions if it may take 2–3 years to work while PMs are judged monthly, quarterly, or by redemption windows.
So the scarce asset is not information.
It is duration of conviction.
That is why some of the best “alpha” in public markets is really just the ability to survive looking wrong longer than everyone else can afford to.
Doesn’t Gotham offer its own index funds now that somewhat replicate these strategies, and have each underperformed the market?
It should be mentioned in the article that in the years following the publishing of “The Little Book That Beats The Market”, the Magic Formula hasn’t worked nearly as well as it used to.
Many thanks for the kind words and for bringing this info to my attention! Together we grow and learn, which I value a lot. I'm thrilled you enjoyed the piece nonetheless. :)
You're right! One of the fairest hits I got on this piece. Thank you.
GVALX has returned 11.18% annualized since inception, compared to 11.76% for the S&P. GSPY is 4.36% against 4.45%, and GINDX turned $10k into about $21k, while SPY did $23.8k.
On the formula pretty much the same story. Roughly 26% against 18% pre-2007, then 12% against 13% post-2010, and one backtest from 2014 to 2024 found not a single year of outperformance.
My fault for leaving this out! I did not actively think about it on a going-forward basis.
There's a scale mismatch in the strongest data point here — the self-managed vs. professionally-managed gap that supposedly proves the behavioral cost of human judgment is measured over just two years, but the piece itself says this needs a 3-5 year holding period to even judge properly. Two years is closer to the noise the strategy is explicitly built to survive than proof of anything permanent about the two approaches.
That said, the jelly bean jar framing is one of the cleanest explanations I've read for why prices and value diverge — worth the read on that alone.
You're right about the way it was constructed, and I'm grateful for the nice things you said about the jelly bean part.
The comparison isn’t showing the strategy’s returns. However, it is looking at two groups implementing the exact same strategy using the same list over the same two-year period in the exact same market. Anything that affected the approach during that time affected both groups equally, so it pretty much cancels out.
There are honestly two genuine weak points in the data, and I do not believe either A or B reveals the true situation.
Thank you for having read the piece carefully enough to notice this!
Fair, and worth conceding directly — pairing the same list against the same window does isolate the behavioral effect from macro noise, so that part holds up better than I gave it credit for.
What it doesn't settle, though, is whether the specific names skipped were actually right to be avoided over a longer horizon than two years — idiosyncratic risk doesn't always show up that fast. Is that one of the two weak points you had in mind, or something else entirely?
Appreciate you walking through the actual mechanics rather than just defending the number!
If the scary names carry tail risk that hadn't shown up in two years, then part of that 84.1% is compensation for risk still sitting in the portfolio rather than evidence of free money. Which would mean the behavioural gap is real but smaller than the headline number.
I don't think that kills it, since the mechanism Greenblatt described was people skipping names for how they felt rather than what they'd modelled. But it does mean the size of the effect is softer than I wrote it.
Adding this to my notes archive, though. You're takes are on point.
In any talk he gives, he mentions it. He gave a talk at Columbia University Graduate School of Business, where I heard it, but he also mentioned it in his Forbes talk 14 years ago. Just look up Joel Greenblatt interviews, and you’ll find it.
This sounds very similar to a method a friend of mine once used. He was not a financial genius, but was technically trained in stats. He said if you look at the market, any stock does up, moves sideways, or goes down. So he simply took the holdings of an index, ranked them by size and then matched the analyst projections of upside and downside for each. He then sorted them into quartile. He shorted the lowest quartile, sold covered calls on the next, and then ran we now call a concentrated portfolio on the other two. He did this once a year in December. As he said, “it’s just math”.
Hi- thanks for the clear and concise explanation- like a super compact, “The Little Book that Beats the Market for Dummies”
One question, if Greenblatt gave details about re-running the screen annually I must have missed it. I’ve heard he originally called for selling off any holdings whose updated financials put them outside the current top 1%, I’ve also heard he later amended those original annual refresh rules somewhat. You state he wants a hold commitment of at least 3 to 5 years. Can you enlighten me please?
Hi Randall, good catch. Two systems got blended together in my sentence.
The Magic Formula from the book holds each name for one year, then sells and replaces it with whatever the fresh screen throws up. Winners a few days past the twelve-month mark, losers a few days before, for tax reasons. Your recollection is right.
The three to five years is a commitment to the process, not to any individual stock. Greenblatt insists on it because roughly one year in four underperforms and people quit at exactly the wrong moment.
The daily-ranked chart in the piece is separate. That's Gotham's own valuation work feeding the long/short funds, not the book's screen.
Thanks so much for the reply and your correction- you’re absolutely right, Greenblatt included the annual refresh/replacement step in TLBTBTM. I was working from memory which isn’t as reliable as it once was (I keep forgetting). Very clear now, 3-5 commitment to the MF system, not individual holdings. Thanks again for the great summarization of one of investing’s “must read” classics.
Err, don’t try to replicate. I felt like I could focus on that one article, learn to implement it and not read others. Replicating it would dilute and damage this seeming clarity!
What made it seem different from many articles that offer a method? Partly the balance of combining/comparing three layers of analysis, partly the track record referred to, and partly specifying the difficulties - that it may take 3 years or more not 18 months before seeing the advantage, and how much worse investors do when making exceptions to the method. Overall it seemed credible and, conceivably, doable.
You’d think your many other fine articles would have a similar effect, but that’s what stood out to me.
Really helpful article. Thank you.
My pleasure! What aspect was the most helpful to you?
The notion that your own valuation matters more than popular opinion, as illustrated by the jelly bean story.
So it’s not really a buy and hold philosophy, but HOLD HOLD HOLD over two or three major market dips over two to three years, and then eval not one year, but at year two or year three?
Nothing is mentioned about adding to the position during those market dips. This is not “playing the market” but adding to your position during those times normal folk would panic-sell. “Buy when others are fearful,” says the Oracle
Again, as you wisely explain, a quarterly or even annual eval may have already sold, skipping a longer-term value analysis, and not even considering adding to the position.
That’s what it boils down to, yes. However, I somewhat disagree with it, since this is a somewhat outdated view of investing. I think you evaluate every annual report to see if the thesis is still intact.
My opinion that the notion of value itself is dynamic and since move to information age the conventional measures like p/e, ev/ebitda were unable to catch it, hence the underperformance if this basis strategy in the last decade
a forty percent return, in plain sight, they say,
yet lines of green gardens hide risks in the grey
if "effortless" wealth is the hook on the line~
the catch is usually your money, not mine~🧠
C’est quoi ce charabia
Try reading it multiple times or read some of his books, it will all make sense by then.
Great read
Many thanks, Fraser!
Ok good post. Can't disparage this.
Thank you. Pleased that this can not be disparaged. 😆
The edge here is not the 40%. It is the mismatch between valuation horizon and career horizon.
Greenblatt’s own evidence is brutal: professionally managed accounts returned 84.1% over the first two years, while self-managed clients turned the same system into 59.4%, even trailing the S&P 500 at 62.7%.
The reason the anomaly survives is structural. A trade can be obviously cheap and still be unholdable for institutions if it may take 2–3 years to work while PMs are judged monthly, quarterly, or by redemption windows.
So the scarce asset is not information.
It is duration of conviction.
That is why some of the best “alpha” in public markets is really just the ability to survive looking wrong longer than everyone else can afford to.
Exactly! And, if we start early, we can afford it to look wrong before it pays off.
Doesn’t Gotham offer its own index funds now that somewhat replicate these strategies, and have each underperformed the market?
It should be mentioned in the article that in the years following the publishing of “The Little Book That Beats The Market”, the Magic Formula hasn’t worked nearly as well as it used to.
Appreciate your openness and intellectual honesty and it’s a very interesting piece nonetheless!
Many thanks for the kind words and for bringing this info to my attention! Together we grow and learn, which I value a lot. I'm thrilled you enjoyed the piece nonetheless. :)
You're right! One of the fairest hits I got on this piece. Thank you.
GVALX has returned 11.18% annualized since inception, compared to 11.76% for the S&P. GSPY is 4.36% against 4.45%, and GINDX turned $10k into about $21k, while SPY did $23.8k.
On the formula pretty much the same story. Roughly 26% against 18% pre-2007, then 12% against 13% post-2010, and one backtest from 2014 to 2024 found not a single year of outperformance.
My fault for leaving this out! I did not actively think about it on a going-forward basis.
There's a scale mismatch in the strongest data point here — the self-managed vs. professionally-managed gap that supposedly proves the behavioral cost of human judgment is measured over just two years, but the piece itself says this needs a 3-5 year holding period to even judge properly. Two years is closer to the noise the strategy is explicitly built to survive than proof of anything permanent about the two approaches.
That said, the jelly bean jar framing is one of the cleanest explanations I've read for why prices and value diverge — worth the read on that alone.
You're right about the way it was constructed, and I'm grateful for the nice things you said about the jelly bean part.
The comparison isn’t showing the strategy’s returns. However, it is looking at two groups implementing the exact same strategy using the same list over the same two-year period in the exact same market. Anything that affected the approach during that time affected both groups equally, so it pretty much cancels out.
There are honestly two genuine weak points in the data, and I do not believe either A or B reveals the true situation.
Thank you for having read the piece carefully enough to notice this!
Fair, and worth conceding directly — pairing the same list against the same window does isolate the behavioral effect from macro noise, so that part holds up better than I gave it credit for.
What it doesn't settle, though, is whether the specific names skipped were actually right to be avoided over a longer horizon than two years — idiosyncratic risk doesn't always show up that fast. Is that one of the two weak points you had in mind, or something else entirely?
Appreciate you walking through the actual mechanics rather than just defending the number!
If the scary names carry tail risk that hadn't shown up in two years, then part of that 84.1% is compensation for risk still sitting in the portfolio rather than evidence of free money. Which would mean the behavioural gap is real but smaller than the headline number.
I don't think that kills it, since the mechanism Greenblatt described was people skipping names for how they felt rather than what they'd modelled. But it does mean the size of the effect is softer than I wrote it.
Adding this to my notes archive, though. You're takes are on point.
Where are your sources for any of this performance data?
In any talk he gives, he mentions it. He gave a talk at Columbia University Graduate School of Business, where I heard it, but he also mentioned it in his Forbes talk 14 years ago. Just look up Joel Greenblatt interviews, and you’ll find it.
This sounds very similar to a method a friend of mine once used. He was not a financial genius, but was technically trained in stats. He said if you look at the market, any stock does up, moves sideways, or goes down. So he simply took the holdings of an index, ranked them by size and then matched the analyst projections of upside and downside for each. He then sorted them into quartile. He shorted the lowest quartile, sold covered calls on the next, and then ran we now call a concentrated portfolio on the other two. He did this once a year in December. As he said, “it’s just math”.
It’s mostly maths and basic human psychology! I feel like your friend is someone we need to listen to a bit more; he’s onto something. ;)
Hi- thanks for the clear and concise explanation- like a super compact, “The Little Book that Beats the Market for Dummies”
One question, if Greenblatt gave details about re-running the screen annually I must have missed it. I’ve heard he originally called for selling off any holdings whose updated financials put them outside the current top 1%, I’ve also heard he later amended those original annual refresh rules somewhat. You state he wants a hold commitment of at least 3 to 5 years. Can you enlighten me please?
Randall
Hi Randall, good catch. Two systems got blended together in my sentence.
The Magic Formula from the book holds each name for one year, then sells and replaces it with whatever the fresh screen throws up. Winners a few days past the twelve-month mark, losers a few days before, for tax reasons. Your recollection is right.
The three to five years is a commitment to the process, not to any individual stock. Greenblatt insists on it because roughly one year in four underperforms and people quit at exactly the wrong moment.
The daily-ranked chart in the piece is separate. That's Gotham's own valuation work feeding the long/short funds, not the book's screen.
Thanks so much for the reply and your correction- you’re absolutely right, Greenblatt included the annual refresh/replacement step in TLBTBTM. I was working from memory which isn’t as reliable as it once was (I keep forgetting). Very clear now, 3-5 commitment to the MF system, not individual holdings. Thanks again for the great summarization of one of investing’s “must read” classics.
Randall
This is actually my favorite article so far.
I’m honored, thank you! What in particular made it your favorite? I’ll try to get more like it!
Err, don’t try to replicate. I felt like I could focus on that one article, learn to implement it and not read others. Replicating it would dilute and damage this seeming clarity!
What made it seem different from many articles that offer a method? Partly the balance of combining/comparing three layers of analysis, partly the track record referred to, and partly specifying the difficulties - that it may take 3 years or more not 18 months before seeing the advantage, and how much worse investors do when making exceptions to the method. Overall it seemed credible and, conceivably, doable.
You’d think your many other fine articles would have a similar effect, but that’s what stood out to me.
I will not try to replicate it, haha. Many thanks for the detailed explanation, I appreciate it.