Advanced3 min read

Modelling Transaction Costs

The gap between a backtest and a live result is usually costs. They are certain, they scale with turnover and size, and they are the most commonly omitted input.

MadStockAlerts Research · Updated August 29, 2026

What to take away

  • Commission is the smallest and most visible component.
  • Spread and market impact are larger and are not itemised anywhere.
  • Impact grows with order size relative to the security's depth.
  • A strategy's capacity is set by the size at which impact consumes the edge.
  • Costs scale with turnover, which is a design parameter of the strategy.

MAD Academy Training Video · 0:47

The Line That Kills Most Strategies

Modelling costs properly is what separates a backtest from a plan, and impact grows faster than order size.

This lesson is part of a Stock Alerts + Tools plan.

See the library

The components

ComponentCharacterTypical treatment in a naive test
CommissionExplicit, often near zeroSometimes included
SpreadHalf the quoted spread per sideFrequently omitted
Market impactGrows with size relative to depthAlmost always omitted
Delay costThe move between the signal and the fillAlmost always omitted
FinancingMargin interest and borrow feesOmitted unless leverage is explicit
TaxesDepends on the account and the holding periodAlmost always omitted

The middle rows are where the money goes. A test using closing prices with no spread and no impact is assuming a fill that nobody receives, and the assumption gets worse as the strategy trades more or larger.

Impact, and how it scales

Market impact is the price movement caused by the order itself. Empirical work has repeatedly found it grows with the square root of order size relative to typical volume, rather than linearly.

impact = k x volatility x sqrt(order size / average daily volume)

  • k is a constant estimated from execution data
  • the square root means doubling the size costs less than twice as much per share, and still costs more in total

The relationship is why capacity is a real constraint rather than a theoretical one. Every strategy has a size at which impact consumes the edge, and it is knowable in advance rather than discoverable only by trading it.

Turnover as a design parameter

A strategy's cost burden is its per-trade cost multiplied by how often it trades. Because the second is a design choice, cost is something a strategy is built with rather than something it encounters.

Annual turnoverAt 20bp round tripEdge required before any profit
25%5 basis pointsNegligible
100%20 basis pointsSmall
400%80 basis pointsSubstantial
2000%4 percentVery large

The last row is where many signal-driven strategies operate, and it is why an apparently strong signal frequently produces nothing after costs. The signal has to beat four percent a year before it earns anything at all.

The edge a strategy has to clear before it earns anything
The edge a strategy has to clear before it earns anything0%1%2%3%4%Where many signal-driven strategiesoperate25%100%400%2000%Annual turnoverAnnual cost drag

Scroll the chart sideways to see all of it.

The per-trade cost is fixed and the frequency is a design choice, so the cost burden is something a strategy is built with rather than something it encounters. At 20 basis points a round trip.

Testing the sensitivity

The most useful diagnostic is to re-run a backtest at several cost assumptions and observe where the result stops being positive.

  • A strategy that survives a pessimistic cost assumption is robust to this particular error.
  • One that only works at zero cost was never a strategy.
  • The cost level at which it breaks even is a direct measure of how much room the edge has.
  • The same exercise identifies whether reducing turnover would improve the net result, which it frequently does.

This is the same reasoning that appears in the risk pillar as frequency being a risk parameter. The quantitative version is more precise and reaches the identical conclusion.

Capacity

A strategy's capacity is the amount of capital it can deploy before its own trading consumes the edge. It is a property of the strategy and it is estimable in advance.

  1. 1Estimate the edge per tradeIn basis points, from the backtest before costs.
  2. 2Model impact as a function of sizeUsing the square-root relationship against typical volume.
  3. 3Find where impact equals the edgeThat size is the capacity.
  4. 4Apply a substantial marginBecause the estimate is derived from a model rather than from experience.

Capacity falls sharply as the universe gets smaller and less liquid, which is precisely where the largest apparent edges in most backtests are found. A strategy with an enormous backtested return in micro-cap securities frequently has a capacity too small to matter.

Educational content only. MadStockAlerts provides market commentary, research, and educational content. It is not personalized investment advice, and nothing here is a recommendation to buy or sell any security. Trading and investing involve substantial risk, including loss of capital. See the Risk Disclosure and Customer Agreement.