What Is Leverage?

November 19, 2026
5 min read

In a nutshell

Leverage means moving results many times larger with small capital or effort.

In poker terms, switching from $1 chips to $4 chips.

What else?

Your base strength does not grow one bit. But both wins and losses grow many times over.

What Is Leverage?

The word comes from lever. Leverage is what a lever does.

The word lever

A lever uses a pivot to lift something heavy with a small force.

In the money world, it means adding borrowed money to make a trade bigger than your own money.

At work, I think it has also come to mean borrowing tools, systems, or other people’s help to get more done than you could alone.

What decides the effect: your base strength

Your base strength: the force you put on the lever.

Zero times any multiplier is still zero. If your base strength is losing, the multiplier makes the loss bigger.

Leverage in stocks and FX

In the money world, the classic users of leverage are stocks and FX.

Both let you deposit money as collateral and trade many times that amount.

For stocks, it is called margin trading. In Japan, you can trade up to about three times your deposit (Japan Exchange Group, “Outline of the Margin Trading System” (Japanese)).

FX, foreign exchange margin trading, works the same way. In Japan, the cap for individuals is 25 times!

Leverage at work, and generative AI

At work, what you borrow is not money but tools and other people’s strength.

Today, the most talked-about tool of that kind is generative AI.

So who gets the most out of AI’s lever? The answer the research measured: the smaller your base strength, the more you grow.

A study of 5,179 call center agents found that with AI help, the number of issues resolved per hour rose 14% on average. The biggest gain, 34%, went to less experienced agents, while experienced agents barely changed (Brynjolfsson et al., “Generative AI at Work,” NBER Working Paper 31161, 2023).

In an experiment with 758 consultants, those below average at the start improved 43%, and those above average 17% (Dell'Acqua et al., “Navigating the Jagged Technological Frontier,” Harvard Business School Working Paper 24-013, 2023).

For now, it seems we know that AI helps even out the work, and plays a role in raising the floor of skill.

Still, in my own experience, the people who can truly work together with AI to get results are veterans!

One veteran, designing a database, put a few structures in key places that kept data flowing efficiently.

Those structures could not be measured by standard rules such as indexes or normalization.

The rest of the ER diagram (the blueprint of how the tables connect) was filled in by AI.

An excellent design was finished at high speed.

That runs opposite to the research averages.

And used where it does not work, it is not just useless. It makes losses bigger.

Takeaways

Leverage.

When a strong player uses leverage, they get stronger. That’s all.

Use it in the gambling world, or in real life: the question is which one is where you should really use it.

Knowing the word and knowing the reality are, after all, quite different.

Trivia

  • FX (foreign exchange margin trading): depositing money as collateral to buy and sell foreign currencies for many times that amount.
  • Margin trading: depositing money with a broker as collateral and borrowing money or shares to trade.
  • Equity capital: money a company holds without borrowing, which it does not have to pay back.
  • Index: in a database, a lookup table for finding the data you want quickly.
  • Normalization: splitting tables so the same data is not written more than once.
  • NBER (National Bureau of Economic Research): a US economics research organization that publishes working papers.

History

The beginning: moving the Earth with a lever

In the 3rd century BC, Archimedes of Greece worked out the principle of the lever.

He is said to have declared, “Give me a place to stand, and I will move the Earth” (Chris Rorres, “Archimedes: The Lever,” New York University).

Even a small force can move something heavy, given a pivot and a long bar. The lever in leverage comes from here.

Walls and turning points: borrowed strength makes losses bigger too

In 1920s America, more and more people bought stocks by paying about 10% of the price and borrowing the rest.

Borrowed money poured into stocks, and prices rose sharply. In 1929, that market collapsed (Federal Reserve History, “Stock Market Crash of 1929”).

In the 2008 financial crisis, banks borrowing too much became a problem again.

So an international rule (the leverage ratio) was made so that a bank’s capital does not fall below 3% of its total exposure (Basel Committee on Banking Supervision, “Basel III leverage ratio framework and disclosure requirements,” January 2014).

In Japan, a cap was put on the multiplier for individual FX: 50 times from August 2010, then 25 times from August 2011 (Financial Services Agency, Japan, July 2009 (Japanese)).

Use a lever wrong, and you end up under what you lifted.

What comes next

Money leverage has settled into living with caps.

Work leverage jumped in multiplier with generative AI. No one has set a cap yet.

Timeline

PeriodKey eventsKey people and organizationsCountry or regionNotes
3rd century BCThe principle of the leverArchimedesGreece (Sicily)Said to have declared, “Give me a place to stand, and I will move the Earth.” [New York University](https://math.nyu.edu/Archimedes/Lever/LeverIntro.html)
1929A market of stocks bought on borrowed money collapses—United StatesMore and more people paid about 10% of the price and borrowed the rest. [Federal Reserve History](https://www.federalreservehistory.org/essays/stock-market-crash-of-1929)
2010A cap on the multiplier for individual FXFinancial Services AgencyJapan50 times from August 2010, 25 times from August 2011. [Financial Services Agency](https://www.fsa.go.jp/news/21/syouken/20090731-6.html)
2014The bank leverage ratioBasel Committee on Banking SupervisionInternationalCapital must be at least 3% of total exposure. [BIS](https://www.bis.org/publ/bcbs270.htm)
2023A study of generative AI in a call centerBrynjolfsson et al.United StatesIssues resolved rose 14% on average, 34% for less experienced agents. [NBER](https://www.nber.org/papers/w31161)
2023An experiment with 758 consultantsDell’Acqua et al.United StatesOutside what AI does well, correct answers fell by 19 percentage points. [Harvard Business School](https://mitsloan.mit.edu/sites/default/files/2023-10/SSRN-id4573321.pdf)