Amusement Arcades OR Why The Best Businesses Might Be Hiding In The Cheapest Aisle,

The first part of this can be found here (https://searchfunder.com/post/just-buy-a-brewery) The series is working toward one thing: a valuation framework built from the ground up rather than borrowed from large- cap finance. This is the second brick. Here is where I think this leads, stated up front so you can argue with it as you read and in-between: one of the better places to look for an acquisition may be the corner of the market where EBITDA margins are healthy but valuation multiples are low. Not despite the apparent contradiction - possibly because of it. In my data, higher-margin industries tend to be negatively correlated with price, and that negative relationship might be the clearest sign that value is sitting there underpriced. I’d hold all of this loosely. It’s a pattern in one dataset, reinforced by some outside research, not a law. More importantly – this is a macro bird-eye vision, that has close to nothing to do with your next fluffed-up deal. But it’s a pattern worth taking seriously, and the rest of this piece is the argument for why. First, the uncomfortable thing about EBITDA I pulled most of the closed transactions I could get out of DealStats - sub- $20M deals, going back about as far as the database runs and pulled as recent as I could manage (N=7,###-###-#### and ran the correlations. The headline result is one that tends not to go over well in a finance classroom: across thousands of small- business sales, very little seems to predict the sale price except SDE. Here’s my raw correlation work, straight off the worksheet: And here it is cleaned up, every metric plotted by its correlation with sale price: Chart 1 - What correlates with sale price (MVIC) I wonder if you’ve noticed the very subtle product placement… Anyway - Two lanes appear to carry real signal; the rest are faded because they look, statistically, like close to nothing. SDE sits at about +###-###-#### the strongest relationship in the dataset, and my number lands close to the Marshall University dissertation that ran a similar test on 5,499 transactions. Raw EBITDA comes in around −0.15. Revenue, assets, operating margin all hover near zero. The likely mechanism is fairly mundane, which is part of why I trust it. Owner- operators generally don’t run their P&Ls to impress an acquirer; they run them to shrink the check they write the IRS in April. Salary gets maxed, the truck gets expensed, the spouse’s “consulting fee” gets booked. Across thousands of firms, unadjusted EBITDA may end up measuring tax aggression more than profitability. SDE tends to survive this because it adds the owner’s full compensation package back in - it’s closer to what a buyer-operator actually keeps. The Marshall study points the same way: it found SDE predicting microbusiness sale price more accurately and more stably than EBITDA, with EBITDA gaining ground mainly once a firm scales into real management. So we all know that applying an EBITDA multiple to a $300K- SDE HVAC shop is a mistake, or at least a worse approximation than SDE. But that’s somewhat table stakes. The part I find more interesting is what happens to margin. Also, there’s very little correlation between MVIC and EBITDA/SDE ratio – which is a great signal for all you business owners to cook your books. A negative correlation that might be pointing at the money When I collapse the data to industry averages, one relationship stands out - and it seems to run backwards from what most people assume. At the industry level, EBITDA margin correlates with sale price at roughly −0.67. On average, higher- margin industries appear to sell for less. That’s worth sitting with, because it’s close to the center of the whole argument. The market doesn’t seem to reward fat margins with fat prices. If anything, the relationship looks inverted. There are a couple of plausible reasons. Part of it may be mechanical: EV/Sales = EBITDA margin × EV/EBITDA, so in a mature sector where the revenue multiple is roughly anchored, a higher margin can mathematically push the profit multiple down. Dohmeyer and Kierulff documented something like this on small- business transactions years ago and cautioned against averaging multiples as if the relationship were linear. Part of it may be behavioral: the market might read a fat, mature margin as “this is roughly as good as it gets” and price the business more like an annuity than a growth story. I’d add an important caveat, though, because it would be easy to overstate this. Margin doesn’t seem to drive the multiple in any clean way: Chart 2 - What correlates with the valuation multiple When you ask what moves the multiple itself, the honest answer looks like: not much. My industry margin- to- multiple figure is a weak −0.16; the most pronounced published number I found, a 2026 study of 263 manufacturing deals that controlled for size, reaches only about −0.30; and the Grant Thornton / University of Warwick work on 2,500+ listed companies found close to no relationship at the firm level. Margin barely seems to touch the multiple, and as I already established in my earlier paper - neither really do default rates. So it may help to hold both observations together, because the gap between them is arguably where the opportunity lives: margin looks strongly and negatively tied to price (around −0.67), but only weakly tied to the multiple (around −###-###-#### One reading of that is that high- margin businesses tend to land at low multiples not because the multiple is punishing the margin, but because the market may be quietly under- pricing a whole category of solid businesses. If so, the mispricing has a recognizable shape - and you can go look where it tends to sit. Where it tends to sit: the blue- on- blue quadrant This is the picture the piece has been building toward. Shade every industry by its EBITDA margin and its multiple, and look for the corner where margin is high (blue) and the multiple is low (blue): Or graphically if that makes it easier, not sure it does though due to TMI for TLS ‘Too Little Space’. Or what I call the “Listening to sales- people” syndrome. And yes – this means that you might have a great friendship with whatever “Other Personal Care” means, and a break-up with small-time (<$20M) Computer infrastructure providers. Running the medians - industry margin around 20%, industry multiple around 2.4x - roughly 24 industries land in that high- margin, low- multiple quadrant. The standouts tend to be the kind of durable, owner- run businesses nobody brags about at a dinner party: • Offices of Physicians - about 26.6% margin, 1.93x • Computer Systems Design - about 26.0% margin, 1.95x • Tax Preparation Services - about 31.5% margin, 2.38x • Engineering Services - about 24.1% margin, 2.38x • Janitorial Services - about 23.4% margin, 2.19x Compare that to the other corner: Beer, Wine & Liquor at roughly a 10.5% margin commanding about 4.25x, or Industrial Supplies Wholesalers near 11.9% margin and 3.87x. Thinner margins, richer prices. If the market priced quality the way the textbook suggests, the table might shade the other way. In this dataset, it tends to shade this way. The reason these higher- margin businesses stay relatively cheap often seems to have little to do with their cash flow - it may be fragmentation, illiquidity, lack of scale, or simply being too small for institutional buyers to chase. Those look like structural discounts, and structural discounts are generally the kind a disciplined buyer can capture. Why the discount may be real money This isn’t a novel retail theory. It rhymes with the private- equity buy- and- build playbook: acquire sub- scale firms at low multiples, consolidate them into a platform that trades at a higher one, and capture the spread. Chart 3 - Valuation multiples in the wild The cheap- entry lanes tend to cluster around 2–4x. The platform premium - per GF Data, companies above $10M EBITDA - sits closer to 8.1x. The gap between them is the arbitrage, roughly to scale. An SSRN study of 161 buyouts puts the “add- on sourcing effect” at something like 8% of equity- value CAGR, and reportedly exit buyers don’t clearly distinguish acquired EBITDA from organic. The spread looks real and reasonably bankable. That said - three caveats, and they’re where I’d expect people to get hurt. One: I wouldn’t bank on the multiple re- rating. Chart 4 - What drives PE value creation Across PE returns, revenue growth tends to do most of the lifting, and operational margin work increasingly leads - KPMG’s 2025 survey of 500 firms has 64% ranking margin growth as their top value driver. Multiple expansion is only about a third of value creation in Gain.pro’s breakdown, and it’s the most market- dependent lever - the one nobody really controls. If a thesis depends on selling at a higher multiple later, that feels closer to a hope than a plan. Two: a low multiple can be a price tag on a problem. This is the value- trap risk, and EBITDA can be the bait, because it leaves out capex, working capital, taxes, and interest - a lot of the reasons a business runs short of cash. A 30%- margin business that has to redirect 25% of revenue into maintenance capex just to hold position might have only a 5% free- cash- flow margin. For example, The UK outsourcer Mitie reported £27.9M of positive EBITDA in one half of 2017 alongside negative £6.7M of operating cash flow over the same stretch, as working capital absorbed it. The multiple looked cheap; the cash position told a different story. Three: peak- cycle margins can be misleading. A cyclical caught near the top might show high-margin-plus-low-multiple that looks like free money - until the cycle turns, margins compress, and the acquisition debt stays exactly as cyclical as it was (which is to say, not at all). So the blue-on-blue quadrant probably rewards a careful second look at what kind of business is sitting there. A physician’s office or a CPA shop may be cheap mainly because it’s sub-scale and owner-dependent - arguably a structural discount you can work on by professionalizing and consolidating. A capital-intensive manufacturer at the same headline margin might be cheap for reasons that follow you home. Same quadrant on the table; potentially very different outcomes. What you might do with this If you’re an SBA buyer, I’d gently set the flip fantasy aside. The edge here probably isn’t the exit multiple - it’s the entry. Buying a genuinely high- margin business at a sub- scale multiple tends to make the day- one cash-on-cash math more forgiving. As a rough illustration: $300K SDE at 2.5x is a $750K deal, perhaps ~$675K of debt and ~$105K of annual service - a DSCR somewhere near 2.85x against a 1.25x floor. In that framing, the fat margin isn’t there to be sold later; it’s there as a buffer when something goes sideways, and as relatively cheap, durable loan coverage. Multiple expansion becomes a free option rather than a required outcome. The hunt, in practice: I’d start in the low-multiple, decent-margin corner of the table - that seems to be where the under- priced quality clusters - and then, within it, screen hard for businesses where the cheap multiple looks like a function of being small, fragmented, and boring rather than a function of capex or a cycle about to turn. Interrogate free-cash-flow conversion, not just the EBITDA line. The negative margin-to-price correlation may be telling you the good businesses are sitting in the cheap aisle. Diligence is what separates cheap- because-overlooked from cheap-because-broken. The shorter version SDE seems to be the metric that tracks price; EBITDA is closer to what the seller told the IRS. Not much appears to move the multiple. And the higher-margin businesses look negatively correlated with price, which may mean some of the better ones are sitting at low multiples in the cheaper aisle of the market. I don’t think that’s a paradox to explain away so much as a place to start looking. The spreadsheet doesn’t have opinions. It has a +0.78, a −0.67, and a table that shades in a direction worth a second look. It’s a macro overview in a micro world, and if you think that for this reason, this analysis might not be helpful at all, who am I to argue with you? Next in the series: leaving the ETA world - what seems to change once you climb above $20M in the private markets.