Honest Horsepower: How the MMM Calculators Are Calibrated

Every engine simulator that came before us earned a nickname from the racers who used it — and the nicknames all mean the same thing: it reads high. Optimistic numbers sell software and flatter parts. They also cost real builders real money. So here's exactly how we calibrate our calculators, in public, so you can hold us to it.

The disease: multiplier stacking

Here's how simulators drift high. Give every part a little bonus — 2% for roller rockers, 2% for good pushrods, 3% for a windage tray — and each one sounds reasonable alone. Stack ten of them and you've invented 10–15% phantom horsepower that no dyno will ever find, because most of those parts don't add power at all. Springs, rods, and pushrods let an engine rev safely; they gate RPM, they don't push the curve up. In our tools, a part that only enables RPM adds zero horsepower by itself. That single rule kills most of the overestimating disease.

The cure: a ladder of anchors

Every MMM calculator is fitted against a ladder of published, verifiable dyno results — not one anchor, a ladder:

Rung one: factory ratings — the stock number, which is public and (post-1972, net-rated) surprisingly honest.
Rung two: per-unit dyno'd crate engines — BluePrint, GM Performance, and the like dyno every engine they ship; those sheets are the gold standard for a known combination.
Rung three: documented magazine and shop builds — Richard Holdener's step-by-step tests, Engine Masters' shootouts, Fuel Moto's same-bike cam comparisons — the same engine with one change at a time.
Rung four: the documented race ceiling — what the best-of-breed builds actually make per cubic inch, which is platform-specific and much lower than forums remember.

A calculator that hits the stock number but misses the crate engines is broken. One that hits the crates but doubles the race ceiling is broken in a more flattering way. The ladder catches both.

Not all sources weigh the same

Our source-quality order, strictest first: same-engine A/B tests beat per-unit-dyno'd crate engines, which beat magazine builds, which beat vendor claims — which run 5–10% optimistic as a class — which beat forum numbers, which we treat as rumors with enthusiasm. And even good dynos lie until calibrated: EC Carburetors published their own dyno reading a stock Predator 212 at 9.4 HP before recalibration and 6.7 after. When a number can't be cross-checked, it doesn't become an anchor.

We let the top end err low — on purpose

Calibration always leaves a little slack at the wild end of the range, where documentation thins out. We aim that slack downward. If our max-effort build reads 5% under the best documented result, a builder gets a nice surprise at the dyno. If it reads 5% over, they've bought parts chasing horsepower that was never real. Between those two errors, only one costs our users money — so we always take the other one.

What honest looks like on your screen

It means the stock build in each tool reads like the number on the box, not the number in the brochure. It means a slip-on exhaust shows you 2 horsepower because that's what slip-ons do, no matter what the ad said. It means every chart carries the same fine print — estimated, actual results vary — because your fuel, altitude, assembly, and tune all move the number. And it means when you compare two builds, the delta between them is the most trustworthy thing on the page, because calibration errors mostly cancel out of a comparison.

Why bother

Because the whole point of these tools is to spend your parts money once, in the right order. A flattering simulator is a slot machine. An honest one is a plan. We're a small shop network built by a working mechanic — honest numbers are the one advantage the big guys never bothered to build, and we intend to keep it.