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

Onyx Patterns

Top mistakes

Trader-defined mistake-tags ('FOMO entry', 'no stop', 'over-sized', 'broke plan') are the cleanest map of where discipline breaks down. The detector sums net P&L across every trade carrying each mistake-tag and surfaces the three most expensive — so you can fix the one that hurts most before chasing the rest.

What it is

Top mistakes is a per-tag aggregation across the user's closed trades, restricted to tags whose category is `mistake` (the discipline-failure bucket — separate from setup tags, psychology tags, or custom tags).

For each mistake-tag the trader has applied: - sum the net P&L of every trade carrying it - count the trades - compute the avg per-trade P&L

The top three by absolute aggregate cost surface on the card. A trade with two mistake-tags contributes its P&L to BOTH aggregates — we're measuring per-mistake-kind cost, not apportioning a single trade's P&L across labels. (If 'FOMO entry' AND 'no stop' both appear on the same -€100 trade, both aggregates absorb the full -€100. The trader sees both as expensive and decides which intervention to prioritise.)

Tags with positive aggregate are excluded. If the trader labelled three +€50 trades with 'over-sized', the aggregate is +€150 and the tag drops out of the top-3 — that's not an expensive mistake to fix, that's a mis-labelled set of profitable trades. Card stays silent if every mistake-tag the user has applied netted positive (no expensive mistake to surface).

Formula
Approach (textbook): group closed trades by every mistake-tag they carry. Per group: count, sum_pl = Σ net_pl, avg_pl = sum_pl / count. Filter out groups where sum_pl is non-negative (the tag isn't an expensive mistake). Sort the rest by sum_pl ascending (most negative first), take the top 3.
 
TradeOnyx-internal: the minimum number of trades carrying any mistake-tag before the detector emits, the per-mistake threshold for inclusion, and the top-N count are calibrated empirically and not published.
Example

Across the analysis window, 22 trades carry at least one mistake-tag. Aggregations: 'no stop' on 6 trades = -€480 total. 'FOMO entry' on 8 trades = -€340 total. 'over-sized' on 4 trades = -€220 total. 'broke plan' on 3 trades = -€90 total. Total mistake cost: -€1,130.

ResultCard emits with 'no stop' as the top mistake (-€480 over 6 trades, avg -€80 per trade). 'FOMO entry' is second, 'over-sized' third. 'Broke plan' is excluded — fourth in cost, only the top 3 surface. The actionable read: write ONE rule against 'no stop' into the journal — 'no entry without a pre-set SL price, no exceptions'. Don't try to fix FOMO + sizing + planning simultaneously. Eliminating the top single mistake compounds over weeks; the others surface again next month if they remain.
How to read it

How to use the card:

The top entry is the single most expensive mistake category in your trading right now. That's the one that gets a journal rule. The runner-ups are noted and re-checked monthly — if eliminating #1 leaves them as the new #1, fix that next.

Why per-tag aggregation, not per-tag count. A 'no stop' tag on 1 trade that lost €500 is more expensive than 'FOMO entry' on 10 trades that lost €30 each (€500 vs €300). Count alone would surface FOMO; the aggregate gets the priority right. The avg-per-trade column is shown so the trader can also see the per-incident severity.

Why a single trade contributes to multiple tags. If a trade has both 'FOMO entry' AND 'no stop', both behaviours contributed to the loss. Apportioning 50/50 would let the trader rationalise away one of them. Counting full attribution to both correctly answers: 'every time I FOMO, what's the average cost' — and FOMO often co-occurs with no-stop, so both numbers are inflated, which is HONEST about the entanglement.

The discipline-elimination math. A trader who eliminates a -€500/quarter mistake by writing a single journal rule clears 4×€500 = €2,000/year of pure friction. Compounding into the equity curve is the entire point of mistake-analysis: it's the highest-leverage trader-improvement work, far beyond strategy-tweaking.

Limitations.

  • Depends on tagging discipline. If the trader doesn't tag mistakes, the detector has nothing to surface. Card stays silent below MIN_TAGGED_TRADES.
  • Trader-defined names. 'FOMO entry' and 'fomo' are different tags. Encourage tag-name normalisation in the trade-detail UI before this card grows useful for cross-user analysis (post-launch).
  • No co-occurrence here. That's TRA-237's job (Mistake-pair co-occurrence). This detector deliberately stays one-dimensional.

Tier: Pro. Wave 6 (Mistake Analysis) opens with this — followed by frequency-over-time (TRA-236), pair co-occurrence (TRA-237), and mistake-free streak (TRA-238).

Where TradeOnyx uses it

How to read the card:

1. Hero (left) — the worst single mistake's name + total cost. The number that hurts. 2. Breakdown table — the top 3 by aggregate cost, with count + total + avg per trade. Read top-down. 3. Hint line — total mistake cost across all flagged categories, plus the discipline-elimination prompt.

Re-look frequency: monthly. Mistake patterns stabilise over months, not days; the card is most actionable at the end of a trading month or quarter when reviewing the journal.

Tier: Pro.

Related reading