Launching soon — TradeOnyx is still in preparation; public sign-up isn't open yet.

Onyx Patterns

News proximity

Trades opened within ±30 minutes of a high-impact economic event sit in a different statistical regime than steady-state trades — wider spreads, brutal slippage, the algos that own intraday liquidity stepping back. The detector tells you whether you're profiting from that window or paying a tax on it.

What it is

News proximity is the time-distance between a trade's open and the nearest high-impact economic event. The detector uses two buckets:

  • Near news — at least one high-impact event lands within ±30 minutes of the trade's open time.
  • No news (steady-state) — no high-impact event within that window.

Why ±30 minutes? It's the market-microstructure-research consensus window for elevated volatility around scheduled releases: spreads typically start widening 5–10 minutes before, peak in the first 1–2 minutes after the release, and decay back to baseline over the following 20–30 minutes. Outside that window, instrument-specific edges reassert themselves.

High-impact is the impact-tier metadata on each economic event from the data provider (Finnhub, FRED). Examples: NFP, CPI, FOMC, ECB / BoE / BoJ rate decisions, GDP releases. Low- and medium-impact events (housing starts, regional Fed indices) are NOT in the bucket — those don't materially move spreads on the major asset classes.

For each closed trade, the orchestrator joins the trade's open_time against the high-impact event calendar, sets a private flag (`_near_news=True/False`), and the detector buckets accordingly.

Formula
Approach (textbook): for each trade with an open_time, scan the high-impact event calendar (loaded from the EconEvent table over the trade's time window). If any event datetime falls within ±30 minutes of the open, mark `near_news`. Bucket and compute per bucket: count, win-rate (winners / count × 100), avg P&L. Compare the two buckets on win-rate gap and avg-P&L relative gap.
 
TradeOnyx-internal: the minimum sample required for each bucket before emission, the win-rate gap threshold, and the avg-P&L relative-gap threshold are calibrated empirically and not published. The conflict-suppression logic (refusing to emit when win-rate and avg-P&L disagree) is also internal calibration.
Example

Across the analysis window: 8 trades opened within ±30 min of a high-impact event (win-rate 25 %, avg P&L -€18). 32 trades opened in steady-state (win-rate 65 %, avg P&L +€11).

ResultCard emits with verdict **Underperforming**: the trader's near-news bucket is bleeding while their steady-state edge is strong. The actionable read: half-size or sit out the ±30-minute window until the volatility tax stops eating the edge.
How to read it

The two verdicts:

  • Underperforming. Near-news trades lose ground on either win-rate or avg P&L (or both, in the same direction). Mechanism: the trader is paying a spread/slippage tax that the steady-state edge can't recover, OR is forcing entries on news-driven price action they don't have a setup for. Both have the same fix.
  • Outperforming. Near-news trades beat the steady-state on either win-rate or avg P&L. Mechanism: the trader has a structural edge in capturing the volatility move (gap-trade, fade-the-spike, breakout-confirmation). Less common than underperforming and more valuable when it shows up.

The detector deliberately suppresses emission when win-rate and avg-P&L disagree (one says better, the other worse) — that's typically one outsized winner or loser distorting one of the two stats; small near-news samples make those distortions common. We'd rather stay silent than emit on a coin-flip signal.

Why news proximity is its own axis. Standard P&L stats average over all market conditions, hiding the news-window structure. A 55 %-overall trader could be 75 % in steady-state and 20 % near news — same overall, three times the actionable insight if you split. The detector forces that split.

The fix for underperforming is mechanical, not motivational.

  • Half-size the ±30-minute window until the gap closes. Cuts the cost of the volatility tax in half while keeping you involved.
  • Skip the ±30-minute window entirely. No new entries open inside it; let the print happen, then trade the reaction. Lower trade-count, but every entry lands inside the steady-state edge.

Pick the one that matches the strategy and write it into the journal as a rule. The detector re-checks weekly; if the gap closes after 4–6 weeks of discipline, the rule has worked and you can revisit the policy.

Limitations.

  • High-impact only. Low- and medium-impact events are excluded — they don't materially move spreads on the major asset classes. If you trade exotic FX or thin small-caps where mid-tier events DO move the book, the detector under-counts your news exposure.
  • Fixed ±30 minutes. Some traders' edges are fragile only ±5 min, others have a longer tail to ±60 min. The 30-minute window is the consensus default; future versions may parameterise it per asset class.
  • Need ≥ 5 near-news trades. Below that the bucket is too small to claim anything statistically. The card stays silent.

Tier: Pro. Wave 5 (Market Context). Pairs naturally with the Briefing-tab Economic-Events module which shows the next high-impact event live — the patterns card shows the historical breakdown, the briefing shows the upcoming risk.

Where TradeOnyx uses it

How to read the card:

1. Hero (left) — verdict word + 'wins/loses near events'. Single-glance read. 2. Four MicroStats — win-rate near vs. steady-state, avg P&L near vs. steady-state. Read the two pairs together: if all four point the same way (e.g. near worse on both), the verdict is unambiguous. 3. Hint line — the rule that addresses the dominant verdict.

Re-look frequency: weekly. News-window habits don't change daily; the signal stabilises after ≥ 5–6 near-news trades.

Tier: Pro.

Related reading