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

Onyx Patterns

Market regime

Your edge is rarely a constant — it's almost always a regime-conditional thing. Bull markets reward long-bias setups; bear markets reward shorts and patience; sideways markets murder both. The detector splits your closed trades by the regime that was in force on each trade's date and shows you where the carry comes from.

What it is

Market regime is the macro state of an asset class, classified by the 252-day price action of its leading proxy:

  • Bull: price ≥ 20 % above the 252-day low AND EMA50 > EMA200. The market is in a sustained uptrend.
  • Bear: price ≥ 20 % below the 252-day high AND EMA50 < EMA200. The market is in a sustained downtrend.
  • Sideways: neither — range-bound or in mid-correction.

The leading proxy is per-asset-class:

  • Indices / stocks → ^GSPC (S&P 500)
  • Commodities → GLD (SPDR Gold Trust)
  • Crypto → BTC-USD (Bitcoin)
  • FX (specific pair) → the pair itself
  • Generic FX → no regime read (relative quoting has no single proxy)

For each closed trade, the detector reads the regime that was in force on the trade's close date, then buckets the trade into bull / bear / sideways. With ≥ 30 classified trades and ≥ 8 trades in each of at least two buckets, the detector compares win-rate and avg P&L across buckets and emits a verdict.

Formula
Approach (textbook): for each closed trade, evaluate the regime classifier at the trade's close date against the leading proxy of the user's primary asset class. Bucket the trades. Per bucket compute: count, win-rate (winners / count × 100), sum P&L, avg P&L. Find the best- and worst-avg-P&L buckets among those with ≥ 8 trades. Decide the verdict by comparing the BEST-bucket lift over the others' mean against the WORST-bucket drag below the others' mean.
 
TradeOnyx-internal: the minimum total of classified trades, the minimum per-bucket sample, the win-rate gap threshold, the relative avg-P&L gap threshold, and the verdict tie-break logic are calibrated empirically and not published.
Example

Across the analysis window: 64 trades classified. Bull bucket: 32 trades, win-rate 68 %, avg P&L +€18. Bear bucket: 18 trades, win-rate 28 %, avg P&L -€12. Sideways bucket: 14 trades, win-rate 50 %, avg P&L +€2.

ResultCard emits with verdict **Specialist**: the bull regime carries the edge — its lift over the others' mean (~€8) is materially larger than bear's drag below it (~€7). The actionable read: when the macro flips bear, the trader's strategy needs a separate playbook or stand-aside rule.
How to read it

The two verdicts:

  • Regime specialist. One regime carries the edge; the others are flat or mildly negative. Don't read this as 'avoid the others' — read it as 'you have a regime-conditional edge'. Two ways to use the read: size up modestly in the leader regime, OR build a separate playbook (different setups, different risk parameters) for the lagging regimes.
  • Regime struggler. One regime drags the edge while the others carry the equity curve. The cleanest fix is to stand aside in the struggling regime until you've designed a strategy that works there. Trying to force the same playbook into a regime where it bleeds is what most retail accounts get wrong.

The Onyx-Engine reports whichever pattern is dominant in your data; balanced regimes (no bucket meaningfully diverges) keep the card silent.

Why regime is a separate axis from win-rate. Standard win-rate stats average over all market conditions, hiding the regime-conditionality. A 55 %-overall trader could be 80 % in bull and 25 % in sideways — same number, three times the actionable insight if you split the buckets. The detector forces that split.

The asset-class proxy is a deliberate simplification. The detector classifies the WHOLE asset class as one regime — not per-symbol. Why? Because retail correlation is brutal: when S&P is in a bear market, 70-80 % of US stocks correlate down with it. When BTC is in a bear market, ETH and the alts crash with it. Per-symbol regime classification would be more accurate but would also drown the trader in noise. The single asset-class regime is the read that actually changes behaviour.

Limitations.

  • Generic FX traders get no read. FX is relative — there's no single proxy that classifies 'currency markets' as bull or bear. The card stays silent. Trade specific pairs and the regime is the pair itself.
  • Regime transitions are slow. Bull-to-bear transitions take weeks to months. The card you see today reflects the state of the market on each trade's actual close date, but the user's near-term planning should always pair this with the live regime tile on the briefing.
  • Need 30+ classified trades. A first-month account doesn't have enough sample. Card stays silent until the data exists.

Tier: Pro. Wave 5 (Market Context). Pairs naturally with the briefing-tab regime tile (which shows the CURRENT regime live) — the patterns card shows the historical breakdown, the briefing tile shows the present.

Where TradeOnyx uses it

How to read the card:

1. Hero (left) — focus regime word + verdict (specialist/struggler). 2. Breakdown table — per-regime trade count, win-rate, avg P&L. Three rows always: bull / bear / sideways. Buckets with too few trades are still shown but didn't drive the verdict. 3. Hint line — the rule that addresses the dominant verdict.

Re-look frequency: monthly. Regime classification is slow-moving; daily noise is irrelevant.

Tier: Pro.

Related reading