← Quantery Blog

How to test operating leverage after ASML's results

October 15, 2026 · 7 min readoperating-leverageearningsbacktesting

Operating leverage appears when revenue grows faster than the operating costs needed to support it. The result is a wider operating margin: more operating profit from each sales dollar. ASML's latest quarter is a vivid example. The useful thesis, though, is broader: do companies with rising sales, expanding operating margins, and cash backing the earnings subsequently behave differently from companies with sales growth alone?

That question turns one earnings release into a market-wide test. It also avoids the easy mistake of treating one exceptional supplier as proof that margin expansion predicts returns. ASML supplies the prompt. Filed results from many companies have to supply the evidence.

What did ASML report this week?

ASML makes lithography systems used to print patterns on semiconductors. Its October 14 earnings release reported third-quarter net sales of €11.662 billion and net income of €4.960 billion. The filed financial statements put the year-earlier figures at €7.516 billion and €2.125 billion. On those figures, revenue increased 55.2 percent while net income more than doubled.

The middle of the income statement moved harder than the top. The same financial statements show gross margin rising from 51.6 percent a year earlier to 57.5 percent, while operating margin rose from 32.8 percent to 44.4 percent. ASML sold 121 lithography systems during the quarter, compared with 72 in the year-earlier period. Service and field-option revenue also rose, from €1.962 billion to €3.289 billion.

Management tied the demand to customers' expectations for the size and duration of AI-related investment. The October 14 release says customers were expressing interest in orders for 2029 and even 2030, and that currently planned supply for 2027 and 2028 wouldn't fully meet expected demand. ASML guided fourth-quarter sales to €14.2 billion to €15.2 billion and full-year sales to €44 billion to €45 billion.

TSMC provides a separate view from the manufacturing side. Its October 15 earnings release reported third-quarter revenue of NT$1.494 trillion, up 50.9 percent from a year earlier, with a 66.4 percent gross margin. Two suppliers in the same production chain reporting strong demand isn't an independent market sample. It does make the mechanism worth stating precisely.

Why did profit rise faster than sales?

Operating leverage is the change in profit that follows when revenue moves against a cost base that doesn't move at the same speed. A factory, engineering team, and service network carry costs before the next machine ships. Once that capacity is in place, another unit of revenue may require less than another full unit of operating expense. Some of the extra gross profit drops through to operating income.

ASML's filed figures show the shape. Revenue rose by €4.146 billion from the year-earlier quarter. Research and development expense rose from €1.109 billion to €1.174 billion, and selling, general, and administrative expense rose from €303 million to €353 million, according to the filed financial statements. Those two operating-expense lines increased far more slowly than sales. Operating income rose from €2.468 billion to €5.173 billion.

That arithmetic doesn't prove every euro of margin expansion came from fixed-cost absorption. Product mix, pricing, system configuration, service work, accounting timing, and foreign exchange can all change the margin. ASML said stronger demand for productivity upgrades helped lift installed-base revenue and gross margin. A filing gives the outcome; the business explanation still needs the notes and management discussion.

Operating leverage also runs backward. Capacity built for expected demand doesn't vanish when orders slow. Revenue can fall before payroll, depreciation, supplier commitments, and facilities adjust. The same cost base that magnifies profit during a ramp can magnify the decline later. Customer interest several years out isn't recognized revenue, and a capacity constraint isn't a guarantee that every prospective order becomes a shipment.

What belongs in an operating-leverage screen?

Start with three claims and keep them separate.

Sales support asks whether demand is visible in filed revenue. Compare the newest quarter with the same fiscal quarter a year earlier. An adjacent-quarter comparison can grade a seasonal business on its calendar. The method in trailing versus forward numbers explains why the filing sequence and publication date matter.

Margin expansion asks whether operating income grew faster than revenue. Calculate operating margin for both comparable quarters and take the difference. A higher margin is the footprint of operating leverage, although it can't identify the cause.

Cash backing asks whether reported earnings arrived with operating cash flow. Operating cash flow divided by net income is a rough conversion measure. It catches some cases where receivables, inventory, or other accruals make profit look stronger than the cash collected. The distinction is developed in operating cash flow versus net income.

The screen should exclude Financial Services. Interest is part of a bank's operating product, so an ordinary industrial-company operating-margin rule doesn't describe that business correctly. Industry comparisons still matter among the remaining companies. A chip-equipment supplier and a grocery chain don't share a useful margin baseline.

How can the thesis be written down?

The shape below is illustrative, and the bundled templates are the reference for exact fields. It was checked against the current Quantery thesis DSL, but the thresholds are proposed research parameters.

# "Operating leverage with cash backing" (illustrative)
params:
  min_history: 5
  sales_growth_floor: 0.05
  margin_gain_floor: 0.02
  cash_conversion_floor: 0.8

quality:
  flags:
    - { when: "history_periods < $min_history", flag: "short_history:{history_periods}" }

features:
  revenue_now: newest(revenue)
  revenue_prior: lag(revenue, 4)
  op_income_now: newest(operating_income)
  op_income_prior: lag(operating_income, 4)
  revenue_ttm: ttm(revenue)
  op_income_ttm: ttm(operating_income)
  ocf_ttm: ttm(operating_cash_flow)
  net_income_ttm: ttm(net_income)

  sales_growth: if(history_periods >= $min_history and revenue_prior > 0,
    revenue_now / revenue_prior - 1, null)
  op_margin_now: if(revenue_now > 0,
    op_income_now / revenue_now, null)
  op_margin_prior: if(history_periods >= $min_history and revenue_prior > 0,
    op_income_prior / revenue_prior, null)
  margin_gain: op_margin_now - op_margin_prior
  cash_conversion: if(net_income_ttm > 0,
    ocf_ttm / net_income_ttm, null)

criteria:
  sales_support:
    rules:
      - { when: "is_null(sales_growth)", score: 0, flag: no_sales_comparison }
      - { when: "sales_growth >= $sales_growth_floor", score: 2 }
      - { when: "sales_growth > 0", score: 1 }
      - { else: 0 }
  operating_leverage:
    rules:
      - { when: "is_null(margin_gain)", score: 0, flag: no_margin_comparison }
      - { when: "margin_gain >= $margin_gain_floor", score: 2 }
      - { when: "margin_gain > 0", score: 1 }
      - { else: 0 }
  cash_backing:
    rules:
      - { when: "is_null(cash_conversion)", score: 0, flag: no_cash_conversion }
      - { when: "cash_conversion >= $cash_conversion_floor", score: 2 }
      - { when: "cash_conversion > 0", score: 1 }
      - { else: 0 }

gate:
  mode: strict

The five-percent sales floor, two-percentage-point margin gain, and 0.8 cash-conversion floor are assumptions to test. They aren't facts about ASML or universal definitions of quality. Keep them round, save them before reading performance, and perturb them later.

The code also exposes a data limit. lag(field, 4) means four observed filing rows back. It doesn't prove that the rows are contiguous quarters. The history flag establishes a minimum count, but a filing gap can still spoil the comparison. Inspect period dates before treating a surprising margin jump as year-over-year evidence.

What would a backtest need to separate?

Don't jump from the full screen to a verdict. Run a small sequence with the universe, dates, rebalance policy, holding rule, and benchmark held fixed.

1. Positive same-quarter sales growth alone.
2. Sales growth plus operating-margin expansion.
3. Sales growth, margin expansion, and cash conversion.
4. The full version within industries, or with sector effects measured separately.

That sequence asks whether each added claim changes the evidence. If margin expansion adds nothing to the sales baseline, the operating-leverage story hasn't earned a vote. If cash conversion removes most qualifiers, inspect whether missing cash-flow data or industry structure did the filtering. Testing a complex thesis against a simple baseline covers the addition and removal method.

Use filing dates. ASML's filing index records a September 27 quarter end and an October 14 filing date. A simulated trade before that release has looked ahead. Use point-in-time fundamentals and enter after the information was public. Any resulting returns are hypothetical and exclude costs.

Then break the outcome apart. Compare mean and median event returns. Check whether one semiconductor cycle supplies most of the apparent edge. Remove one sector at a time. Test nearby round thresholds. Reserve an untouched period after the rules are set. A result that works only for a huge margin jump may be measuring rare boom quarters rather than a repeatable operating trait.

The test can't establish that AI spending caused later returns. It can establish something narrower: whether companies that converted sales growth into wider filed operating margins, with cash behind the earnings, behaved differently over the tested history. That's enough work for one thesis.

ASML's quarter gives the idea unusually clean lines. Revenue accelerated, expenses rose more slowly, and margins widened. Management described demand beyond current supply plans. Now take the company name out of the rule. Change the thresholds, compare the simpler versions, and make the pattern earn its place across filings the market had seen.

Want to try this on your own rules? Quantery is free for 14 days: the full app, no card required.

← All articles