Skip to content

Core Concepts

What is Trade Geometry?

Trade geometry refers to the path-dependent behavior of trades after entry. Instead of just looking at final P&L, we analyze:

  • MFE (Max Favorable Excursion): Best achievable profit within a time horizon
  • MAE (Max Adverse Excursion): Worst drawdown experienced
  • Time sequencing: When did profit vs pain occur?
  • Path dependencies: Can the trade hit TP before SL?

Why This Matters

Traditional backtesting shows you if a signal works. Trade geometry shows you:

  • HOW it works (profit-first vs pain-first)
  • WHERE to place stops and targets (based on empirical paths)
  • WHEN to exit (time-based vs price-based)
  • WHAT risk management suits the signal (trailing vs fixed)

This is NOT

  • A signal generator or optimizer
  • A backtesting engine
  • A portfolio management tool

This IS

  • A diagnostic tool for analyzing signal behavior
  • An empirical guide for exit rule design
  • A path-dependent risk analysis framework

The Problem

You have a trading signal that fires entry triggers. You need to answer:

  1. What stop-loss size is realistic?
  2. What take-profit targets are achievable?
  3. Should I use trailing stops or fixed stops?
  4. Does the signal need "room to breathe" or does profit come quickly?
  5. Are there different trade archetypes within my signal?

The Solution

For each entry signal, we:

  1. Extract the forward price path over horizon H
  2. Compute MFE/MAE along the path
  3. Record timing: when did MFE/MAE occur?
  4. Analyze the 2D distribution (MFE, MAE)
  5. Derive actionable insights for exits

The Process Flow

Signal → Entry Events → Trade Paths → Geometry Analysis → Exit Rules
  ↓          ↓              ↓               ↓                 ↓
 ±1        indices      MFE/MAE/t     Distributions     TP/SL/Time

Key Terminology

MFE (Max Favorable Excursion)

The best profit the trade could have achieved if you had a perfect exit. For long trades, it's the highest high minus entry price. For shorts, it's entry minus lowest low.

Example: You enter long at $100. Price touches $105, then falls to $98. Your MFE is 5% even if you exited at $98.

MAE (Max Adverse Excursion)

The worst drawdown experienced during the trade. For longs, it's entry minus lowest low. For shorts, it's highest high minus entry.

Example: You enter long at $100. Price drops to $97, then recovers to $103. Your MAE is -3% even if final P&L is +3%.

Time to MFE (t_mfe)

How many bars it took to reach the MFE. Critical for deciding between immediate exits vs patience.

Time to MAE (t_mae)

How many bars it took to reach the MAE. Reveals if pain comes immediately (hard to avoid) or later (can be stopped out).

Ordering

Whether MFE or MAE occurred first:

  • MFE-first: Profit before pain → Trailing stops work well
  • MAE-first: Pain before profit → Need room, use wide fixed stops
  • Tie: Both at same time (rare)

Horizon (H)

The number of bars to look forward from each entry. Typically 10-50 bars depending on your strategy timeframe.

Example: If you trade hourly bars and typically hold for 2 days, use H=48 (48 hours).

The Six Analysis Sections

Section A: Geometry Overview

Answers: What does the overall distribution of outcomes look like?

Key Outputs:

  • Scatter plot (MFE vs MAE)
  • Marginal distributions
  • Summary metrics (median, win rate, tail risks)

Interpretation: Tight cluster = consistent behavior, wedge shape = good asymmetry, dense at (0,0) = weak follow-through.

Section B: Frontiers

Answers: Where are the natural risk/reward boundaries? What stop size makes sense?

Key Outputs:

  • Risk-constrained frontier (max MFE for given DD cap)
  • Opportunity-constrained frontier (DD needed for target)
  • Knee points (optimal stop sizing)

Interpretation: Knee point = where more risk doesn't buy much more upside. Use it for stop sizing.

Section C: Ordering

Answers: Does profit come before pain, or vice versa?

Key Outputs:

  • MFE-first vs MAE-first proportions
  • Trailing suitability metrics
  • "Needs room" indicators

Interpretation: >60% MFE-first = use trailing stops. >60% MAE-first = signal needs room, use wide stops.

Section D: TP/SL Feasibility

Answers: What TP/SL combinations are realistic? What's the true probability of hitting TP before SL?

Key Outputs:

  • Probability heatmap (path-dependent)
  • Expected value heatmap
  • Robust zones (high EV, stable)

Interpretation: Green zones = profitable TP/SL combos. Avoid fragile peaks with low sample counts.

Section E: Volatility Normalization

Answers: Is my edge regime-dependent? Should stops scale with volatility?

Key Outputs:

  • Percentage vs vol-normalized scatter comparison
  • Regime-split analysis (low/mid/high vol)
  • Stability metrics

Interpretation: If geometry changes dramatically in vol space → use ATR-adaptive exits.

Section F: Clusters (Trade Archetypes)

Answers: Are there distinct trade types within my signal?

Key Outputs:

  • Cluster scatter plots (colored by archetype)
  • Per-cluster statistics
  • Suggested exit rules per cluster

Interpretation: Fast winners vs needs-room vs noise. Apply different exits to each cluster.

Common Questions

Q: Why analyze geometry instead of just final P&L?

A: Because two trades with same P&L can have vastly different paths. One might go straight to profit, another might drawdown 5% first. This affects what exit rules will work.

Q: What's a good sample size?

A: Minimum 100 trades per side. 500+ is better. 1000+ is ideal for Section D heatmaps.

Q: Should I optimize TP/SL based on Section D?

A: No. Use it as a guide for realistic zones. Look for robust zones with stable EV, not single peaks.

Q: My strategy shows 50/50 ordering. What does that mean?

A: Your signal is regime-mixing (sometimes catches trends, sometimes mean-reverts). Consider filtering by regime or using hybrid exits.

Next Steps