Scope and learning objectives

This article synthesizes the supplied claims about strategy comparison, historical and displayed performance, backtest interpretation, and adjustment review. Examples remain specific to their stated structures, configurations, and contexts.

  • Identify when two strategy results are not comparable on a like-for-like basis.
  • Explain why annual returns, displayed gains, and unusually perfect backtests require contextual review.
  • Evaluate an observed outcome by considering alternative market paths and adjustment behavior.
  • Distinguish an adjustment exercise used for learning from a universal management rule.

01

Establish a Like-for-Like Comparison

Before judging relative performance, examine each structure independently and identify material differences in strike location, trade size, duration, and the underlying's location relative to the position. [2][4][1]

  • Similar-looking configurations at different strikes are not a like-for-like comparison. [4]
  • The cited comparison process evaluates structures in isolation and accounts for differences in trade size and duration. [2]
  • In that specific example, the diagonal uses weekly options seven to fourteen days out and is held for about seven days; those durations are example-specific. [2]
  • Do not directly compare the closing proceeds or cost of one butterfly with the opening cost of another when the underlying occupies substantially different locations relative to their tents. [1]

02

Interrogate the Performance Record

Reported performance is interpretable only in context: annual-return calculations depend on capital definitions and compounding, while a displayed gain may cover only part of the position record. [3][6]

  • Historical annual return should not be the primary basis for selecting a strategy or projecting future performance. [3]
  • Return calculations depend on how capital is defined and on compounding over a changing base. [3]
  • In the cited display example, a $30,000 gain was incomplete because only open positions were included. [6]

03

Challenge Perfect-Looking Backtests

A loss-free backtest can generate impressive-looking but unrealistic statistics and does not provide a realistic representation of future trade performance. [5][7]

  • A backtest that never loses can report a 100% win rate alongside a meaningless positive value for average loss. [5]
  • The absence of losses prevents that backtest from realistically representing how the trade will perform in the future. [7]
  • Perfect-looking summary statistics should therefore be interpreted in light of the loss-free record that produced them. [5][7]

04

Review Alternative Paths and Adjustment Behavior

Observed results are only one path. A fuller comparison considers other possible market paths, what followed the initial move, and how each strategy would be adjusted, while treating adjustment examples as tools for understanding position dynamics rather than universal prescriptions. [8][9][10]

  • Before drawing conclusions, consider what could have happened under other market paths and what occurred after the initial move. [8]
  • Include how each strategy would be adjusted when comparing outcomes. [8]
  • In the cited strategy-specific setting, repeated alternation between upward and downward adjustments signals that the position or its thresholds may be too tight for a back-and-forth market. [9]
  • If that alternation persists in the cited setting, transaction and repositioning losses make a losing outcome increasingly likely. [9]
  • The positive-10-delta adjustment in the described exercise examines the trade-off between reducing downside risk and taking on upside risk; its purpose is to understand position dynamics, not to win every scenario. [10]

Review

Key takeaways

  1. Make structural comparisons only after accounting for strike location, size, duration, and relevant position context. [2][4][1]
  2. Treat historical returns and displayed gains as context-dependent records rather than self-sufficient conclusions. [3][6]
  3. A loss-free backtest can make summary statistics unrealistic and does not realistically represent future performance. [5][7]
  4. Review alternative paths and adjustment behavior, but keep configuration-specific adjustment examples within their original scope. [8][9][10]

Self-check

Review questions

Why can two similarly configured positions still fail a like-for-like comparison?

Different strikes can invalidate direct comparison, and differences in size, duration, or the underlying's location relative to each position can further change what is being compared. [2][4][1]

What should a reviewer ask before relying on an annual return or displayed gain?

Ask how capital and compounding were defined and whether the display represents a complete record; the cited $30,000 example included only open positions. [3][6]

Why is a 100% win rate insufficient evidence of realistic future performance?

In a loss-free backtest, the win rate and even the average-loss field can be unrealistic, and the record does not realistically represent future trade performance. [5][7]

How should adjustment behavior inform a comparison without becoming a universal rule?

Consider how each strategy would respond under alternative paths and after the initial move. Treat repeated alternating adjustments as strategy-specific feedback and configuration-specific delta exercises as studies of risk trade-offs, not general prescriptions. [8][9][10]

Traceability

Evidence index

Canonical source claims used in this guide. Open a session link to verify the underlying passage at its original timestamp.

[1]The proceeds or cost of closing one butterfly should not be compared directly with the cost of opening another when the underlying price is in a substantially different location relative to each position's tent.
[2]When comparing how two trade structures react, the speaker recommends evaluating each structure in isolation and accounting for differences in trade size and duration; the diagonal in this example uses weekly options seven to fourteen days out and is held for about seven days.
[3]Historical annual return should not be the primary basis for selecting a strategy or projecting future performance, because return calculations depend on capital definitions and compounding on a changing base.
[4]Positions at different strikes are not a like-for-like comparison even when their configurations look similar.
[5]A backtest that never loses can produce a meaningless positive 'average loss' and a 100% win rate, making its performance statistics unrealistic.
[6]A displayed $30,000 gain was not a complete result because the display included only open positions.
[7]A backtest with no losses does not provide a realistic representation of how the trade will perform in the future.
[8]When comparing strategy outcomes, the speaker advises considering what could have happened under other market paths, what occurred after the initial move, and how each strategy would be adjusted before drawing conclusions from the observed result.
[9]Repeated alternation between upward and downward adjustments is feedback that a position or its adjustment thresholds may be too tight for the current back-and-forth market; if it persists, transaction and repositioning losses make a losing outcome increasingly likely.
[10]In the described exercise, the speaker uses a positive-10-delta adjustment to study the trade-off between reducing downside risk and taking on upside risk, with the goal of understanding the position's dynamics rather than winning every scenario.