PineConnector Pulse · Open research · August 2026
The state of retail trading automation. Measured.
Since January 2021, PineConnector has carried retail trading signals from TradingView into MetaTrader — around 200 million of them. Every month we publish what that flow looked like: which markets it went to, which way it pointed, and when it happened. Aggregates only, methodology attached, free to cite.
01 · What changed
The month in one line
Overall instruction activity was flat, but its composition rotated sharply: entry instructions rose while management instructions fell, and the market mix moved out of indices and crypto into FX and gold.
All instruction activity
Flat
vs matched July window
Entry instructions
+19.1%
management instructions fell 23.0%
Entry share of activity
65.1%
was 54.6% in July
Long share of entries
56.2%
activity-weighted, historical entries
Composition, not volume
Total instruction activity barely moved. What changed was the mix: for every 100 instructions, July 2026 produced about 55 entries and 45 management instructions; August 2026 produced about 65 and 35.
- Entry instructions19.1% increase
- Management and other instructions23.0% decrease
- All instruction activityno change
- Active connected accounts sending entries1.7% decrease
- Mean entries per active account21.1% increase
The last figure is an arithmetic mean, not a median. It says the average rose — not that a typical account became 21% busier. A smaller number of high-volume accounts can move it on their own.
02 · Markets
The most automated markets
Two different questions, two different answers. Reach is how many connected accounts touch a market — how widespread it is. Activity share is how much of the instruction flow it carries — how intense it is. Gold (XAU/USD) leads both, but underneath them the two rankings diverge sharply.
Reach — share of active connected accounts
August 2026 · change vs July 2026 at right · Markets shown meet the publication threshold in both windows: at least 100 distinct connected accounts. Accounts trade several markets, so these do not sum to 100%.
Widespread is not the same as busy
Activity share against reach, for the markets where both are published. Gold is both. The FX pairs are busier than they are widespread; the index products are the reverse.
Data table — markets, August 2026
| Market | Reach | Reach change | Activity share | Activity change |
|---|---|---|---|---|
| Gold (XAU/USD) | 59.4% | +2.5 pp | 31.5% | +32% |
| Nasdaq 100 | 16% | -3.2 pp | 8.2% | -6.6% |
| Bitcoin (BTC/USD) | 15.4% | +0.3 pp | 11.4% | -4% |
| EUR/USD | 14.4% | 0 pp | 8.3% | +52.1% |
| USD/JPY | 10.2% | +2.7 pp | 4.6% | +45.6% |
| GBP/USD | 7.3% | -0.5 pp | 6.3% | +37.5% |
| Nikkei 225 | 6.4% | -0.4 pp | — | — |
| Dow Jones (US30) | 6.3% | -2.8 pp | — | — |
| DAX (GER40) | 6.1% | -0.7 pp | — | — |
| S&P 500 (US500) | 5.5% | +0.2 pp | — | — |
| AUD/USD | 4.3% | -0.4 pp | — | — |
| Silver (XAG/USD) | 4% | 0 pp | — | — |
| USD/CAD | 3.7% | 0 pp | — | — |
| GBP/JPY | 3.6% | -1 pp | — | — |
On the threshold
Markets need at least 100 distinct connected accounts in both windows to be published. Two sat on the line this month:
- EUR/JPY — cleared the threshold in August, just missed it in July — growing
- USD/CHF — cleared in July, missed in August — shrinking
03 · Asset classes
Where the flow lives
Foreign exchange carried the most instruction activity this month, butmetals reached the most accounts — 61.4% of them. Most active and most widespread are not the same category, and the gap between them is the story.
Share of instruction activity
- Foreign exchange35.5%
- Metals31.8%
- Equity indices19.6%
- Crypto11.6%
- Other1.5%
Activity against reach
The two bars answer different questions. FX carries the most instructions from the fewest accounts; metals is the mirror image — a majority of accounts, sending fewer instructions each.
Who drove the change
Share of the net change in entry activity contributed by each class. FX and metals together contributed more than 100% because the other classes offset part of their gain.
- Foreign exchange+67.8%
- Metals+47.5%
- Equity indices−8.6%
- Other−3.6%
- Crypto−3.1%
Data table — asset classes, August 2026
| Asset class | Activity share | Share change | Activity change | Reach |
|---|---|---|---|---|
| Foreign exchange | 35.5% | +6.2 pp | +44.1% | 27.1% |
| Metals | 31.8% | +3 pp | +31.5% | 61.4% |
| Equity indices | 19.6% | -5.3 pp | -6.6% | 28.9% |
| Crypto | 11.6% | -2.9 pp | -4.1% | 16.9% |
| Other | 1.5% | -1 pp | -27.3% | — |
Foreign exchange
- Activity share35.5%
- Activity change+44.1%
- Reach27.1%
Largest asset class by activity, but reaches barely a quarter of accounts — the most concentrated major class.
The three leading pairs account for 53.8% of FX activity, effectively unchanged from July.
Metals
- Activity share31.8%
- Activity change+31.5%
- Reach61.4%
The broadest category by far: three in five active accounts touch it.
Gold is 99.1% of metals activity — the metals category is functionally a gold category.
Equity indices
- Activity share19.6%
- Activity change-6.6%
- Reach28.9%
Lost both activity share and reach — the clearest retreat of the month.
Nasdaq 100 held 42.0% of index activity in both months.
Crypto
- Activity share11.6%
- Activity change-4.1%
- Reach16.9%
Reach held steady while activity fell: the same accounts, sending fewer instructions.
Bitcoin is 98.0% of crypto activity — the crypto category is functionally a bitcoin category.
Other
- Activity share1.5%
- Activity change-27.3%
No individual market here meets the publication threshold.
Energy, single stocks and ETFs, synthetic indices and the remaining long tail.
04 · Direction
Which way the instructions pointed
56.2% of classified entry instructions carried a long direction, up from 51% in July 2026 — every major asset class moved the same way. These are historical entry instructions, not open positions, exposure, sentiment or a forecast, and a long instruction says nothing about whether it worked.
By asset class
Net long bias in percentage points (long − short). Right skews long, left skews short.
By market
Markets meeting the publication threshold, activity-weighted.
Two ways to count, two different answers
Weighting every instruction equally lets the busiest accounts dominate. Weighting every account equally asks what a typical connected account did. Where the two disagree, the crowd and the volume disagree.
- USD/JPY73.4% / 70.6%−2.8
- EUR/USD67.1% / 58.4%−8.7
- Bitcoin (BTC/USD)62.2% / 60%−2.2
- Nasdaq 10055% / 61.1%+6.1
- Gold (XAU/USD)53.1% / 62.9%+9.8
- GBP/USD49.2% / 56.1%+6.9
Activity-weighted — every instruction counts oncePer-account — every account counts onceScale: 0% (all short) → 50% (balanced) → 100% (all long)
Gold (XAU/USD) shows the widest split this month. Where the per-account figure sits higher, the busiest accounts were less long-leaning than the wider base — a difference that a single headline percentage would hide. Market-level activity-weighted direction remains provisional pending a contributor-concentration check.
Data table — direction, August 2026
| Category | Activity-weighted long | Change | Per-account long | Change |
|---|---|---|---|---|
| Foreign exchange | 57.3% | +5.6 pp | 62.9% | +5 pp |
| Metals | 53% | +4.1 pp | 63% | +7.2 pp |
| Equity indices | 55.5% | +6.8 pp | 64.5% | +2 pp |
| Crypto | 61.9% | +3.7 pp | 60.3% | +2.1 pp |
| USD/JPY | 73.4% | +14.8 pp | 70.6% | +5.6 pp |
| EUR/USD | 67.1% | +5.8 pp | 58.4% | +3.4 pp |
| Bitcoin (BTC/USD) | 62.2% | +3.8 pp | 60% | +2 pp |
| Nasdaq 100 | 55% | +4.2 pp | 61.1% | +0.2 pp |
| Gold (XAU/USD) | 53.1% | +4.2 pp | 62.9% | +7 pp |
| GBP/USD | 49.2% | -2.4 pp | 56.1% | +2.6 pp |
05 · The clock
When the machines trade
Automation does not mean around the clock. One hour of the day carries 5.68% of instructions and another carries 1.78% — a 3.19× swing between the busiest and quietest hour, and the same two hours held those positions in July 2026. The shape of the day is one of the most stable things in this dataset.
Flow by hour of day
Share of the period's entry instructions per hour. Pick a timezone, or let it follow yours — the chart rotates the same buckets, it does not re-cut the data.
Data table — source hours
| Hour | Share | Change |
|---|---|---|
| 00:00–01:00 | 4.52% | +0.1 pp |
| 01:00–02:00 | 4.8% | +0.4 pp |
| 02:00–03:00 | 4.25% | +0.1 pp |
| 03:00–04:00 | 3.79% | -0.2 pp |
| 04:00–05:00 | 3.67% | -0.3 pp |
| 05:00–06:00 | 4.21% | 0 pp |
| 06:00–07:00 | 4.24% | 0 pp |
| 07:00–08:00 | 4.25% | -0.1 pp |
| 08:00–09:00 | 4.19% | -0.3 pp |
| 09:00–10:00 | 4.06% | -0.2 pp |
| 10:00–11:00 | 3.99% | -0.3 pp |
| 11:00–12:00 | 4.25% | -0.3 pp |
| 12:00–13:00 | 4.9% | +0.1 pp |
| 13:00–14:00 | 5.68% | +0.5 pp |
| 14:00–15:00 | 5.29% | +0.3 pp |
| 15:00–16:00 | 4.79% | +0.2 pp |
| 16:00–17:00 | 4.18% | -0.1 pp |
| 17:00–18:00 | 4.05% | 0 pp |
| 18:00–19:00 | 4% | 0 pp |
| 19:00–20:00 | 4.01% | +0.1 pp |
| 20:00–21:00 | 3.4% | -0.1 pp |
| 21:00–22:00 | 1.78% | +0.2 pp |
| 22:00–23:00 | 3.74% | -0.1 pp |
| 23:00–00:00 | 3.98% | 0 pp |
Flow by day of week
Indexed so the average weekday = 100, which corrects for months containing unequal numbers of each weekday.
Data table
| Day | Index | Change |
|---|---|---|
| Mon | 101.7 | -5.6 |
| Tue | 101.4 | -4 |
| Wed | 101.7 | +0.7 |
| Thu | 101.6 | +1.6 |
| Fri | 91.4 | +3.8 |
| Sat | 9.4 | -1.4 |
| Sun | 16.5 | -3.5 |
Fast facts
August 2026 · source hours
- Busiest hour13:00–14:00 · 5.68%
- Quietest hour21:00–22:00 · 1.78%
- Peak-to-quiet swing3.19× · July 2026: 3.15×
- Strongest 4-hour block12:00–15:59 · 20.7%
- Weekend vs weekday12.9% · of average weekday activity
Monday through Thursday sat within 0.3 index points of each other — effectively tied, so no single weekday is "the busiest." Friday runs about 8.6% below the average weekday.
06 · Instruction anatomy
What the instructions carried
Beyond which market and which direction, every entry instruction either carries exit levels or does not. The share carrying a stop-loss field rose while the share carrying a take-profit fell — driven almost entirely by a jump in stop-loss-only instructions, up 6.5 points in a month.
Exit levels attached to entry instructions
Every entry instruction falls into exactly one of these four groups.
- Both SL and TP52%4.2 pp decrease
- SL only15.5%6.5 pp increase
- TP only3.4%1.6 pp increase
- Neither field29.1%3.9 pp decrease
What this does not say. Measures whether the parsed entry instruction carried a recognised stop-loss or take-profit field. It does not establish that a position was protected: the field may be invalid, protection may come from EA settings, a later instruction, or strategy logic. It is a measurement of instruction text, not of whether anyone managed their risk.
Data table — exit fields, August 2026
| Group | August 2026 | July 2026 | Change |
|---|---|---|---|
| Both SL and TP | 52% | 56.2% | -4.2 pp |
| SL only | 15.5% | 9% | +6.5 pp |
| TP only | 3.4% | 1.8% | +1.6 pp |
| Neither field | 29.1% | 33% | -3.9 pp |
Terminal split
Share of accounts downloading or updating the connector in the window. It reflects setup and update activity, not a live census of running terminals.
- MetaTrader 51.1 pp increase
- MetaTrader 41.1 pp decrease
07 · Method
How the numbers are made — and what they are not
The unit
A tracked instruction is one instruction received by the bridge. It is not a broker fill, an executed order, an open position, or a trading decision.
Reach
Reach is the share of active connected accounts with at least one instruction in a market. Accounts trade several markets, so reach does not sum to 100%.
Direction
Direction describes historical entry instructions, never positions, exposure, sentiment or advice.
Never published
No account, licence or customer identifiers. No absolute instruction, account or revenue totals. No broker or prop-firm attribution. No market below the publication threshold. No performance or profitability claims.
08 · Reports · RSS
Written editions
The narrative version of each release — what moved, what reversed, what to be careful about.
- Pulse 001 — August 2026: the month the mix changedAugust 28, 2026