Lodash _.round Tie-Breaking Logic Explained

This article examines the exact tie-breaking mechanism executed by the Lodash _.round utility when handling midpoint values. It details how the function delegates rounding to JavaScript's standard runtime, the behavior of its "round half toward positive infinity" rule across positive and negative values, and how exponential string manipulation is used to preserve precision during mid-point evaluation.

Lodash Implementation Mechanism

Lodash does not implement a custom rounding algorithm or banker's rounding (round half to even). Instead, _.round is generated by an internal factory function called createRound, which delegates directly to JavaScript’s native Math.round.

When a precision parameter is supplied, Lodash converts the input number into exponential notation via string manipulation to avoid binary floating-point rounding errors. It shifts the decimal point by the specified precision, calls Math.round on the shifted value, and shifts the decimal back using exponential notation. Consequently, the tie-breaking logic executed on the resulting midpoint is strictly defined by ECMAScript's specification for Math.round.

The Tie-Breaking Rule: Round Half Toward Positive Infinity

When _.round encounters an exact midpoint value (where the fractional component is precisely .5), it resolves the tie using round half toward positive infinity (often referred to as round-half-up).

The rule behaves as follows:

• If the fractional part is exactly 0.5, the number is rounded to the next integer in the direction of positive infinity (\(+\infty\)).
• For positive numbers, this rounds away from zero. For example, _.round(2.5) evaluates to 3.
• For negative numbers, rounding toward positive infinity means rounding toward zero. For example, _.round(-2.5) evaluates to -2, not -3.

Precision-Adjusted Midpoints

When rounding to specific decimal places using the precision argument, the same rule applies after the decimal shift:

• _.round(1.55, 1) shifts to '15.5', evaluates Math.round(15.5) to 16, and shifts back to return 1.6.
• _.round(-1.55, 1) shifts to '-15.5', evaluates Math.round(-15.5) to -15, and shifts back to return -1.5.

Because Lodash relies on exponential notation (${number}e${precision}) rather than basic multiplication (number * 10 ** precision), it minimizes floating-point inaccuracies that frequently cause numbers like 1.005 to be represented internally as 1.0049999999999999. Once converted to a clean decimal representation, any exact .5 boundary reliably breaks toward positive infinity.