Directional Overlap
Directional overlap (concordance S, 0–1): how much two reports hold the same instrument on the SAME side at the same time — the side-aware refinement of symbol overlap. A hedge (one long, one short) scores 0. Tells same-direction crowding from offsetting positions.
- Computed from
- Trades list
- Scope
- Across reports
- Range
- 0 – 1
- Direction
- Lower is better
Directional overlap is the side-aware refinement of symbol overlap. Symbol overlap asks whether two reports are in the same instrument at the same time; directional overlap adds one condition — they must be on the same side. Both long EURUSD counts; one long while the other is short does not. It separates genuine doubled-up crowding from a hedge — when one report bets the price up and the other bets it down, so the positions partly cancel instead of compounding.
How it's calculated
The mechanics are identical to symbol overlap — a time-domain concordance on each trade's [open, close] interval, max-normalized, lot size ignored, measured inside the pair's common active window, no Epps-effect binning. The only change is the unit of analysis:
S = Σ (time both hold the same symbol AND the same side)
───────────────────────────────────────────────────────
Σ over each shared symbol × side: max(Tᴀ, Tʙ)- symbol × side
- the unit — EURUSD-long and EURUSD-short are SEPARATE units, scored independently and then blended by weight
- Tᴀ, Tʙ
- each report's time on that symbol×side unit, within the pair's common active window
- S
- similarity ∈ [0,1]; NaN if the pair never shares a symbol at all
What it tells you
| Value | Reading | Notes |
|---|---|---|
| 0.7 – 1.0 | Doubled up | Same instrument, same direction, same time — effectively one position at twice the size. |
| 0.3 – 0.7 | Partial | Meaningful same-side overlap on some instruments. |
| 0.0 – 0.3 | Distinct | They rarely take the same side of the same market at once. |
| NaN | No shared market | They never traded a common symbol — nothing to overlap. |
Symbol overlap vs directional overlap
Read them together — the gap between the two numbers is itself the signal:
| Symbol overlap (default) | Directional overlap | |
|---|---|---|
| Unit | symbol (long ∪ short merged) | symbol × side (long, short separate) |
| Two reports overlap when… | both are in the instrument | both are in it on the same side |
| A hedge (one long, one short) | counts as full co-exposure | scores 0 |
| Answers | "are they in the same markets?" | "are they running the same trade?" |
- High symbol, high directional — they hold the same instruments the same way. True crowding; the portfolio is concentrated and a move against the shared side hits both books at once.
- High symbol, low directional — same markets, opposite sides. They appear to hedge each other on that instrument. Treat that "hedge" with suspicion: the offset is only as clean as the two legs are matched in size and timing — different lots, or one report closing first, leaves a net position that reappears exactly when it's least wanted. And a cross-strategy hedge still pays spread, swap and margin on both legs, so it nets price risk, not cost. It smooths the equity curves in calm markets without removing the market's gap risk.
- Low on both — genuinely separate exposure (still worth a currency-leg check — see symbol overlap).
- It's not signed. Directional overlap has no −1: a hedge reads 0 (different units), not "negative." For anti-correlation of results, read return correlation — a different question.
- A strategy that routinely holds both sides at once (a grid or martingale, which stacks buys and sells on the same pair) is split across the long and short units, so it can only ever match half of a one-sided copy of itself — directional overlap is pulled toward ~0.5 even though the two clearly share the market. For such net-hedging strategies, directional overlap under-reads the crowding; trust symbol overlap for that.
- Same NaN and size blind spots as symbol overlap — no shared symbol → NaN; lot size ignored (read with trade leverage).