Troubleshooting

Find your symptom below. Each entry says the likely cause, what to check, and what to do. Several are the system working as designed, and those are marked.

The connection test fails

Symptom. Test Connection on Accounts returns a failure, or the connection card shows Failed with an error underneath it.

Likely cause. Almost always the credentials or their permissions. The error text on the card comes straight from the exchange, so read it first.

What to check, in order.

1. Every field, retyped. Credentials are write-only once saved, so a stray space or a truncated paste is invisible afterwards.
2. Permissions. The key needs trade permissions on futures or perpetuals. Read-only keys pass a balance check and then fail at the first order.
3. IP allowlist. Our requests do not come from a fixed address, so a key restricted to an IP will not work.
4. The exchange-specific trap. See below.

Bitget

• Passphrase. The one you chose when creating the key, not your account password and not your 2FA code. A wrong passphrase gives an authentication error that reads like a bad secret.
• Position mode. The system requires one-way mode, because hedge mode makes the target book ambiguous. Switch in Bitget’s futures settings; the exchange refuses the change while positions are open, so close them first.
• Futures account, not spot. Balances must be in the USDT-M futures account.

Gate

• Unified or futures account. The key must have futures trading permission, and the balance must be in the account the key can reach.
• Read-only keys. Gate defaults new keys to read-only. Enable trade access explicitly.

Hyperliquid

• The two fields are different objects. The Wallet Address is your main account address, starting 0x. The API Wallet Private Key is the key of an API wallet you generate under Hyperliquid’s API settings.
• Do not paste your main wallet’s private key. The most common Hyperliquid failure, and it hands over far more authority than the system needs.
• The API wallet must be approved and unexpired. If yours lapsed, generate a new one and update the connection.

What to do. Fix the field or setting, then test again. If it still fails, delete the key at the exchange, create a fresh one, and add the connection again rather than editing the broken one.

My exposure dropped by half overnight

This is probably not a fault

The most common “something is wrong” report, and usually a deliberate risk control firing.

Symptom. Gross exposure on Portfolio is roughly half what it was, across the whole book. Positions are smaller but the mix of names looks similar, and net exposure is still near zero.

Likely cause. Regime scaling. When our risk rules classify the market as unfavourable, the system scales the whole book’s gross exposure down, to roughly half. It applies to the book rather than to individual names, hence the uniform halving.

What to check.

• Did every position shrink by a similar proportion? A uniform reduction points at regime scaling; one or two names points at per-symbol caps or ordinary rebalancing.
• Is net exposure still near zero? Regime scaling reduces gross and leaves the long/short balance intact.
• Did your leverage setting change? Nothing we do changes it automatically, so a changed setting means someone changed it. Being over your plan’s cap has a different consequence — see My plan is changing.
• Did your account equity fall? Position sizes derive from equity, so a drawdown shrinks the book on its own.

What to do. Nothing. The scaling reverses when conditions change and the book is rebuilt over the following runs. There is no override, by design.

An order did not go out

Symptom. A rebalance ran, but Trade History shows adjustments listed as held or not placed rather than filled.

Likely cause. Most are too small for the exchange to accept. Every venue enforces a minimum order size and a quantity step, and a small rebalance can land under it. The system declines to send those rather than have them rejected or rounded to a size the model did not ask for.

What to check. Trade History splits these into two groups, and the split matters:

• Held for a later run. Routine. Each was below the exchange minimum or below one tradable unit. The group header states how many, the largest, and the total.
• Other outcomes. Not routine: no usable price, missing market data for the symbol, or an outcome this build does not recognise. Anything here is worth reading.

A held order is not queued

The system does not promise a held adjustment will be placed later. With a stable target and a stable price, an adjustment that is too small stays too small. What you get is a bound: your exposure in that name stays within one exchange unit of target. If a reduction has been held for many runs in a row, Trade History labels it as such rather than calling it normal, and that one is worth raising with us.

What to do. For the held group, nothing. If the same names are held every run, the cause is account size relative to the venue’s order minimum, and a larger account or a venue with a lower minimum fixes it. For anything in the other group, read the reason on the row and contact support with it. Full detail in Orders that were held.

My system says inactive, or no trades are happening

Symptom. The system shows as inactive or paused, or runs happen but never place anything.

Likely cause. One of three gates. All three must be open before your account is traded:

1. Payment. The subscription must be active or in its grace period. A failed payment moves it to past due or unpaid and execution stops there; the banner in Settings states the grace-period deadline.
2. Exchange connection. The system must be linked to a connection in the connected state. A pending or failed connection is not traded.
3. The system itself. A system never started, or paused, produces no runs. The dashboard says System is paused rather than showing a countdown.

What to check. Work down that list in order, since a payment problem makes the other two irrelevant. Then check Settings: if you are at your plan’s system limit or over the leveraged-capital cap, the app says which.

What to do. Update the payment method through the billing portal, re-test the connection, or start the system. Once all three are open, the next scheduled run picks the account up. See Plan limits for the cap.

My plan is changing, or I am over my plan limit

We change the plan, not your leverage

We no longer reduce the leverage you selected in order to keep you inside a plan. If you remember that behaviour, or read it in older documentation, it has been replaced by the plan change described here.

Symptom. Settings or the dashboard’s tier usage card shows Over Limit, or names a plan and a date you did not pick.

Likely cause. Your leveraged capital — account equity multiplied by your leverage setting — has been above your plan’s cap on most of the recent daily runs. A single day over does nothing; a sustained breach schedules a move to the plan that covers your account, with at least 14 days’ notice and effective no earlier than your next renewal.

What to check. Open Settings and read the Tier Usage card: equity, leverage, leveraged capital, and the cap. Deposits are the usual trigger, strong performance the other. Your leverage slider on Systems will read exactly what you set it to — that is expected, not a sign the warning is stale.

What to do. Nothing, if the larger plan suits you: the change happens on the stated date. Otherwise, before that date, either bring usage below the cap (lower your leverage yourself, or withdraw) or pick a plan directly in Settings. Above 80 percent of the cap the app warns you well before anything is scheduled.

If your account later settles back down, we move you back down automatically at a subsequent renewal, though never below the plan you originally chose.

Numbers on Dashboard and Portfolio disagree

Symptom. Equity, exposure, or P&L reads differently on two pages at the same moment.

Likely cause. Different scope, or a different window, or both. Neither figure is wrong; they answer different questions.

• Scope. The dashboard aggregates across every connected exchange. Portfolio and per-connection cards show one connection, so with more than one exchange connected they will not match, and should not.
• Window. A P&L figure is a change over a period, so a different period gives a different number from the same data.
• Level versus change. Open P&L is a level: how far your positions sit above or below their entry prices, measured from whenever each was opened. A chart or a return figure is a change over a window. These can point in opposite directions at once and both be correct.
• Freshness. A figure labelled as last verified at a time is a point-in-time record from the last connection test, not a live reading.

What to check. Read the label under each figure: tiles state scope, window, and whether the value is live or stale. An unavailable figure shows a dash, never a zero.

What to do. Compare like with like: same connection, same window. If two tiles with the same stated scope and window genuinely disagree, report it. Portfolio explains each tile individually.

The Next Run time keeps moving

Symptom. The Next Run countdown shifts, or shows a time that does not match what you expected.

Likely cause. Two sources feed that figure. When a scheduled run has been recorded, the stored time is shown. When it has not, the app estimates from the schedule in your browser and marks it (estimated), and an estimate recalculates as the clock moves.

Times are computed in UTC and displayed in your local timezone, so daylight-saving changes shift the local time without the schedule changing.

What to check. Look for the word “estimated” under the countdown. If it is there, no scheduled run has been recorded yet, which is expected before your first run and after a pause.

What to do. Nothing, unless the countdown reaches zero and stays there with no new activity for well over an hour; runs can start a little late under load. If a full day passes with no run and the three gates above are open, contact support.

Still stuck

Include these when you get in touch, they cut the diagnosis in half:

• Which exchange and which system.
• The exact error text from the connection card or the Trade History row.
• The date and time of the run in question, and your timezone.
• Anything that changed just before it: a deposit, a leverage change, a new API key, or a manual trade in the same account.